outdoor mimo wireless networks

Outdoor MIMO Wireless NetworksVersion 1.0

Outdoor MIMO Wireless Networks Validated Reference Design

Copyright© 2012 Aruba Networks, Inc. AirWave®, Aruba Networks®, Aruba Mobility Management System®, Bluescanner, For Wireless That Works®, Mobile Edge Architecture®, People Move. Networks Must Follow®, RFprotect®, The All Wireless Workplace Is Now Open For Business, Green Island, and The Mobile Edge Company® are trademarks of Aruba Networks, Inc. All rights reserved. Aruba Networks reserves the right to change, modify, transfer, or otherwise revise this publication and the product specifications without notice. While Aruba uses commercially reasonable efforts to ensure the accuracy of the specifications contained in this document, Aruba will assume no responsibility for any errors or omissions.

Open Source CodeCertain Aruba products include Open Source software code developed by third parties, including software code subject to the GNU General Public License (“GPL”), GNU Lesser General Public License (“LGPL”), or other Open Source Licenses. The Open Source code used can be found at this site:

http://www.arubanetworks.com/open_source

Legal NoticeARUBA DISCLAIMS ANY AND ALL OTHER REPRESENTATIONS AND WARRANTIES, WEATHER EXPRESS, IMPLIED, OR STATUTORY, INCLUDING WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, TITLE, NONINFRINGEMENT, ACCURACY AND QUET ENJOYMENT. IN NO EVENT SHALL THE AGGREGATE LIABILITY OF ARUBA EXCEED THE AMOUNTS ACUTALLY PAID TO ARUBA UNDER ANY APPLICABLE WRITTEN AGREEMENT OR FOR ARUBA PRODUCTS OR SERVICES PURSHASED DIRECTLY FROM ARUBA, WHICHEVER IS LESS.

www.arubanetworks.com

1344 Crossman AvenueSunnyvale, California 94089

Phone: 408.227.4500Fax 408.227.4550

Aruba Networks, Inc. 2

http://www.arubanetworks.com/open_source


Table of Contents

Chapter 1: Introduction 9About the Outdoor MIMO Wireless Networks VRD 9Outdoor Deployment Types 9

Campus Extension 10Outdoor Mesh with AirMesh 11

Aruba Reference Architectures 12Outdoor Wireless Integrators 13Assumptions 13Reference Documents 14

Icons Used in this Guide 15

Chapter 2: Outdoor Networking Deployment Methodology 17Network Discovery 17Preliminary (High-Level) System Design 18Site Acquisition 18Final (Low Level) System Design 19Installation and Configuration 19Coverage and Throughput Verification 20Final Network Acceptance 20

Chapter 3: Outdoor Access Points and Multichannel Backhaul 21Choosing the Deployment Type 21Understanding Single-Channel and Multi-channel Backhaul 21

The Evolution of Mesh Technology 22Comparing End-to-End Performance 24

ArubaOS AP for Campus Extension 26AP-175 (Campus Extension) AP 26

AirMesh APs for Outdoor Mesh Networks 27MSR4000 Quad-Radio Mesh Router 27MSR2000 Dual-Radio Wireless Mesh Router 28MST200 Single-Radio Wireless Mesh Router 29MSR1200 Dual-Radio Indoor Mesh Router 30

AP Model Summary 31

Chapter 4: Outdoor Antennas and RF Coverage Strategies 33Antenna Beamwidth, Pattern, and Gain 33

Omnidirectional Antenna Types 35Directional Antenna Types 36

Aruba Networks, Inc. Table of Contents | 3


Effect of Mechanical Down Tilt on Directional Antenna Coverage 37Directional Antenna Conclusions 40

RF Coverage Strategies for Outdoor WLANs 41Understanding Side and Overhead Coverage 41Sparse Side Coverage 44Dense Side Coverage 45Dense Overhead Coverage 46

Selecting an Aruba Outdoor Antenna 47Understanding Aruba MIMO Antenna Part Numbers 47Access Layer Antennas 48Backhaul Layer Antennas 51

Chapter 5: 802.11n Multiple-In and Multiple-Out 53Ratification and Compatibility 53Understanding MIMO 53

802.11n Spatial Streams 54

Other 802.11n Technologies to Increase Throughput 5440 MHz Channels 54Improved OFDM Subcarriers 56Space Time Block Coding and Maximal Ratio Combining 57Short Guard Interval 57

Understanding MAC Layer Improvements 57A-MSDU 58A-MPDU 58Block Acknowledgement 59

Putting It All Together – From 54 Mb/s to 600 Mb/s 59802.11 Terminology 60

Transmit, Receive, and Spatial Stream Designation 60Modulation and Coding Scheme Index 612.4 and 5 GHz Support 62

Backward Compatibility 63Maximizing Rate vs. Range with MIMO Outdoors 63

Direct vs. Indirect Multipath 64Correlation and Decorrelation 65Polarization 65Leveraging Polarization Diversity to Improve Decorrelation 66

Chapter 6: AP Selection for Common Outdoor Topologies 67Single-Radio Point-to-Point Bridge: MST200 67Single-Radio Leaf Node: MST200 67



Dual-Radio Outdoor-to-Indoor: MSR1200 68Dual-Radio Mobile Applications: MSR1200 or MSR 2000 68Dual-Radio Client Access: AP-175 and MSR2000 69Single Hop Point-to-Point: AP-175 or MSR2000 69Multi-hop Linear Mesh: MSR2000 70Parallel Point-to-Multipoint: MSR2000 or MSR4000 70High Capacity Mesh Core: MSR4000 71Remote Thin AP Endpoints Overlaid on AirMesh 72

Chapter 7: Aruba Software Technologies 73Choosing an Outdoor Operating System 73ArubaOS for Campus Extension 74AirMesh for Outdoor Mesh Networking 75

Radio Frequency Management 76Adaptive Wireless Routing 78Path Distance Factor 86Active Video Transport 91Virtual Private LAN over Mesh 93MobileMatrix and Seamless Session Persistent Roaming 96

Chapter 8: Planning the Access Layer 101Discovery 101

Define the Coverage Footprint 102Identify Siting Constraints 103Identify Quality-of-Service or Special Service Level Agreement Zones 103Specify Key Network Design Parameters 103

Capacity Planning 104Offered Loads of Typical Network Services 104Bandwidth vs. Throughput 104Client Throughput Requirements 105Oversubscription Ratio 106Strategic Throughput Reservation 106

Determining Cell Size 107Matching Client and AP Power 107Free-Space RF Propagation 108Effect of Path Loss on Data Rate and Throughput 109Estimate Path Losses 110Link Budget Calculation and Link Balance 111Path Loss Due to Cumulative RF Absorption 112Path Loss Modeling for Indoor Coverage by Outdoor APs 114



Summary 115

Using the Aruba 3D Outdoor RF Planner 115Finished RF Plan Examples 116

Chapter 9: Planning the Mesh Backhaul Layer 119Identify Portal Candidates 119Choose RF Backhaul Topology 120

Serial Point-to-Point Connections 120Parallel Point-to-Multipoint Connections 120Full Mesh in a Multi-Gateway Design 121

Choose Capacity Injection Topology 122End-Fed Injection Topologies 122Center-Fed Injection Topologies 124Hybrid Topologies 125Maximum Hop Count 126Maximum Number of Children 126Ratio of Mesh Portals to Mesh Points 126

Capacity Planning 127Determine Number of Usable Backhaul Channels 127Compute Ingress Load 129Compute Egress Load 129Estimate Bandwidth of Individual Mesh Links 130Mesh Capacity Math for Single Channel Backhaul Systems 131Model End-to-End Traffic Flows 133

RF Design 134Planning Mesh Layers with the Aruba 3D Outdoor RF Planner 135

Chapter 10: Site Surveys for Large Outdoor Networks 137Create a “Soft” RF Plan 137General Considerations for Choosing Mounting Assets 139

Identifying RF Absorbers, Reflectors, and Interferers 140

Selecting Mounting Locations for Mesh Points 140Performing the Survey 141Choosing a Pole 142Evaluating Pole Power From the Ground 143Reading Pole Tags 144Measuring Pole Dimensions 144

Radio LOS Path Planning 146Antenna Height 146



Surveys for Mesh Portal Mounting Locations 148Wired Backhaul Assessment 148Antenna Position and Orientation 148Radio Interference 149Weather Conditions 149Ethernet Cabling 149Grounding 150

Civils Approvals 150Final Network Design 151Best Practices for Conducting Outdoor Surveys 151

Personal Safety & Security 151Building a Complete Outdoor Survey Kit 152

Chapter 11: IP Planning for Aruba AirMesh 157Configure a Router ID 157Mesh Backhaul Links 157Access Links and Client Devices 157Wired Network Ethernet Link Parameters 157IP Addressing and Networking 157

Chapter 12: Installation, Validation, and Optimization 159MeshConfig 159Staffing Expectations 161Aruba Outdoor AP Antenna Weatherproofing 162

Installing Antennas 162Weatherproofing Connections 162

RF Coverage Validation 171Reconciling Drive Test Data with Predictive Models 172

Mesh Network Optimization 172

Appendix A: Allowed Wi-Fi Channels 1732.4 GHz Band 1734.9 GHz Band 1735 GHz Band 174

Appendix B: DFS Operation 177Behavior of 5 GHz DFS Radios in the Presence of Radar 177

Appendix C: Campus Extension Example 179



Appendix D: Intermodal Transportation Example 183Application Types 183Dense Overhead Coverage Strategy 184Sparse Side Coverage Strategy 185

Appendix E: Open Pit Mine Example 187

Appendix F: Aruba Stadium Design Summary 189RF Design 189“Picocell” Coverage Strategy 190Understanding the Structure of a Picocell 191Picocell Coverage & Performance Validation Test Results 193

Appendix G: Terminal Doppler Weather Radars 197

Appendix H: Aruba Contact Information 201Contacting Aruba Networks 201



Chapter 1: Introduction

This Solution Guide is designed to help customers understand the Aruba system architecture and the individual components that are needed to deliver reliable, high-capacity outdoor networks using 802.11n with multiple-in and multiple-out (MIMO) radios.

About the Outdoor MIMO Wireless Networks VRDAruba has extensive experience designing complex outdoor WLAN solutions for applications like stadiums, outdoor transportation terminals, oil and gas facilities, municipalities, and large campus environments. Aruba outdoor solutions meet the needs of emerging applications by increasing the speed of each connection. This increase in speed is achieved using MIMO-based radio techniques and mesh routing for very large outdoor areas.

This guide describes these main points: The lifecycle of an outdoor wireless network deployment Typical basic processes and tools that are used in outdoor wireless networking Products and technologies to meet the needs of a wireless network operator MIMO antenna selection and placement for maximum capacity Design recommendations for common deployment scenarios Regulatory rules that must be incorporated into all outdoor RF designs

Outdoor Deployment TypesThis guide addresses two distinct types of deployments, each of which has its own technical requirements, coverage strategies, and implementation methodologies:

Campus extension with AP-175: Customers that have standardized on a controller-based thin AP architecture for indoor coverage often want to extend the role-based access control (RBAC) security model to the outdoor environments on their properties.

Outdoor wireless mesh with AirMesh: Some customers operate a wireless network that is almost exclusively outdoors. Indoor connections can be provided from the outdoor network, usually via remote bridge links or special-purpose indoor repeaters.

Aruba offers a choice of two different mesh-capable operating systems. The best choice typically depends on which deployment type best fits the intended outdoor wireless network.

Both types of deployment use: Mesh portals: Connected to the high-speed wired network (also known as wired gateways). Mesh points: Unwired radios that connect to mesh portals using an RF backhaul link. Mesh

points are fully capable of multihopping over long distances.

Aruba Networks, Inc. Introduction | 9


Campus Extension

ArubaOS outdoor solutions are used to extend secure indoor enterprise coverage to outdoor areas. Some common examples of these applications include:

Campus coverage for universities, hospitals, and large enterprises Manufacturing plants Industrial yards Ports, rail yards, and airports Stadiums, arenas, and other large public venues for Internet access or 3G offload

In these environments, controller-based wireless LANs (WLANs) are generally running indoors using a wired backbone to connect thin APs to an Aruba controller. For example, in the case of an intermodal transportation facility or manufacturing plant, it is often true that the business offices either are using or migrating to a controller-based architecture. For this reason, IT departments want to have the same security model for outdoor facilities. Also, consistent equipment and configuration procedures can reduce IT operating costs.

From a hardware perspective, a campus extension network generally requires a rugged version of the dual-radio access point (AP) that is used indoors. A campus extension network is illustrated in Figure 1. In this case, we assume an existing indoor ArubaOS WLAN, which is extended out via mesh to cover the outdoor portions of the facility.

Figure 1 Campus extension network (Container Port)

For campus environments, both radios are often used to provide client access, with occasional short mesh hops to connect remote buildings or provide spot coverage from utility poles nearby. Mounting assets tend to be buildings; consequently, AP power is primarily power-over-Ethernet (PoE). Using PoE leverages the existing indoor infrastructure and makes sense given the limited number of AC- or DC-powered nodes.



Generally, campus extension networks should use ArubaOS, with outdoor APs managed by the same controller(s) that support the indoor network. ArubaOS is an “overlay” network, which assumes that a reliable wired LAN or WAN interconnects the APs with their controller.

Outdoor Mesh with AirMesh

When a green-field outdoor wireless network is considered, as shown in Figure 2 the driving application may or may not include some indoor coverage. But these large area networks use mesh routing technology instead of extending an indoor controller-based architecture.

Figure 2 Green-field outdoor wireless network topology (City Grid)

In the long-term, multiple applications and new users must be supported on these outdoor networks. During the planning stage, consider how network capacity can be increased in the future. Examples of common green-field wireless networks include:

Municipal Wi-Fi® for video surveillance and public/private network access Mines, oil fields, and other large, outdoor, industrial facilities Emerging smart-grid applications

In these green-field wireless networks, the outdoor mesh network provides the backbone for delivering all applications and services. These networks can cover extremely large areas, measured in square kilometers (km2) or square miles (mi2). Any viable mounting asset in the vicinity of a desired mesh node location must be supported. Therefore, a wide variety of single-, dual-, and quad-band radio options are necessary to provide the wireless architect with maximum flexibility. AC- and DC-power dominates outdoor mesh networks, with some PoE at mesh portals. The 4.9 GHz licensed band can be used in countries that allow it.

Outdoor mesh networks should generally use Aruba AirMesh™ on standalone Multi-Service Router (MSR) routers. MSR routers provide LAN-like layer 3 (“L3”) and layer 2 (“L2”) traffic forwarding across

Ethernet

DSL



links of varying quality and availability. These routers also provide a range of other features to maximize the performance of various IP network services over a large area.

Aruba Reference ArchitecturesThe Aruba Reference Design series is a collection of technology deployment guides that include descriptions of Aruba technology, recommendations for product selections, network design decisions, configuration procedures, and best practices for deployment. Together these guides comprise a reference model for understanding Aruba technology and designs for common customer deployment scenarios. Each Aruba VRD network design has been constructed in a lab environment and thoroughly tested by Aruba engineers. Our customers use these proven designs to rapidly deploy Aruba solutions in production with the assurance that they will perform and scale as expected.

The VRD series focuses on particular aspects of Aruba technologies and deployment models. Together the guides provide a structured framework to understand and deploy Aruba wireless LANs (WLANs). The VRD series has four types of guides:

Foundation: These guides explain the core technologies of an Aruba WLAN. The guides also describe different aspects of planning, operation, and troubleshooting deployments.

Base Design: These guides describe the most common deployment models, recommendations, and configurations.

Applications: These guides are built on the base designs. These guides deliver specific information that is relevant to deploying particular applications such as voice, video, or outdoor campus extension.

Specialty Deployments: These guides involve deployments in conditions that differ significantly from the common base design deployment models, such as high-density WLAN deployments.

Figure 3 VRD core technologies

arun

_033

4Foundation

Base Designs

SpecialtyDeployments

Applications



Outdoor Wireless IntegratorsOutdoor wireless networks are the most labor-intensive and challenging type of WLAN to design and deploy. Many different disciplines and trades must come together for a successful outdoor network, including:

Project management RF engineering LAN and IP network engineering Construction and fabrication Tower erection, climbing, and rigging Grounding and electrical safety AC, DC, battery-assist, and solar power systems Municipal attachment rights agreements and city council testimony

Few IT departments have access to experts in all of these areas. Therefore, Aruba strongly recommends that every customer that intends to deploy an outdoor system of any size engage an experienced outdoor wireless network integrator. These companies can provide any type of resource required for a successful project, and they can help navigate the many issues that inevitably come up during an outdoor project.

Your local Aruba account manager can help direct you to a qualified outdoor integrator. You can also explore the Aruba ServiceEdge™ provider network, which includes many integrators who specialize in outdoor work: http://www.arubanetworks.com/partners/deployment_partners/locate.php.

AssumptionsIn this guide we make several assumptions about the level of experience that a network planner has. We provide references to some basic material, but we make the following assumptions:

Reader is familiar with unlicensed band plans. Reader understands RF link budget planning in outdoor environments. Reader understands MIMO fundamentals. Reader is experienced with physical installation of outdoor radio equipment.


http://www.arubanetworks.com/partners/deployment_partners/locate.php


Reference DocumentsThe following documents are recommended for further reading on 802.11n, MIMO, and outdoor wireless networking technologies.

Designed for Speed: Network Infrastructure for an 802.11n World, Peter Thornycroft, Aruba, 2008

Next Generation Wireless LANs: Throughput, Robustness, and Reliability in 802.11n, Eldad Perahia and Robert Stacey, Cambridge University Press, 2008

Hardware Installation Guides - Aruba AP-175 and MSR2000 Outdoor APs Certified Wireless Network Administrator (CWNA) Study Guide, David A. Westcott & David D.

Coleman, John Wiley & Sons, 2006 Aruba Networks 3D Outdoor RF Planner Aruba Antenna Matrix

The following reference materials and discussion groups are recommended for learning about Aruba products and solutions:

For information on Aruba Mobility Controllers and deployment models, see the Aruba Mobility Controllers and Deployment Models Validated Reference Design, available on the Aruba website at http://www.arubanetworks.com/vrd

The complete suite of Aruba technical documentation is available for download from the Aruba support site. These documents present complete, detailed feature and functionality explanations beyond the scope of the VRD series. The Aruba support site is located at: https://support.arubanetworks.com/. This site requires a user login and is for current Aruba customers with support contracts.

For more training on Aruba products or to learn about Aruba certifications, visit the Aruba training and certification page on our website. This page contains links to class descriptions, calendars, and test descriptions: http://www.arubanetworks.com/training.php/

Aruba hosts a user forum site and user meetings called Airheads. The forum contains discussions of deployments, products, and troubleshooting tips. Airheads Online is an invaluable resource that allows network administrators to interact with each other and Aruba experts. Announcements for Airheads in person meetings are also available on the site: http://airheads.arubanetworks.com/

The VRD series assumes a working knowledge of Wi-Fi®, and more specifically dependent AP, or controller based, architectures. For more information about wireless technology fundamentals, visit the Certified Wireless Network Professional (CWNP) site at http://www.cwnp.com/


https://support.arubanetworks.com/

https://support.arubanetworks.com/

http://www.arubanetworks.com/training.php/

http://airheads.arubanetworks.com/

http://www.cwnp.com/

http://www.arubanetworks.com/vrd


Icons Used in this Guide

Figure 4 shows the icons that are used in this guide to represent various components of the system.

Figure 4 VRD icon set

Video camera

ResidenceBuilding

Mobile phone Smart phoneDirectionalantenna

Network cloud

AP

Laptop

Arubacontroller AirWave

server

Switch

Router

Server

arun

_044

5

RAP5

S3500wired AP

WiredMUX

MST200(physical)

MSR1200(physical)

Surveillancecenter

MSR2000(physical)

Attenuator

Tunnels

MSR 2Kor AP-175

MST200(logical)

MSR4000(logical) AP with

camera & light



Chapter 2: Outdoor Networking Deployment Methodology

For many existing Aruba customers, an outdoor network often is an extension of their indoor network that delivers coverage across a large enterprise or hospital campus. After mounting locations are selected, installation for these customers is like adding coverage indoors: select the right APs and antennas and make sure the controller supports the required licenses. For other customers who want to build larger outdoor Wi-Fi networks, mesh radios are used and the selection of mounting locations becomes more complex. This chapter describes a general methodology that is common to campus extension and outdoor-mesh networks.

Whether you are extending an indoor network or building a large outdoor mesh network, the planning process generally includes the steps displayed in Figure 5 to create a scalable, manageable network with the required coverage and capacity:

Figure 5 Outdoor network deployment process

These steps can be completed quickly when an Aruba network is being extended because customers are familiar with existing locations for outdoor antennas and radios. However, large outdoor networks often require very detailed plans and may require civil approvals and permits for mounting locations that are not owned by the network operator.

Network Discovery Like all IT projects, an outdoor wireless network begins with a discovery process. An outdoor discovery includes these components:

Map of the expected coverage area Statement of desired operating capacity List of potential mounting assets under the control of the network operator Primary network users, in order of priority Primary applications, in order of priority Desired project schedule, broken into relevant phases Available budget for initial construction and ongoing operation

Existing Aruba customers who plan campus extensions often can provide accurate mounting location and terrain information that can be used during the outdoor planning process. These outdoor networks may cover limited areas or be simple point-to-point solutions to bridge multiple buildings or locations

arun

_042

3

Step 1Networkdiscovery

Step 2Preliminary

systemdesign

Step 3Site

acquisition

Step 4Final

systemdesign

Step 5Installation

andconfiguration

Step 6Coverage and

throughputverification

Step 7Final

networkacceptance

Aruba Networks, Inc. Outdoor Networking Deployment Methodology | 17


together. For these customers, the locations of radios identified in the preliminary system design and the final system design can be very close.

For large outdoor mesh networks, the objective of the discovery step is to deliver a realistic overview of the whole network, by outlining wired and wireless resources, which provides the foundation for more meaningful planning during later steps.

Preliminary (High-Level) System Design The preliminary system design establishes clear coverage and capacity expectations for each outdoor area. After the high-level coverage area is identified, break the area into smaller logical sections of about 1-2 km2 or mi2 for further detailed planning. A preliminary design always includes the initial site survey and an RF spectrum analysis. Depending on the size of the area to be covered, these two tasks require the largest labor component of the preliminary design.

Large outdoor mesh networks rely on cells of coverage that communicate using layer 3 mesh routing. First identify the number of active users that can be expected in each area and the peak bandwidth the network is expected to deliver. Then use the following key metrics for further planning:

Number of cells per kilometer or square mile The ratio of mesh points (unwired radios) to mesh portals (wired radios) For each area, identify mounting assets with access to usable power

The preliminary system design generally includes these components: Site survey and spectrum analysis report First draft of the RF design model for the network, possibly including IP design Preliminary bill of materials Proposed mounting locations and wired network access locations Radio propagation models and antenna selections for each mounting location Testing tools needed to verify coverage and capacity Preliminary budget estimate for integration and construction services

Site AcquisitionSite acquisition often involves two types of radio mounting assets:

Assets that are owned or under the control of the network operator, like buildings Assets that may require permits and payment to a third party, like street lights

For example, a university that wants to expand the network to cover outdoor common areas can generally assume that they can mount radios on the buildings and streetlamps within the campus. On the other hand, if the preferred mounting assets for that same area require mounting on third-party building rooftops or city-owned lights, then negotiations and timing can take longer. During site surveys that include these types of locations, it is common to identify alternate mounting locations in case the preferred sites are unavailable (which can be quite common).

Each mounting site must support the weight of radios and any wind load and have access to continuous, unswitched electrical power. Each radio location must also have a suitable grounding



path. The antennas and mounting methods for each site are selected to provide the desired client coverage and to complete a reliable RF path to other mesh points along the path to the mesh portal and the wired network.

Final (Low Level) System DesignThe final design should provide a detailed RF design and include detailed mounting location information, such as GPS coordinates step-by-step cable pathway instructions to help with radio installation planning. The final design must also include detailed IP addressing information and other back-end system interfaces that may be required, such as captive portals for public networks. For Aruba customers, outdoor networks are often simple extensions of the current role-based access controls. However, new multiuse outdoor networks may require implementation of new authentication models that should be carefully considered when planning the network.

The final design typically includes this information: Radio specifications for each validated mounting location User device characterization for network planning Clear coverage and capacity expectations by area Mesh portal radio locations and wired network connections Mesh point mounting locations and electrical powering plan RF frequency plan if required IP network design for the mesh network, wired network and back-office equipment An agreed-upon method of testing and validating coverage and capacity Deployment-related services and other resources

Installation and ConfigurationTo install and configure each radio, follow the steps in the hardware installation guides, as identified in the final system design. As equipment is installed, carefully record the GPS coordinates of each radio and document these for later use. Take pictures of each installation from multiple angles because each location may not be visited for long periods of time. Aruba recommends that installers label each cable and the port to which it is attached. Sometimes it is necessary to affix customer-specific labels that identify the network owner or operator or other asset tracking information. This information is invaluable for troubleshooting elevated radios.

To simplify installation in the field, always preconfigure each remote radio. Be sure to follow the IP network design to include the mesh radios and back-office equipment.

Aruba strongly recommends that only experienced outdoor wireless integrators install outdoor radio equipment. A licensed electrician must complete all radio grounding, and must install low-voltage or high-voltage power systems required by the network.



Coverage and Throughput VerificationWhile the network is being installed, it is common to measure coverage periodically using GPS-enabled tools such as Air Magnet Survey Professional or Ekahau SiteSurvey Professional. When an entire area or subarea is completely installed, drive tests are performed. Drive test results show “heat maps” of the signal strength, which document the level of coverage. However, common best practice is to measure the Receive Signal Strength (RSSI) using independent third-party tools. Doing so ensures coverage in the required bands:

2.4 GHz 802.11 b/g/n 5 GHz 802.11a/n Municipal use of the 4.9 GHz bands (optional)

Then compare these results with the original system design to identify coverage gaps or holes. Address these gaps by identifying additional mounting locations and adding equipment and installation resources from a pool that is reserved for this purpose.

RF signal strength heatmaps only tell part of the coverage story, namely the AP-to-client radio propagation. Properly done with the AP power matching the expected client power, it can also indicate the likely return path. However, it does not necessarily tell you anything about actual two-way data throughput. This is especially true because the capacity of the network may increase based on MIMO spatial streams in each location. As you will learn in Chapter 5, the ability of radios to decorrelate individual spatial streams does not necessarily depend on SNR.

To test two-way throughput, one must take performance measurements from sample points around the area using active testing tools such as iperf or Ixia IxChariot. Aruba recommends a uniform test suite at each test point:

TCP upstream, downstream and bidirectional UDP upstream, downstream and bidirectional Repeat each of the above on each major type of client device to be used

In general it is important to use multiple streams (2-4 each way) whether using iperf or IxChariot to generate sufficient load through the IP layer of the network driver stack. Once the throughput results are obtained, additional optimization of the network may be advisable. It is also possible to test different pathways across the network by using multiple traffic “sink” locations at various points in the mesh.

During this phase, it is common to install monitoring systems and begin to measure the network reliability. Additionally, the network operator is trained on how to use the monitoring systems.

Final Network AcceptanceDuring the final acceptance step, a coverage heat-map and throughput testing results from a drive test are usually summarized and a final report is prepared with the assistance of the customer. The network documentation should include the street address and GPS coordinates of every installed radio, pictures of the majority of installations, and detailed IP network diagrams.



Chapter 3: Outdoor Access Points and Multichannel Backhaul

Aruba offers a wide range of APs, antennas, and related accessories to enable campus extension and outdoor mesh wireless networks. The choice of which hardware and operating system to use for a given network is driven by the deployment type and often by the need for single-channel or multi-channel backhaul.

Choosing the Deployment TypeAruba has two families of outdoor APs: the AP-175 series and the MSR series. These APs are similar in design, but run different operating systems. An outdoor area can be covered by extending an existing Aruba indoor network through the use of the AP-175 outdoor AP with the proper antennas. This AP runs ArubaOS and is managed by a controller. The AP-175 can interoperate with Aruba indoor APs, can be used as mesh portals, and can be used with other ruggedized AP-175s that are operating as unwired mesh points. Role-based user access policies are preserved across the combined indoor and outdoor network.

In large outdoor networks, the AirMesh MSR series of wireless mesh routers are mounted on rooftops, radio towers, street lights, and even traffic lights to extend coverage across large areas. When considering outdoor Wi-Fi networks, good coverage is generally equated to the availability of suitable mounting assets in combination with Aruba hardware and antenna flexibility. The MST1200 is an indoor AP that is suitable for use as a mesh portal with external antenna connectors that can be connected to remote antenna outdoors for a low-cost solution. The MSR series runs the Aruba AirMesh operating system.

To provide scalable coverage over large outdoor areas, wireless networks use combinations of mesh portals. Mesh portals are connected to the wired network and wireless mesh points. For each radio, its role, frequency band, and channel are defined in the software configuration. Mesh links connect mesh points to other mesh points and to mesh portals, which then connect to a high-speed wired network.

Table 1 lists the AP models that should be used for each deployment type.

Understanding Single-Channel and Multi-channel Backhaul A key factor in choosing which AP family is best for your outdoor network is the number of radio channels that will be used for backhaul. In general, campus extension networks with the AP-175 tend to have very few hops and can utilize a single-channel for intramesh backhaul, whereas outdoor mesh

Table 1 AP model based on deployment type

Deployment Type Dual-Radio Rugged

Quad-Radio Rugged

Single-Radio Rugged

Single-Radio Indoor

Dual-Radio Indoor

Campus Extension (ArubaOS)

AP-175 - - AP-92 AP-124/AP-134

Outdoor Mesh (AirMesh) MSR2000 MSR4000 MST200 - MSR1200

Aruba Networks, Inc. Outdoor Access Points and Multichannel Backhaul | 21


networks built with the AirMesh family typically have many hops and use multiple channel backhaul links to extend capacity.

The Evolution of Mesh Technology

Mesh networking technology has been enabling production networks for many years. In that time, it has gone through several generations, culminating in the fourth generation AirMesh solution from Aruba. Figure 6 illustrates the progression of technology enhancements:

Figure 6 Summary of wireless mesh technology evolution

To help put the value of the AirMesh solution into perspective, it is useful to consider how mesh technology has evolved over the years:

First generation - Single radio L2 mesh. The earliest mesh implementations used single radio APs in the 2.4GHz band for both client and backhaul service. Since there is only one radio, all mesh nodes are on the same channel. This means that when one radio is transmitting, whether a client or another mesh node, no other radio can transmit. This approach suffered from two major performance limitations. First, client transmissions had to be received by the AP, and then retransmitted on to the upstream mesh node(s). This meant that the offered load at the access layer could not exceed 50% of the uplink bandwidth to avoid saturation. Second, if there was more than one mesh hop, the same effect occurred again on the backhaul. This further reduced the allowable offered load at the access layer. First generation meshes operated at layer 2.

Second generation - Dual radio L2 mesh. An obvious solution to the client performance limitation was to use separate radios for client and backhaul service. Second generation mesh APs typically used 2.4GHz for client access and the 5GHz band for backhaul. In this design, all mesh radios shared the same channel, though client radios could use typical 1, 6, 11 channelization. Now the AP could serve clients simultaneously with backhaul traffic. However, when relaying frames between mesh nodes, the 50% throughput drop per hop was experienced

arun

_043

7

Performance&

scalability

Technology evolution

1st Generation

Indoor access• Single radio• Omni-directional antenna• Layer 2 bridging

2nd Generation

Indoor & outdoorHot spots

• Dual radio• Omni-directional antennas• Layer 2 bridging

3rd Generation

Hot zonesLow-res video

• Multi-radio• Directional antennas• Layer 2 bridging

4th Generation

Municipal coverageHD-quality video

Voice, and mobility

• Multi-radio 802.11n• Directional antennas• Layer 3 routing



because each mesh node had to receive a transmission before repeating it upstream. Second generation meshes also operated at layer 2.

Third generation - Multichannel Layer 2 backhaul. Some vendors eliminated the first and second generation intermittent send-receive-send cycle by using two radios for the backhaul. These radios are generally operating on separate non-interfering channels. Simultaneous send and receive is now possible. This dramatically improves latency over multiple mesh hops. However, due to the layer 2 topology, the mesh had a fixed tree structure such that all traffic flowing through the mesh must pass through the “root” node. For some traffic flows this is no problem. However, for peer-to-peer applications such as connecting a mobile police car to a remote video camera, the root node bottleneck imposed significant performance degradation. Also, intra-mesh roaming of mobile vehicles was typically not possible due to IP address changes by the client.

Fourth generation - Multichannel Layer 3 backhaul. Aruba has delivered the industry's first fourth generation mesh solution, combining the power of multiple backhaul radios with an RF-aware layer 3 routing protocol inside the mesh. This allows the construction of high-speed mesh “cores” which feed distribution and access tiers. Traffic flows directly where it is needed inside the mesh, without imposing arbitrary paths or bottlenecks inside root nodes that are not the least cost path. Further, AirMesh provides for seamless high speed roaming via a MobileIP-like implementation. In addition, AirMesh includes the unique Virtual Private LAN over Mesh (VPLM). VPLM presents a L2 appearance at the mesh ingress/egress points, while allowing the mesh to operate internally in layer 3 mode. This combines the simplicity and compatibility of L2 with the performance and efficiency of L3.

Realizing the potential of a fourth generation mesh is the subject of most of this Design Guide. In Chapter 9: Planning the Mesh Backhaul Layer on page 119, you will learn how to create an RF design for a multichannel backhaul. In Chapter 11: IP Planning for Aruba AirMesh on page 157, you will learn about the IP planning for the L3 features of AirMesh.



Comparing End-to-End Performance

Single-channel backhaul was the dominant network design for most first and second generation outdoor mesh networks. They remain an appropriate solution for campus extension use cases with low hop counts, but their capacity limitations make them a poor choice for today’s mesh networks that need to deliver high capacity for multiple HD video streams across multiple hops. Traditional single-radio/single-channel multihop links experience a 50% or more decrease in throughput for each network hop. Throughput is decreased because a single channel radio must share the air and repeat transmissions from upstream to downstream nodes and vice versa. Single channel outdoor networks generally use omnidirectional antennas, as shown in Figure 7. Using this strategy, nodes are placed much closer together than the required Wi-Fi coverage dictates due to the lower combined gain of the omni antennas.

Figure 7 50% per-hop throughput loss on single-channel mesh networks

By contrast, it is possible to maintain high end-to-end throughput with low latencies by employing multiple channels in the backhaul network, as shown in Figure 8. This architecture is mandatory as more mesh client devices use 802.11n and as fixed high-bandwidth sources such as video cameras or vehicle-mounted digital video recorders become commonplace. Multichannel mesh networks generally employ directional antennas between individual mesh nodes, creating a mesh from a large number of discrete point-to-point or point-to-multipoint links.

Figure 8 Throughput is maintained when using multiple backhaul channels

arun

_035

3

50 Mb/s100 Mb/s

Throughput25 Mb/s

12 Mb/s

Internet

Ch. 149Ch. 149Ch. 149

Radio 1 Radio 1 Radio 1 Radio 1

Ch. 149

arun

_035

4

100 Mb/s

Throughput

100 Mb/s 100 Mb/s 100 Mb/s

Ch. 161Ch. 153Ch. 157Ch. 149

Internet



Aruba has developed specific antennas, deployment practices, and software calibration controls that work with the mesh routing algorithms to deliver reliable high- capacity RF coverage across very large areas using multi-channel backhaul.

The performance difference between single-channel and multichannel backhaul architectures can be easily demonstrated with any IP load generation tool, such as iPerf or Ixia IxChariot. To illustrate the point, Aruba measured end-to-end throughput across 4 hops using a single-channel and multi-channel configuration. The single-channel testbed used 4 mesh nodes, each with a single backhaul radio using omnidirectional antennas. Figure 9 illustrates the multichannel mesh testbed on which the data in Figure 10 was obtained. Both tests were conducted inside a Faraday cage to eliminate outside interference.

Figure 9 Multichannel mesh testbed

Figure 10 clearly shows the 50% per hop performance limitation of early mesh generations, as well as the ability of AirMesh to maintain nearly constant end-to-end throughput and latency over large distances.

Figure 10 Multi-channel vs. single-channel backhaul performance: four hops

arun

_043

0

MSR1200

MSR 1200MSR2k

Attenuator Attenuator Attenuator AttenuatorR0 R0 R1 R0 R1 R0 R1 R0

MSR2k MSR2k

IXIA


http://www.arubanetworks.com/products/access-points/antennas.php


Multichannel backhaul generally requires that directional antennas be used between radio pairs within the mesh. This topology blends the best of outdoor mesh and point-to-point architectures into a single platform. This is desirable for maintaining end-to-end throughput as shown these figures and also to increase the allowable distance between mesh nodes. For the same range, a radio pair that uses directional antennas can achieve a higher signal-to-noise ratio (SNR) in line of sight (LOS) and non line of sight (NLOS) conditions. Higher SNRs translate directly into higher physical-layer data rates and more overall network capacity.

To keep the management overhead low, AirMesh allows automatic software configuration of each radio using a feature called Radio Frequency Management (RFM). RFM ensures the flexibility to deploy each system using the frequencies, channels, and maximum power that are allowed within each country. AirMesh is a layer 3 system, and RFM is capable of automatically provisioning IP addresses on all multichannel radio pairs. For more information on RFM, see Chapter 7: Aruba Software Technologies on page 73.

ArubaOS AP for Campus ExtensionThis section presents the Aruba AP-175 campus extension access point.

AP-175 (Campus Extension) AP

The multifunction AP-175, shown in Figure 11 is an affordable, fully hardened, outdoor 802.11n AP that provides maximum outdoor deployment flexibility. The high-performance AP-175 delivers wire-like performance at data rates up to 300 Mb/s per radio. The AP-175 is the outdoor radio of choice for Aruba customers with installed controllers that are expanding coverage to adjacent outdoor areas.

Figure 11 AP-175 dual-radio access point

The AP-175 features two 2x2:2 MIMO radios, with one radio dedicated to 2.4 GHz and the other dedicated to 5 GHz.The unit includes four N-type antenna interfaces, two per radio, in an over/under configuration. The AP-175 can be mounted on the wall or on a mast in any outdoor area.

AP

175_

01

1 2 3 4

65

7

8

1 Radio 1 (Antenna 1)

2 USB Console Interface

3 Reserved or Power Interface




7 Ethernet Interface (PoE)

8 Grounding Point


http://www.arubanetworks.com/products/mobility_controllers.php


The AP-175 carries an IP66 rating and has been engineered to operate in harsh outdoor environments. The AP-175 can withstand exposure to high and low temperatures, persistent moisture and precipitation, and is fully sealed for protection from airborne contaminants.

As an 802.11n AP, the AP-175 works with centralized Aruba Mobility Controllers to enable the use of existing role-based authentication systems.

The multifunction AP-175 can be configured through the Aruba Mobility Controller to provide WLAN access with part-time or dedicated air monitoring for wireless intrusion prevention systems.

The AP-175 comes in three different versions, depending on the desired power source. The AP-175P operates from standard 802.3at PoE+. The AP-175AC model is used for a 100-240 volt AC power source. The AP-175DC is used for a 12-48 volt DC power supply or solar bus.

The AP-175 and MSR2000 share the same underlying hardware platform. However, they ship from the factory with different firmware images. An AP-175 cannot be converted to an MSR2000 or vice versa.

AirMesh APs for Outdoor Mesh Networks

MSR4000 Quad-Radio Mesh Router

The Aruba MSR4000 wireless mesh router, shown in Figure 12 delivers high-performance wireless back haul and Wi-Fi access to outdoor environments where wired connectivity is impractical or unavailable.

Figure 12 MSR4000 quad-radio mesh router

The MSR4000 is ruggedized and hardened to withstand extreme environmental conditions, and it is ideal for deployment in metro areas, oil and gas plants, retail centers, business parks, and transportation hubs.

1 Radio 0 (Antenna 2) 6 Radio 1 (Antenna 2)


3 Ethernet Interface 8 Radio 3 (Antenna 1)


5 Radio 1 (Antenna 1) 10 Console Interface



A multiradio, multifrequency architecture combined with adaptive layer 3 technology makes the MSR4000 unique. Together, these features provide unparalleled speed and reliability, low latency, and seamless handoffs for voice, video, and other real-time applications across long-distance, outdoor wireless mesh networks.

The MSR4000 consists of four independent 802.11a/b/g/n radios to create flexible outdoor wireless mesh topologies that can use the 2.4 GHz and 5 GHz bands as well as the 4.9 GHz public safety band. Each radio is capable of providing a maximum throughput of 300 Mb/s.

Each individual radio can be configured to operate as a client access AP or as a point-to-point or point-to-multipoint node to deliver full-mesh backhaul. This four-radio architecture separates client access and mesh backhaul and optimizes radio resources for both types of traffic to ensure high throughput and low latency. The MSR4000 fully participates in the Aruba Adaptive Wireless Routing™ (AWR™) algorithms, which automatically optimize traffic flow between multiple wireless mesh routers for maximum user capacity.

MSR2000 Dual-Radio Wireless Mesh Router

The Aruba MSR2000 dual-radio mesh router, shown in Figure 13 provides unparalleled speed and reliability at the edge of large-scale mesh networks. The two radios deliver low latency and seamless handoffs for voice, video, and other real-time applications across long-distance, outdoor wireless mesh networks.

Figure 13 MSR2000 dual-radio mesh router

The MSR2000 consists of two independent 802.11a/b/g/n radios to create flexible outdoor wireless mesh deployments that use the 2.4 GHz and 5 GHz bands or the 4.9 GHz public safety band. Each radio provides a maximum throughput of 300 Mb/s.

Note: The interface layout for the MSR2000 is identical to the interface of the AP-175 (see Figure 11 on page 26).



Each individual radio can be configured to operate as a client-access AP or as a point-to-point or point-to-multipoint node to deliver full mesh backhaul. If necessary, both radios in the MSR2000 can be configured for backhaul on different channels. This configuration allows the MSR2000 to serve as an unwired relay in a multichannel architecture and maintain high end-to-end throughput and low latency. The MSR2000 fully participates in the Aruba AWR algorithms, which automatically optimize traffic flow between multiple wireless mesh routers for maximum user capacity.

MST200 Single-Radio Wireless Mesh Router

The Aruba MST200 wireless mesh access router is considered a true edge router and connects devices such as video surveillance cameras and IP phones to high-performance Aruba outdoor wireless mesh networks. The MST200 uses the AWR protocol to determine the best path for each device to send data to the wired network.

Figure 14 MST200 single-radio wireless mesh router

The MST200 is also an ideal solution for delivering wired network connectivity to buildings inside a mesh or at the end of a mesh. MST200s can also be used in pairs to construct low-cost, high-throughput point-to-point bridge links between two buildings when a full mesh is not required. The integrated 14dBi dual-polarization 5GHz MIMO antenna greatly simplifies the installation process while providing a clean, attractive look.

The MST200 is ruggedized and hardened to withstand extreme environmental conditions. The MST200 is ideal for deployments in outdoor environments to support applications like video and perimeter surveillance, metro area networks, electronic billboards, and mass transit networks. The

1 Ethernet Interface (PoE In)

2 USB Console Interface

3 Status LEDs

4 Integrated Antenna (14dBi 60°x14°)

1 2

3

4



MST200 is also ideal for public-safety monitoring along transportation corridors and for rapid deployments at large-scale public events or during emergency response.

The MST200 provides a maximum throughput of 300 Mb/s and delivers unprecedented stability and reliability. The MST200 and all MSR routers also employs Active Video Transport™ (AVT™) traffic shaping and load balancing algorithms for use across RF links. These algorithms enable the MST200 to deliver HD-quality video from fixed surveillance cameras to headquarters locations.

MSR1200 Dual-Radio Indoor Mesh Router

The Aruba MSR1200 wireless mesh router, shown in Figure 15 delivers high-performance networking to remote indoor environments where wired connectivity is impractical or unavailable. Typically, large outdoor networks require the ability to deliver wired and/or wireless connections to indoor users through a bridge or repeater node. The MSR1200 is perfect for this application, whether serving as a mesh point or a low-cost mesh portal. When mounted inside a vehicle, the MSR1200 is designed to support seamless roaming with Aruba's MobileMatrix feature and participating in the mesh routing protocols as a mobile mesh point.

Figure 15 MSR1200 dual-radio indoor mesh router front panel LED

Figure 16 MSR1200 dual-radio indoor mesh router rear panel interfaces

1 Power LED

2 Radio 0 port LED

3 Radio 1 port LED

4 Ethernet port LED

1

2

3 4



A multiradio, multifrequency architecture and adaptive layer 3 technology make the MSR1200 a full participant in AWR dynamic routing. Together, the indoor and outdoor APs provide unparalleled speed and reliability, low latency, and seamless handoffs for voice, video, and other real-time applications.

The MSR1200 consists of two independent 802.11a/b/g/n radios to create flexible indoor wireless mesh deployments that use the 2.4 GHz and 5 GHz bands. Each radio can be configured to operate as a mesh portal or mesh point. By providing client access on one radio and using the second radio for mesh backhaul, the client access and mesh backhaul data are separated. Separating this data optimizes radio resources for both types of traffic to ensure high throughput and low latency.

AP Model SummaryTable 2 presents a quick-reference summary of the entire family of Aruba mesh APs presented in this chapter.

Table 2 Outdoor features on each AP

Function / Model AP-175 MSR2000 MSR4000 MST200 MSR1200

Controller-Managed AP with ArubaOS

Autonomous APs with AirMesh

Number of Radios 2 2 4 1 2

4.9GHz Public Safety Band

Outdoor Rating IP66 IP66 IP66 IP66 n/a



Chapter 4: Outdoor Antennas and RF Coverage Strategies

The information provided in this section will help you understand antenna basics and Aruba best practices for how to cover common outdoor environments. For those new to RF engineering, Aruba highly recommends the vendor-neutral Certified Wireless Network Professional training classes and certifications which provide in-depth education on RF fundamentals. For more information, visit www.cwnp.com.

Antenna Beamwidth, Pattern, and Gain Antenna gain is a relative measure of how the antenna compares to an ideal isotropic radiator. The gain of an antenna is specified in dBi, which is the directional gain of the antenna compared to an isotropic antenna. An isotropic antenna is an ideal (theoretical) antenna that spreads energy in all directions (in a sphere) with equal power. You may think of the sun as a good analog for an isotropic antenna.

Figure 17 Isotropic antenna

Antenna gain is often confused with power because the gain of an antenna can increase the transmitted or received signal levels. However, it is important to note that gain is usually only stated as a maximum value and this value will increase signal levels only in a particular direction. This is because antenna gain is achieved only by compressing the radiated power into a tighter region in 3D space, and antennas (by themselves) do not create increased power. Antenna gain is more correctly described as a focusing of radiated power rather than an amplification of it. This means that any antenna with gain > 1 dBi will provide higher signal levels than the isotropic radiator in some directions, but will actually reduce signal levels in other directions. With increasing maximum gain, the area in 3D space with reduced signal level grows inversely with increasing gain. This means that higher gain antennas focus the power into a tighter and tighter region of space, which can actually result in much worse coverage if clients are not in the region of higher gain.

Equal signal strength radiated over a sphere

Aruba Networks, Inc. Outdoor Antennas and RF Coverage Strategies | 33

http://www.cwnp.com


To help visualize the idea of focusing energy in some directions at the expense of others, imagine that the sphere in Figure 17 is a rubber ball. Using the same ball with the same available surface area, how would you be able to stretch the ball farther out? One way is to press down on the top of the ball and squash it down vertically. The same basic shape is kept in the horizontal plane (round), but the ball is forced to stretch, which creates a pancake shape in the vertical direction. Figure 18 represents the concept of the omnidirectional antenna, which achieves a greater coverage distance in the horizontal direction at the expense of coverage in the vertical areas of the radiating sphere.

Figure 18 Omni-directional antenna

What would happen if instead of squeezing the ball, you were to pinch it on one end? This concept is illustrated in Figure 19. The ball is forced into a conical shape whose length depends on how much the body of the cone is compressed. This represents the concept of a high gain directional antenna.

Figure 19 High gain directional antenna

It is not necessary for the cone to face sideways, parallel to the ground. It is also possible to pinch the top of the ball and cause the cone to stretch down towards the ground. This is known as a “squint” or “downtilt” pattern, and will be discussed extensively in the balance of this solution guide as it is Aruba’s preferred antenna type for large outdoor yard and plant environments.



Omnidirectional Antenna Types

Omnidirectional antennas fall into two categories. The classic omni - also known as the “stick” omni due to its appearance - is a tall, thin radome whose length varies with the intended frequency band. Both vertically polarized and horizontally polarized stick omnis are available, including 2x2 MIMO kits that include one of each from Aruba for use in outdoor networks.

The other type of omni is known as the “squint” or “downtilt” omni. The squint is technically a directional antenna because it faces down. However, the antenna is designed to provide standard vertical polarization. It also operates as a full 360-degree omnidirectional antenna in the horizontal plane. The antenna has a very low gain (3-5 dBi) depending on frequency, creating a tight, well formed “cell” with the bulk of the signal focused down toward clients. See Figure 20 for an illustration of these antenna patterns.

While squint antennas are common indoors, Aruba developed and brought to the market one of the first outdoor models. This antenna is the result of our experience of providing coverage to intermodal facilities that cover large areas and that require coverage behind and inside container stacks and mobile equipment. However, this antenna is used in an increasing number of high-capacity outdoor networks. This antenna is intended to be mounted high up—such as on a high mast, light pole, or tall communications tower—where it has good LOS behind most obstructions. This antenna enables wireless designers to use a “dense outdoor deployment” strategy in a manner similar to providing consistent coverage indoors.

Figure 20 H-plane comparison of “stick” omni and down-tilt omni antenna patterns

The horizontal range of the squint antenna is much less than the high-gain antenna due to the lower overall gain as well as the shape and directivity of the pattern.

Pair of 8-dBi high-gain omnis Pair of 3-dBi down-tilt omnis



However, the power of the squint antenna becomes obvious when we consider the E-plane pattern. Figure 21 shows the vertical coverage of the same two antennas, which are mounted at a height of 12 meters (40 ft). One can immediately see that the -67dBm cell edge in the vertical plane does not even reach the ground, whereas the squint omni not only reaches all the way but also has a clear LOS behind any obstructions.

Figure 21 E-Plane comparison of stick omni and squint omni antenna patterns

Directional Antenna Types

Though it is true that higher-gain antennas increase the range in the direction of the antenna gain, it is not true that the signal strength is the same everywhere in that direction. High gain directional antennas - also known as narrow vertical beamwidth antennas - achieve the range by “stretching” the pattern. However, this stretch of the pattern also causes the area of reduced coverage that exists between every antenna and the beginning of its main lobe to stretch out as well, as shown in Figure 22.

Figure 22 Null zone of a narrow vertical beamwidth antenna

This diagram is typical of a 12-14 dBi antenna with an 8-degree vertical beamwidth (hence the term “narrow vertical beamwidth”). It is assumed to be mounted at 30 meters with no down tilt. In the area before the main lobe hits the ground, there will be some illumination by side lobes of the antenna pattern. While there may be some signal, it will be anywhere from 20dB to 40dB lower than inside the main lobe.

Pair of 8-dBi high-gain Omnis Pair of 3-dBi down-tilt Omnis

Ret

ail_

139

447 m1,500 m

Ground level

50% radiated towards sky

50% radiated towards ground



Contrast the size of this area with that of a low-gain directional antenna - also known as a wide vertical beamwidth antenna - as shown in Figure 23. In this case, a 5-dBi, 60-degree sector has a reduced coverage zone of just 50 meters or so from the same mounting height.

Figure 23 Null zone of a wide vertical beamwidth antenna

Effect of Mechanical Down Tilt on Directional Antenna Coverage

Mechanical down tilt is used on a directional antenna that is mounted high up to “aim” it toward its intended coverage zone. Our experience is that professional wireless designers are often casual about the actual angle of the mechanical down tilt. Generally they are content to estimate down tilt based on a quick ground-based visual inspection of a site without fully considering the 3D implications on the shape of the delivered coverage at ground level. However, the actual results of a high mounting height and modest down tilt can often surprise even experienced wireless engineers. The following examples show how important it is to use mechanical down tilt correctly, and where it is not suitable.

Ret

ail_

14053 m

600 mGround level

50% radiated towards sky

50% radiated towards ground



We begin by showing (in Figure 24) the relative horizontal and vertical beamwidths of two commonly used directional antenna types. On the left is a 12 dBi antenna (Aruba ANT-82) and on the right is a 7 dBi antenna (Aruba ANT-83). Both offer 90 degrees of horizontal beamwidth. This makes it easy to see how the increased gain of the higher-gain antenna comes at the expense of vertical beamwidth (60 degrees on the 7 dBi antenna versus only 10 degrees for the 12 dBi antenna). In this example, the antennas were modeled at a height of 30 meters.

The lighter area in the diagram in the upper right (and in the diagrams that follow in this section) shows the main lobe of the antenna in contact with the ground.

Figure 24 Effect of Higher gain on vertical beamwidth

Note how narrow the vertical beamwidth of the high-gain antenna is and that the main lobe does not touch the ground. And while the wider vertical beamwidth of the lower-gain antenna does touch the ground, it is only the bottom portion of the main lobe, meaning that most of the signal is wasted overhead. Both antennas could benefit from mechanical downtilt.

Plan ViewPlan View

Elevation View

12 dBi gain (ANT-82)90 degrees horizontal beamwidth

10 degrees vertical beamwidth

7 dBi gain (ANT-83)90 degrees horizontal beamwidth

60 degrees vertical beamwidth



In Figure 25, 10 degrees of mechanical down tilt is added to a narrow vertical beamwidth antenna on the left (10 degrees) and an antenna with a wider vertical beamwidth antenna (60 degrees) on the right.

Figure 25 Azimuth view with 10 degrees of mechanical down tilt

In Figure 26, the narrow vertical beamwidth antenna on the left sacrifices close-in coverage to achieve greater range. Mechanical down tilt cannot fully compensate for this null area underneath the antenna before the pattern hits the ground.

Figure 26 Elevation view with 10 degrees of mechanical down tilt

On the right, no null area exists, because more of the main lobe of the wide vertical beamwidth antenna now hits the ground.

12-dBi gain: 90 degree 7-dBi gain: 90 degree

12-dBi gain: 90 degree 7-dBi gain: 90 degree



Figure 27 shows the results when the down tilt is further increased to 30 degrees for the narrow vertical beamwidth antenna (the antenna on the left). This is done in an attempt to obtain better coverage close to the AP. The result is a distorted and narrow coverage pattern with even less coverage that actually reaches the ground.

Figure 27 Narrow vertical beamwidth with 30 degrees mechanical downtilt

A common mounting height for outdoor networks is 12-15m (40–50 ft). Even at this relatively modest mounting height, a small mechanical down tilt (10–30 degrees) creates a narrow vertical beamwidth antenna that creates only a small “stripe” of coverage on the ground. This limited coverage is the opposite of what the wireless designer intended, which was to provide uniform coverage throughout the coverage area.

Directional Antenna Conclusions

We have seen in this section that high-gain antennas are primarily intended for long-distance, point-to-point connections, not close-in client coverage. We have further established that:

Vertical beamwidth is more important than horizontal beamwidth in determining the experience of clients.

Mechanical down tilt is not a good solution to compensate for narrow vertical beamwidth. It reduces the size of the main antenna lobe that reaches the ground.

High mounting heights are not compatible with narrow vertical beamwidth antennas due to the size of the null zone between antenna and the 3 dB point.

Low mounting heights are easily obstructed by ground level equipment or buildings.

12-dBi gain: 90 degreehorizontal view

12-dBi gain: 90 degreevertical view



Assuming that the wireless designer is determined to use a narrow vertical beamwidth antenna for client coverage, two methods are available to reduce the size of the null area:

Use mechanical down tilt. However, as we have seen, a relatively small amount of down tilt (just 15 degrees) produces the “striping” affect and reduces the overall coverage area.

Reduce the mounting height. The best way to maximize the coverage area of a narrow-vertical beamwidth antenna and minimize the null is to reduce the mounting height. For this reason, Aruba recommends that high-gain directional antennas that are used for client coverage (as opposed to point-to-point links) should never be mounted higher than about 30 feet with a maximum of about 5 degrees of mechanical down tilt.

It may seem that if you reduce the mounting height of a narrow vertical beamwidth directional antenna, the coverage issues described here would be solved. Unfortunately, doing so renders the main lobe of the signal more vulnerable to LOS obstructions that occur more often at lower mounting heights. The network planner must constantly balance these trade-offs.

RF Coverage Strategies for Outdoor WLANsA coverage strategy is a specific method or approach for locating APs inside a wireless service area. Generally, any given coverage strategy will also call for a specific antenna pattern providing required directionality (even if it is just using integrated antennas in an AP). Three basic coverage strategies are generally used to provide 2.4 GHz and 5 GHz high capacity Wi-Fi coverage in outdoor environments:

Sparse side coverage Dense side coverage Dense overhead coverage

Coverage is sparse when a relatively small number of irregularly-spaced locations cover a large space, often using high-gain, narrow-vertical beamwidth directional antennas. Coverage is dense when many APs are relatively evenly spaced to cover a large area from many locations use lower-gain, wide-vertical beamwidth antennas.

Understanding Side and Overhead Coverage

From a horizontal perspective, sparse and dense coverage are very easy to understand and to visualize. Side and overhead coverage are more complex and will be considered in depth in this section.

Side Coverage

Coverage is from the side when the main lobe of the antenna is approximately the same elevation as the clients being served. If mechanical down tilt is in use, the elevation difference may be greater, but it is still considered side coverage.



Viewed from the side, the main lobe of the antenna pattern spreads out to a precisely engineered limit all around the AP. A common misconception is that each pole-mounted AP serves the area directly below. However, a client standing immediately underneath such an AP using a stick omni will not benefit from the antenna pattern because the main beam is passing overhead. Instead, the client may well be associated to the next AP over. Also, the 50% of the signal that is directed upwards from a typical stick omni antenna is immediately wasted, as illustrated in Figure 28.

Figure 28 Side coverage

Overhead Coverage

Overhead coverage refers to the use of “squint” or “downtilt” omnidirectional antennas that face downwards but are electrically designed to provide a full 360 degrees of coverage with standard vertical polarization, as shown in Figure 29. All of the antenna gain is focused in the direction of the clients underneath.

Figure 29 Overhead coverage

Viewed from the azimuth, or overhead, both antennas provide full 360 degree coverage in a circular shape. However, the downtilt omni will have a smaller, tighter pattern, whereas the side coverage AP will spread its signal further out.

arun

_043

3

Wastedsignal

Reduced coverage area outsidemain antenna lobe

Side Coverage

60°

3dBbeamwidth

10 m

arun

_043

4Overheadcoverage

120°120°

20 m



Choosing Between Side and Overhead Coverage

Side coverage from low-gain directionals or omnis is recommended as the best and lowest-cost solution for campus extension coverage at up to 9 meters (30 feet) of building height. In a standard campus deployment, multiple APs on adjacent buildings work together to provide complete, overlapping coverage of the target area.

For mounting positions higher than 12 m (40 ft), Aruba strongly recommends the use of squint omni antennas. The reason for this is illustrated in the following diagram. For a standard 60 degree directional antenna such as the ANT-2X2-D607 or ANT-2X2-D805, the -3 dB point where the main lobe intersects the ground moves out 5.2 m (17 ft) from the AP for every additional 1 m (3.2 ft) of mounting height. We have already shown that mechanical downtilt is limited in its ability to compensate for increasing height.

Figure 30 Effect of increasing AP height on main lobe reaching ground level

In summary, for steep down angles and mounting heights over 12 m (40 ft) in outdoor areas, the low-gain squint omnidirectional antenna is ideal:

It limits range to a predictable area around each AP and reduces AP-to-AP interference It reduces client density per AP by employing more, smaller cells Its antenna pattern provides users at ground level with a higher signal than APs see to each

other Adaptive radio management functionality is improved for auto-calibration of the RF network and

automation of ongoing operations.

arun

_043

5

60°

90°30°

17 m43 m

69 m

10 m

25 m

40 m

40 msin(30°)

= 80 m=

= 50 m=

= 20 m=

25 msin(30°)

10 msin(30°)

MSR2000MSR2000

MSR2000MSR2000

MSR2000MSR2000

MSR4K/2Kwith

ANT-2x2-D607



Sparse Side Coverage

The sparse side coverage strategy is used when outdoor areas have very limited vertical mounting assets and usable electrical service. We start by using these few existing buildings, towers, and structures that have power and data services. These are also typical locations for other transmitters such as two-way radios and even cellular telephone base stations, so we often co-locate AP-175s or AirMesh routers in the same positions. Figure 30 is a real customer example of a 5 km2 (2 mi2) seaport showing the handful of locations with wired backhaul. Note the uneven distribution of locations throughout the yard, making it impossible to achieve uniform signal levels.

Figure 31 Sparse side coverage example

In this deployment scenario we use very high-gain (≥13 dBi), 60-degree sector, moderate elevation (50 degree) antennas to cover as much range as possible from each radio position.

It’s Aruba’s experience that this strategy alone is inappropriate to deliver reliable outdoor coverage for clients. Frequent LOS obstructions cause signal drop-outs and a poor user experience. The exception to this observation is that side coverage remains a good alternative for covering fixed wireless cameras, which are often at similar elevations. This coverage strategy also does not comply with vendor RF design best practices from Cisco, SpectraLink, or Vocera when planning wireless Voice over IP (VoIP) networks because it is not capable of delivering a consistent -67dBm signal level or predictable roaming transitions throughout a coverage area.



Dense Side Coverage

Dense side coverage networks are most often seen in a campus environment where common areas are surrounded by buildings that are accessible to the network operator. In a yard environment such as pictured in Figure 32 below, dense side coverage can be achieved using existing light poles to mount mesh radios at regular intervals. In these networks, AP-175s or AirMesh routers are deployed densely using omnidirectional or sectored side coverage from buildings or utility poles. In dense side coverage networks the radio density is high, providing good RF reliability because another radio always is working nearby.

Figure 32 Dense side coverage example

Aruba typically recommends mid-gain (5 - 7 dBi) antennas rather than high-gain antennas in this scenario to minimize close-in nulls. The mid-gain antennas deliver consistent client coverage throughout as a result of delivering homogenous signal levels across large areas. These antennas can also deliver good roaming performance. When AP-175s are deployed, ArubaOS can utilize Adaptive Radio Management (ARM). Consistent AP spacing and the homogenous antennas on the building walls enables the system to respond dynamically to ambient RF changes and is good for delivering VoIP coverage.

Dense side coverage radio deployments can be consistent with voice handset vendor best practices documented by Cisco, SpectraLink, and Vocera.



Dense Overhead Coverage

The dense overhead coverage strategy is often seen in transportation, manufacturing and industrial deployments where antennas can be mounted overhead. But it can be equally well applied to metropolitan networks, and offers some advantages in terms of decreasing the channel reuse distance. AP-175s or AirMesh routers are deployed densely and antennas are mounted higher up, between 15-35m (50–120 ft) above ground level. Existing light poles, high masts, and communication towers are used to mount AP-175s or AirMesh routers every 200-300m (650-950ft), resulting in a high number of alternate paths and a very reliable system.

Figure 33 Dense overhead coverage example

Aruba sells a specialized low-gain (typically 3-5 dBi), squint, omnidirectional antenna that faces down to create very uniform cells. These antennas work reliably and deliver consistent performance in cluttered outdoor environments like container ports and rail yards because they usually have clear LOS behind ground obstructions that would block side coverage solutions.

The dense overhead coverage strategy results in excellent voice support and a dense number of radios with LOS to many APs. This strategy is consistent with voice handset vendor best practices.



Selecting an Aruba Outdoor AntennaIn outdoor networks, antenna types are always used for specific purposes. For example, directional antennas are used for each backhaul link and omnidirectional antennas are used for access radios. Aruba has invested heavily in research for MIMO antennas that deliver the highest possible performance even in multipath-poor outdoor environments. The line of Aruba MIMO antenna products represents the state of the art in rate-versus-range performance for outdoor extension and outdoor mesh applications.

Understanding Aruba MIMO Antenna Part Numbers

Aruba has introduced a proprietary line of MIMO antennas for use with the AP-175 and MSR series APs and mesh routers. To minimize cost and maximize performance, these antennas include multiple elements with polarization diversity.

These antennas also introduce a new part number system that should make it easier to select the right part and understand existing networks with these antennas installed. The system works like this:

Figure 34 Guide to Aruba outdoor antenna part numbers

N O T E

Aruba MIMO antennas contain special multiple-polarization arrays that have been designed to maximize decorrelation of MIMO spatial streams, and minimize intra-array coupling between antenna elements. Aruba does not warranty the performance of outdoor networks using non-Aruba antennas. The use of third-party antennas is at the customer’s own risk.

N O T E

Be sure to check whether the models you choose require a separate low-loss RF cable to connect to the AP. Some Aruba antennas include pigtail connectors and may not need RF cables for attaching to the AP. Your Aruba representative can help you determine what parts are necessary.

ANT - NxM - ABCCNxM=

“2x2” for 2x2 MIMO antennas

A = D for dual-band 2 for 2.4 GHz single-band 5 for 5 GHz single-band

B =Single digit representing H-plane 0 = omnidirectional 1 = 10 degrees or less 2 = 20 degrees 3 = 30 degrees 4 = down-tilt

omnidirectional 5 = {reserved} 6 = 60 degrees 7 = 70 degrees 8 = 120 degrees 9 = 90 degrees

CC =Two digits that represent gain in dBi



Based on this system, the Aruba line of dual-band 2x2 MIMO antennas at the time of writing is as follows:

ANT-2x2-D805: 120 Degree Sector, 5 dBi, ±45 Polarization ANT-2x2-D607: 60 Degree Sector, 7 dBi, ±45 Polarization

For squint omni deployments, Aruba offers the following two element antenna: AP-ANT-90: Down-tilt Omnidirectional, 3 dBi, V Polarization

Similarly, the Aruba single-band 2x2 MIMO antennas include: ANT-2x2-2005: 2.4 GHz, Omnidirectional, 5 dBi, H/V Polarization ANT-2x2-2714: 2.4 GHz, 70 Degree Sector, 14 dBi, H/V Polarization ANT-2x2-5005: 5 GHz, Omnidirectional, 5 dBi, H/V Polarization ANT-2x2-5010: 5 GHz, Omnidirectional, 10 dBi, H/V Polarization ANT-2x2-5614: 5 GHz, 60 Degree Sector, 14 dBi, ±45 Polarization

For the latest listing of Aruba’s line of antenna products, visit our web site on http://www.arubanetworks.com/products/access-points/indoor-and-outdoor-antennas/. From this page, you may also wish to download the Aruba Antenna Matrix, which is a handy quick reference guide to the entire Aruba antenna line in table format.

Access Layer Antennas

For access layer radios, omnidirectional antennas can provide good all-around coverage for client devices. These antennas can be applied in outdoor extension or outdoor mesh networks when mounting locations like street lights have a clear view in all directions. The Aruba ANT-2x2-2005 is good for this purpose. It is a kit of two 5dBi 2.4GHz antennas, one horizontally polarized and one vertically polarized.

Aruba's squint antenna for outdoors is the ANT-2x2-D403. This is a 3dBi dual-band antenna with vertical polarization.

For sectored coverage, Aruba offers a range of horizontal beamwidths such as the 5dBi 120 degree ANT-2x2-D805 and the 7dBi 60 degree ANT-2x2-D607. A 14dBi 70 degree option is also available. All Aruba directional antennas feature multiple polarizations.


http://www.arubanetworks.com/products/access-points/indoor-and-outdoor-antennas/


Table 3 shows the Aruba MIMO antenna family typically used for client connectivity. Table 3 Omnidirectional antenna typically for access connections

Vertical Horizontal

ANT-2x2-20055 dBi

Vert. Beamwidth: 30°2.4 GHz

ANT-2x2-50055 dBi

Vert. Beamwidth: 29°5 GHz

AP-ANT-903 dBi

Vert. Beamwidth: 57-61°Horiz. Beamwidth: 360°

Dual-band



ANT-2x2-D8055 dBi

Vert. Beamwidth: 70°Horiz. Beamwidth: 120°

Dual Slant: ±45°2x 30" pigtails

Dual-band

ANT-2x2-D6077 dBi

Vert. Beamwidth: 50°Horiz. Beamwidth: 60°

Dual Slant: ±45°2x 30" pigtails

Dual-band

Table 3 Omnidirectional antenna typically for access connections (Continued)

Vertical Horizontal



Backhaul Layer Antennas

For backhaul radio links, narrow beamwidth MIMO antennas in 5 GHz are popular because more channels are available that are generally much cleaner than 2.4 GHz. In addition, narrow-beamwidth MIMO antennas have improved interference rejection and can achieve higher SNRs based on good LOS. Omnidirectional antennas generally are not used for backhaul links because they are exposed to interference from a full 360-degree radius. For client connections, the Aruba MIMO-based omnidirectional antennas work particularly well because a “pair” includes one vertical and one horizontally polarized antenna. These antennas should be mounted above and below each other to maximize decorrelation of multiple spatial streams. Table 4 shows a typical directional or sectored antenna, typically used for backhaul or point-to-point links. Table 4 Directional antennas typically used for backhaul or mesh links

Vertical Horizontal

High Gain DirectionalANT-2x2-5614

14 dBiVert. Beamwidth 14°

Horiz. Beamwidth 60°5 GHz

ANT-2x2-5010


5 GHz

ANT-2x2-2714


Horiz. Beamwidth 70°

Dual Slant ±45°Linear

2xN-type female 2.4 GHz



When you select the specific antennas to be used for each site, consider both the horizontal and vertical beamwidth for each frequency. Previous sections described the result of poor planning or poor installations using even small amounts of mechanical down tilt. Also, remember that some Aruba Wi-Fi antennas are dual-band and may combine horizontal and vertically polarized antenna elements for improved performance and ease of installation.

Figure 35 Azimuth follows the visible beam of antenna gain

During planning, the antenna azimuth or direction, as shown in Figure 35 should be specified for each location, including combinations of built-in antenna down tilt plus any mechanical down tilt that should be added by the installer using physical adjustments on the mounting brackets. In many cases, it may be necessary to remotely locate the antennas from the AP or AirMesh router. For these sites, make sure to identify the type and length of RF cable with the proper connectors and also adjust the RF link budget to account for the added signal loss from this cable.

When selecting the antenna for each mounting location, refer to the Aruba Antenna Matrix for detailed understanding of the antenna patterns and gain.



Chapter 5: 802.11n Multiple-In and Multiple-Out

The promise of 802.11n networks is to provide “wire like” speeds to the end user, eventually as much as 600 Mb/s per radio. This speed is achievable by using multiple technologies, including the use of multiple-input and multiple-output (MIMO) technology. MIMO technology combines multiple send and receive antennas, and multiple streams of data being sent at the same time. In addition, the 802.11n specification adds new encoding algorithms and wider channels. This all comes together to increase the data transfer rate significantly.

Ratification and CompatibilityThe IEEE ratified the 802.11n amendment in September of 2009, but by that time 802.11n APs and clients based on an early draft of the 802.11n standard were already actively deployed. In many organizations, deployment was driven when the Wi-Fi Alliance® used an early draft of the amendment and certified “draft-n” products as interoperable. Interoperability certification gave customers the confidence to deploy the products. This certification also gave the vendors the ability to start actively producing and deploying 802.11n capable devices.

The devices produced under the pre-n certification are still in production today and all Aruba APs meet the final standard. Backward compatibility between 802.11n APs and legacy clients is a key part of the amendment. Backward compatibility means that stations that previously connected to 802.11a, b, or g APs are still capable of connecting to 802.11n APs. New networks are now being deployed with 802.11n APs even where the clients do not support the standard.

Understanding MIMOUnlike traditional 802.11a/b/g radios, which use single-input and single-output (SISO), 802.11n radios use MIMO technology to increase throughput by increasing the number of radio transmit and receive chains. An AP or client may have up to four transmit and four receive chains, and it is possible to have a different number of transmit vs. receive chains. Figure 36 shows the difference between a SISO and MIMO transmission.

Figure 36 SISO vs. MIMO

arun

_031

2

WirelessChannel

WirelessChannel

Transmitter

SISOAP

Receiver

SISOClient

Single in, single out Multiple in, multiple outTransmitter Receiver

MIMOAP

MIMOClient

Aruba Networks, Inc. 802.11n Multiple-In and Multiple-Out | 53


802.11n Spatial Streams

The concept of spatial streams of data is related to the ability to transmit and receive on multiple radios. More transmitters and receivers allow the AP to send independent streams of data. Much like adding additional lanes to a road, multiple spatial streams allow the wireless AP to transmit more data simultaneously. Spatial streams split data into multiple parts and forward them over different radios, and the data takes different paths through the air. Figure 37 demonstrates the concept of multiple spatial streams of data.

Figure 37 A MIMO transmission with two spatial streams of data

Part of the advantage of MIMO and spatial streams is that APs can use multipath transmissions to their advantage. SISO systems see performance degradation due to multipath transmissions because the multipath may add to signal degradation. However, 11n APs use multipath transmission to reach their full speeds. The delay in the propagation of paths at different rates allows MIMO and spatial streams to be received correctly at the other end of the transmission link. In a SISO system, that delay can cause interference.

Multiple antennas are needed to transmit and receive multiple spatial streams. Depending on hardware, an AP or client can transmit or receive spatial streams equal to the number of antennas it has. However, the AP may have more antennas than spatial streams.

Other 802.11n Technologies to Increase ThroughputTwo spatial streams allow us to double the transmission rate. But this alone is not adequate to get us from 54Mbps in 802.11a/g to 300Mbps with 802.11n. The 802.11n standard includes four other physical-layer technologies that work together to deliver 300Mbps. They are 40 MHz channels, improved OFDM subcarriers, short guard interval, and space-time block coding.

40 MHz Channels

Previously, 802.11 transmissions were transmitted using 20 MHz data channels. Anyone who has deployed an 802.11a/b/g AP has worked with 20 MHz channels, with each AP set to a single, non-overlapping channel. With 802.11n, two channels can be bonded, which actually more than doubles

N O T E

Though many 802.11a/b/g APs have two antennas, they are not capable of using both antennas at the same time. Instead, the two antennas provide diversity. Each antenna receives a different receive signal strength and the AP selects the strongest one to use for each reception. To send a signal, typically the AP uses the antenna that was last used to receive a signal.

arun

_031

3

Stream 1

Stream 2Client



the bandwidth because the guard channels in between also are used. Figure 38 shows the difference is width for a 40 MHz spectral mask as opposed to the 20 MHz mask originally specified for 802.11 transmissions.

Figure 38 Spectral mask, 40 MHz vs. 20 MHz channels

In the 5 GHz band, multiple 40 MHz channels are available, and depending on the regulatory domain, additional channels are available with dynamic frequency selection (DFS) enabled. Figure 39 outlines the available 40 MHz channels in the 5 GHz band. At the time of this writing (January 2011), some channels have recently become unavailable for new AP models.

Figure 39 40 MHz channels in the 5 GHz band

The limited number of channels in the 2.4 GHz band makes 40 MHz channels unsuitable for use. The 2.4 GHz band has only three 20 MHz non-overlapping channels available in most regulatory domains. If a single 40 MHz channel is deployed in the 2.4 GHz band, the channel covers two of the three usable channels. Aruba recommends that 40 MHz channels only be deployed in the 5 GHz band where more non-overlapping channels are available for use. As you can see in Figure 40 a 40 MHz

arun

_031

9

Spectral mask for 40 MHz channel Spectral mask for 20 MHz channel

-30 MHz-20 MHz

-10 MHz fc +10 MHz+20 MHz

+30 MHz

-9 MHz0 dBr

+9 MHz0 dBr

+11 MHz-20 dBr

-11 MHz-20 dBr

+20 MHz-28 dBr

-20 MHz-28 dBr

+30 MHz-40 dBr

-30 MHz-40 dBr

-30 MHz-30 MHz -20 MHz

-10 MHzfc

+10 MHz+20 MHz

+30 MHz+30 MHz

-19 MHz0 dBr

+19 MHz0 dBr

+21 MHz-20 dBr

-21 MHz-20 dBr

+30 MHz-28 dBr

-30 MHz-28 dBr

+30 MHz-40 dBr

-40 MHz-40 dBr

arun

_031

4

36 40

5180

Bandedge

Channel

Frequency (MHz)

Bandedge

5150 53505200

44 48

5220 5240

52 56

5260 5280

60 64

5300 5320

149 153

5745

Bandedge

Channel

Frequency (MHz)

Bandedge

US UNII I and UNII II bandsUNII I: 5150-5250 MHzUNII II: 5250-5350 MHz8x 20 MHz channels4x 40 MHz channelsUNII II requires DFS

UNII II extendedUS intermediate band

5450-5725 MHz11x 20 MHz channels5x 40 MHz channelsRequires DFS Channels in red are currently unavailablein the US, Canada, and the EU due to new regulatory requirements.

US UNII III / ISM band

Channels defined for 5 GHz band (US regulations), showing common 20 MHz channel plan and 40 MHz options

5725-5850 MHz5x 20 MHz channels2x 40 MHz channels5725 58505765

157 161 165

5785 5805

100 104

5500

Bandedge

Channel

Frequency (MHz)

Bandedge

5450 57255520

108 112

5540 5560

116 120

5580 5600

124 128

5620 5640

132 136 140

5660 5680 5700



channel overlaps two of the three available channels in the 2.4 GHz frequency band.

Figure 40 The 2.4 GHz band is not suitable for 40 MHz channels

Improved OFDM Subcarriers

Orthogonal frequency-division multiplexing (OFDM) is the encoding scheme that is used in Wi-Fi transmissions. OFDM splits a single channel into very small subcarriers that can transport independent pieces of data as symbols. Each symbol represents some amount of data, which depends on the encoding scheme. The data subcarrier count has increased from the original 48 to 52 subcarriers in 20 MHz channels and 108 subcarriers in 40 MHz channels. This increase means that more data channels are available to carry traffic. Each additional subcarrier can carry data over the channel, which increases throughput. In Figure 41 you can see the difference in sub-carriers that 802.11n brings to 20 MHz channels, as well as the number of carriers available with 40 MHz channels.

Figure 41 Increase in subcarriers increases throughput

WARNING

Aruba recommends that customers do not use 40 MHz channels in the 2.4 GHz band due to the lack of available bandwidth and high chance of interference with legacy 802.11b/g networks. While it is possible to enable these channels, the end result is fewer overall channels and a decrease in throughput.

arun

_034

5

Channel 1 2 3 4 5 6

1 6

7 8 9 10 11

11

12 13

Center frequency 2412 2417 2422 2427 2432 2437 2442 2447 2452 2457 2462 2467 2472

3 4 5 6 7 8 9

arun

_031

7

57carriers

-20 MHz -10 MHz +20 MHz+10 MHzCenter frequency

57carriers

28carriers

-10 MHz +10 MHz

28carriers

26carriers

-10 MHz +10 MHz Center frequencyCenter frequency

26carriers

52 subcarriers (48 usable) for a 20 MHznon-HT mode (legacy 802.11a/g) channel

114 subcarriers (108 usable) for a 40 MHz HT mode (legacy 802.11n) channel

56 subcarriers (52 usable) for a 20 MHz HT mode (802.11n) channel



To see how this directly affects data rates, Table 5 shows the difference in speeds that occur from legacy rates to high throughput (HT) rates. Wi-Fi engineers can use this information to compare rates used under 802.11a/g to the new HT rates used in 802.11n. For more information about this comparison, see Modulation and Coding Scheme Index.Modulation and Coding Scheme Index on page 61.

Space Time Block Coding and Maximal Ratio Combining

MIMO also uses diversity techniques to improve the performance. Between two communicating stations, one station can have more antennas than the other. If there are more transmit antennas than receive antennas, Space Time Block Coding (STBC) can be used to increase the signal-to-noise ratio (SNR) and the range for a given data rate. For STBC, the number of transmit antennas must be greater than the number of spatial streams.

The operation of Maximal Ratio Combining (MRC) is dependent on the number of available receive radio chains. When there is more than one receive chain, the MRC technique combines the signals received on multiple antennas. The signals can come from one or more transmit antennas. When the signals are combined, the SNR is improved and the range for a given data rate is increased.

Short Guard Interval

The guard interval is the spacing between OFDM transmissions from a client. This interval prevents frames that are taking a longer path from colliding with subsequent transmissions that are taking a shorter path. A shorter OFDM guard interval between frames, from 800 ns to 400 ns, means that transmissions can begin sooner in environments where the delay between frames is low.

Understanding MAC Layer ImprovementsMoving up the OSI reference model, the 802.11n standard also includes several MAC-layer technologies to greatly improve the efficiency and throughput of wireless transmissions. These are A-MSDU, A-MPDU and block acknowledgements.

Table 5 802.11a/g vs. 802.11n (one spatial stream) HT rates with 800 ns guard interval

802.11a/g 802.11n (1 SS)

6 6.5

12 13.0

18 19.5

24 26.0

36 39.0

48 52.0

54 58.5

N/A 65.0



A-MSDU

Aggregate MAC Service Data Unit (A-MSDU) allows stations that have multiple packets to send to a single destination address and application to combine those frames into a single MAC frame. When these frames are combined, less overhead is created and less airtime is spent on transmissions and acknowledgements. A-MSDU has a maximum packet size of 7935 bytes. Figure 42 shows how A-MSDU aggregation occurs.

Figure 42 A-MSDU aggregation

A-MPDU

Aggregate MAC Protocol Data Unit (A-MPDU) combines multiple packets that are destined for the same address but different applications into a single wireless transmission. A-MPDU is not as efficient as A-MSDU, but the airtime and overhead is reduced. The maximum packet size is 65535 bytes. Figure 43 shows the operation of A-MPDU operation.

Figure 43 A-MPDU aggregation

arun

_031

5

Applications

MSDU (MAC Service Data Unit)

MAC processing

MPDU (MAC Protocol Data Unit)

PHY layer

MAC processing

Aggregated MSDU format (A-MSDU)

P1 P2 P3

P3P2P1

P3P2P1MACheader

arun

_031

6

Applications

MSDU (MAC Service Data Unit)

MAC processing

MPDU (MAC Protocol Data Unit)

PHY layer

MAC processing

Aggregated MPDU format (A-MPDU)

P1 P2 P3

P3P2P1MACheader

MACheader

MACheader



Block Acknowledgement

Block acknowledgements confirm that a set of transmissions has been received, such as from an A-MPDU. Only the single acknowledgement must be transmitted to the sender. Block acknowledgements also can be used to acknowledge a number of frames from the same client that are not aggregated. One acknowledgement for a set of frames consumes less airtime. The window size for the block acknowledgement is negotiated between AP and client. Figure 44 shows the two cases of block acknowledgement in action.

Figure 44 Block acknowledgement of multiple frames

Putting It All Together – From 54 Mb/s to 600 Mb/sFigure 45 illustrates how 802.11n increased transmission speed so dramatically by showing how the technologies of the 802.11n standard are combined to increase throughput. As each of these technologies is combined, the speed increases dramatically.

Figure 45 MIMO increases data throughput to APs up to 600 Mb/s

arun

_031

8

Aggregate MPDU is a special case requiring block acknowledgement

Block acknowledgement covers many framesin one acknowledgement

P4 header

P1P2P3 headerheader Ack P1, P2, ... P3

header Ack P1, P2, ... P4P3 header P2 header P1 header

arun

_031

1

4 x 150300 Mb/s -2 spatial streams,

2x2:2 minimum

450 Mb/s - 3 spatial streams,

3x3:3 minimum

600 Mb/s - 4 spatial streams,

4x4:4 minimum

Original802.11a,g

OFDM54 Mb/s

Improved OFDM 52 subcarriers vs.48 for original

65 Mb/s

40 MHz channelsinstead of 20 MHz

135 Mb/s

Reduced guard interval between OFDM symbols400 ns instead of 800 ns

150 Mb/s

11n allows for up to 4 streamsof MIMO spatial multiplexing

600 Mb/s



802.11 TerminologyWhen discussing 802.11n networks, certain terms are commonly used by engineers to denote the general capacity of a system, and the instantaneous connection rates between individual stations.

Transmit, Receive, and Spatial Stream Designation

In 802.11a/b/g, only a single antenna and a single stream of data are involved. But 802.11n adds multiple antennas and multiple streams of data to increase the transmission capabilities of APs and stations. It is important to understand the nomenclature that is used to describe the capabilities of the system to transmit data at certain rates. Figure 46 shows this nomenclature.

Figure 46 Transmit, receive, and spatial stream nomenclature

Transmit: The number of antennas that are dedicated to transmitting data. Receive: The number of antennas that are dedicated to receiving data. Spatial streams: The number of individual data streams that the radio is capable of

transmitting. An 802.11 a/b/g AP (1 x 1 : 1) is capable of one stream of data, or one transmission, to a client at a time. An 802.11n AP is capable of transmitting multiple streams of data at the same time to the same client. The number of spatial streams must be less than or equal to the number of transmit or receive antennas, depending on which way traffic is flowing.

arun

_030

4

Number oftransmit antennas

Number ofreceive antennas

Number ofdata spatial streams



Modulation and Coding Scheme Index

The modulation and coding scheme (MCS) index is used to arrive at the data rates for a connection. When speeds are discussed, the MCS rate is often used as a short hand for the modulation type and spatial streams. The actual data rate is dependent on the guard interval and channel width as well. The network engineer can determine the maximum expected connection speed of the client if the following information is known:

Number of spatial streams Modulation type in use Channel width Guard interval

Table 6 shows the change where one spatial stream is used to map older 802.11 a/g rates to newer 802.11n 1x1:1 rates with an 800 ns guard interval. It includes the modulation method and MCS rate. Use this information when examining the full MCS rate table that follows.

Table 6 802.11a/g vs. 802.11n (one spatial stream) HT rates with 800 ns guard interval

802.11a/g Modulation Method 802.11n (1 SS) MCS (1 SS)

6 BPSK 1/2 6.5 0

12 QPSK 1/2 13.0 1

18 QPSK 3/4 19.5 2

24 16QAM 1/2 26.0 3

36 16QAM 3/4 39.0 4

48 64QAM 1/2 52.0 5

54 64QAM 3/4 58.5 6

N/A 64QAM 5/6 65.0 7



Table 7 shows the HT rates used in 802.11n. Some steps in the higher rates have been skipped over, but the same process of speed increases repeats with higher the higher rates. Thinking back to Figure 45 with the pipes, these rates show the active combinations of modulation, channel width, guard interval and spatial streams. The data rate can change on a packet-by-packet basis depending on the environment and other factors.

2.4 and 5 GHz Support

For an engineer with Wi-Fi experience but new to 802.11n, it can be difficult to understand that the amendment is not synonymous with the frequencies on which the network operates. Previously, the most commonly referenced amendments to the 802.11 standard operated in only one band. 802.11a networks operated only in the 5 GHz band and 802.11b/g networks operated only in the 2.4 GHz band. The amendment and band could be referred to interchangeably when discussing network operations. However, 802.11n applies to the 2.4 and 5 GHz bands.

Table 7 802.11n MCS index

Index Spatial Streams

ModulationType

20 MHz Channel in Mb/s 40 MHz Channel in Mb/s

800 ns GI 400 ns GI 800 ns GI 400 ns GI

0 1 BPSK 6.5 7.2 13.5 15.0

1 1 QPSK 13.0 14.4 27.0 30.0

2 1 QPSK 19.5 21.7 40.5 45.0

3 1 16-QAM 26.0 28.9 54.0 60.0

4 1 16-QAM 39.0 43.3 81.0 90.0

5 1 64-QAM 52.0 57.8 108.0 120.0

6 1 64-QAM 58.5 65.0 121.5 135.0

7 1 64-QAM 65.0 72.2 135.0 150.0

8 2 BPSK 13.0 14.4 27.0 30.0

9 2 QPSK 26.0 28.9 54.0 60.0

10 2 QPSK 39.0 43.3 81.0 90.0

11 2 16-QAM 52.0 57.8 108.0 120.0

12 2 16-QAM 78.0 86.7 162.0 180.0

13 2 64-QAM 104.0 115.6 216.0 240.0

14 2 64-QAM 117.0 130.0 243.0 270.0

15 2 64-QAM 130.0 144.4 270.0 300.0

Repeats for Three Streams

23 3 64-QAM 195.0 216.6 405.0 450.0

Repeats for Four Streams

31 4 64-QAM 260.0 288.9 540.0 600.0



When we discuss clients and APs, it is important to specify the band that each can operate on in addition to the 802.11n features that are supported. For single radio APs, the categorization is 802.11a/n for 5 GHz and 802.11b/g/n for 2.4 GHz, which signifies that 802.11n speeds and features are available in each band.

Backward CompatibilityThe 802.11a/b/g only APs and clients are no longer being produced in many cases, so it is important to understand that backward compatibility for legacy clients is built into the 802.11n standard. Many organizations have specialized devices that have not yet reached the end of their service and that cannot be upgraded to take advantage of the new 802.11n standard. In these cases, the 802.11n APs will continue to support legacy devices by default in a compatibility mode.

In much the same manner as 802.11g is able to co-exist with 802.11b, older clients will cause APs to use a compatibility mode so that they can work with legacy clients. This degradation of service has the expected results on performance: all clients are forced to operate around the lowest common denominator in the area. Slower clients require that faster clients perform certain portions of the transaction, such as a request to send data, at a lower speed. Faster clients are unable to operate at optimal speeds. To combat the loss of throughput that is experienced by 802.11n devices, the Aruba Adaptive Radio Management (ARM) feature implements airtime fairness to prevent slower legacy clients from starving higher-speed clients.

Maximizing Rate vs. Range with MIMO OutdoorsTo achieve the highest PHY-layer data rates, MIMO relies on unique spatial streams arriving at the receiver from more than one direction. These multiple inputs and multiple outputs can be used to dramatically increase the link capacity. However, in the outdoor environment it is difficult, if not impossible, to predict the spatial diversity that can be expected on any given link. If some multiple inputs and multiple outputs are occurring due to surrounding buildings or terrain, it is likely the AP will experience some benefit. On the other hand, if the AP is connecting to another AP across a wide open flat area, there is no guarantee that the receiver will see anything but the single originally transmitted signal. The next section describes what modern outdoor RF planners should consider when performing active surveys for MIMO networks.



Direct vs. Indirect Multipath

When designing outdoor MIMO networks, it is vital to understand different types of multipath and what types are helpful to maximize performance. Two types of multipath are available:

Direct Multipath: created between the direct paths of the antennas Indirect Multipath: created by scattering and reflections in the environment

Figure 47 Direct multipath example

Figure 48 Indirect multipath example

Compare Figure 47 and Figure 48. Direct multipath is created by the use of multiple antennas and indirect multipath is created by reflections in the environment. Though these figures show only the paths for Antenna 1, the 802.11n standard can use the full matrix of direct and indirect paths between antennas 1 through 6.

6

1,6

1,5

1,4

5

4

3

2

1

arun

_042

5

Tower 1 Tower 2

6

1,6

1,5

1,4

Ground

5

4

3

2

1

arun

_042

5

Tower 1 Tower 2



Based on Aruba research and customer deployments, it is expected that typical long-range bridging links will have the following characteristics that will limit the available throughput based on multipath:

A lack of significant indirect multiple paths based on mounting height Inability to discriminate the direct multiple paths with increasing link distance

For shorter range links, indirect multiple paths may also be low compared to indoor environments. However, discrimination of the direct multiple paths should make multiple direct paths available if the antenna spacing is adequate on the tower (or horizontally on a boom), which means that link capacity is increased significantly.

Correlation and Decorrelation

Correlation is a complex mathematical parameter that is related to the statistical degrees of freedom between the antennas in a MIMO array and the paths available. Attempts to decorrelate paths in an 802.11n setting are related to the available multipath (direct and indirect) and the patterns and polarizations of the antennas used. Ideally, all direct and indirect paths multiple paths between antennas 1 through 6 would be available and uncorrelated. This deployment would result in the highest possible throughput.

However, in a real environment, the objective of decorrelation and the multiple available paths between the antennas can be inversely related. For example one way to decorrelate antennas 1 through 3 in the Figure 47 would be to use directional antennas and aim the antennas in three different directions (120 degrees separated). Similarly, antennas 4 through 6 could be aimed in different directions. This configuration would decorrelate the antennas at both ends of the link, but at the same time may reduce the number of available paths between the antennas. For example if fewer of antennas 4 through 6 are in the direct or indirect pattern of antenna 1, the throughput could drop significantly due to the lack of available multiple paths.

For the outdoor case, direct multipath is the more important consideration, leaving Polarization diversity as the best available technique for decorrelation of small MIMO arrays.

Polarization

Polarization is a description of the orientation of the electric field within an electromagnetic wave. Linearly polarized antennas produce a wave that has a polarization which is commonly referenced to the surface of the earth in outdoor applications. If the electric field is horizontal with respect to the Earth’s surface, the wave is termed “horizontal polarization”. Similarly, if the electric field is vertical, it is considered “vertical polarization”. Polarization diversity can be used in MIMO systems as a method of reducing correlation. Because antennas in a typical MIMO array are linearly polarized, it is often possible to isolate reception between horizontal and vertical polarization by 10-15 dB. This isolation results in a high degree of decorrelation between the antennas. Although the use of decorrelation through polarization diversity could be helpful to increase throughput, it is also possible that decorrelation through polarization diversity could reduce the number of available direct multipath opportunities and thus reduce throughput in some settings.



Leveraging Polarization Diversity to Improve Decorrelation

In indoor environments, it is generally understood that multipath created by reflections from walls and objects can enhance throughput by MIMO system by providing the ability to support multiple de-correlated spatial streams. In outdoor links, the availability of multiple beneficial reflections can not typically be assumed. In fact, in many outdoor designs, clear Line of Sight (LOS) is generally recommended to avoid signal degradation due to attenuation and obstructions in the environment. In the worst case example for an outdoor clear line of sight link, the direct radio path may be the only path available with no useful reflections to enhance MIMO performance. In general, at long distances from the AP and assuming relatively short spacing between the MIMO antenna elements, the outdoor case could revert to SISO data rates, meaning that all available paths between the antennas are “correlated” and only a single independent data stream can be supported.

Fortunately, however, the use of multiple polarizations for transmit and receive can be substituted for multipath reflections to enhance MIMO decorrelation and recover multiple data stream MIMO performance. For this reason Aruba outdoor accessory MIMO antenna arrays use multiple polarizations. This better ensures that multiple data streams can be supported over the available range of coverage.

N O T E




Chapter 6: AP Selection for Common Outdoor Topologies

Aruba offers a flexible product line for designing reliable, high-capacity, outdoor networks. This section reviews some typical application use cases and describes the hardware options that a wireless architect might select for each case.

Single-Radio Point-to-Point Bridge: MST200The MST200, shown in Figure 49 provides a low-cost point-to-point connection between any two remote Ethernet networks. The integrated 14dBi multiple-polarization antenna in the MST200 provides for simplified installation and good aesthetics. AirMesh operating system software provides for a complete PtP solution with no additional hardware required.

Figure 49 MST200 operating as a point-to-point bridge

Single-Radio Leaf Node: MST200When operating as a leaf node at the mesh edge, the MST200 is a low-cost solution to expand the mesh network and increase network capacity.

Figure 50 MST200 operating as a leaf node

The long range of the MST200 enables flexible deployments for wired devices or low-power Wi-Fi clients. With proper high-gain antennas, two MST200 or MSR2000/4000 Series routers can establish mesh links over more than 5 km distance and can achieve 100 Mb/s or more of TCP/IP throughput at distances of up to 1 km.

IPnetwork

IPnetwork

MST 200

Mesh link

MST 200

arun

_040

0ar

un_0

401

IPnetworkMesh

link

MST 200

MSR4K/2K

MSR4K/2K

Aruba Networks, Inc. AP Selection for Common Outdoor Topologies | 67


Dual-Radio Outdoor-to-Indoor: MSR1200Figure 51 shows the MSR1200. Most frequently, the MSR1200 is used to connect to an outdoor mesh network and to distribute coverage inside select buildings. This scenario can also be reversed such that the MSR1200 can serve as a low-cost mesh portal mounted indoors, connected to antennas that are mounted outside the building using low-loss coaxial cable.

Figure 51 MSR1200 mounted indoors with outdoor antennas for backhaul

Dual-Radio Mobile Applications: MSR1200 or MSR 2000Though it is not ruggedized for outdoor installations, the MSR1200 can be installed inside the cabin or cargo area of vehicles, with antennas mounted remotely. Unlike the MSR1200 that must be mounted in a dry area on a vehicle, the MSR2000 can be exposed to the elements, which makes mounting more flexible on larger public safety and other vehicles. By using both radios, one for client access and one for backhaul, the MSR2000 and MSR1200 can be used to create a mesh network to support mobile clients.

Figure 52 MSR1200 or MSR2000 mounted inside a vehicle compartment

arun

_040

2

IPnetwork

MSR 4K/2K

MSR 1200

Mesh link

arun

_040

3

IPnetwork

Mesh link

MSR 1200in car

MSR 4K/2K



Dual-Radio Client Access: AP-175 and MSR2000Either the AP-175 or the MSR2000 can be used to establish a reliable outdoor Wi-Fi hot-spot with dual-band coverage.

Figure 53 AP-175 or MSR2000 mounted to a building or wired pole to provide campus extension

Single Hop Point-to-Point: AP-175 or MSR2000Either the AP-175 or the MSR2000 can be configured as a mesh portal and connected to the wired network (see Figure 54). The AP-175 is used to extend an Aruba indoor network to outside areas of relatively modest coverage, like a university or business campus. The MSR-2000 is typically used when larger areas that require more mesh points are anticipated.

Figure 54 AP-175 or MSR2000 configured to provide one-hop mesh to provide remote client service

IPnetwork

MSR 2Kor AP-175

IPnetwork

Mesh linkMSR 2K

or AP-175MSR 2K

or AP-175

arun

_039

7



Multi-hop Linear Mesh: MSR2000The MSR200 can be used to alternate channels and extend a very long bridged connection, perhaps around physical obstacles or over very long distances (see Figure 55).

Figure 55 MSR2000 extends a long bridged connection

Parallel Point-to-Multipoint: MSR2000 or MSR4000Using AWR, any MSR radio can maintain a maximum of six mesh links with six different neighbors, as shown in Figure 56. For simple networks, the MSR2000 can be used to create a shared point-to-multipoint network for connecting remote sites to a core site. The second radio is used on the mesh points to provide client devices access to the network. A single MSR4000 providing sectored coverage on each radio can support up to 24 directly-connected child mesh points.

Figure 56 MSR2000 maintaining multiple mesh links

IPnetwork

Mesh linkMesh link

MSR 2K MSR 2K

MSR 2K

arun

_039

9

Ch. 157Ch. 149

arun

_039

8

IPnetwork

MSR 2K

MSR 2K

MSR 2K

Mesh links



High Capacity Mesh Core: MSR4000The MSR4000 quad-radio router, shown in Figure 58 is purpose-built to operate as a high-capacity mesh portal or mesh point. Aruba customers often deploy a network of MSR4000s as a mesh “core”, leveraging multichannel mesh to maintain end-to-end performance. Then, MSR2000s are used to create a mesh “distribution layer” which in turn connects with MSR1200 and MST200 edge access nodes. This can be visualized in Figure 57.

Figure 57 Using MSR4000 to create a multi-tier mesh with dedicated mesh 'core' for backhaul

Each radio can be configured to operate in the 2.4 GHz, 5 GHz, or 4.9 GHz bands. Careful channel assignments and frequency reuse plans allow networks to scale to very large sizes and sustain very high throughput even across many mesh hops.

Figure 58 MSR4000 operates as a high-capacity mesh portal or mesh point

arun

_035

2

Wirednetwork

MST200MST200

MST200

MSR2000

MSR1200

MSR4000s

Mesh core

Edge access

MST200MST200

MST200

MSR1200MSR1200

MSR1200MSR1200

IPnetwork

MSR 4K

MSR 4K

MSR 4K

MSR 4K

arun

_039

6



Remote Thin AP Endpoints Overlaid on AirMesh Figure 59 shows a standard ArubaOS controller-based thin AP network that is overlaid on an AirMesh routed network connecting different facilities, combining an ArubaOS and Aruba AirMesh solution in order to cover difficult areas for signal.

Figure 59 ArubaOS overlay network using AirMesh routers

In this example, a customer wants to use standard Aruba APs in remote buildings in a campus or yard environment that require multiple hops to reach. These buildings do not have other wide-area connectivity, for instance guard shacks or off-campus housing. By combining the ArubaOS and Aruba AirMesh solutions, the customer can operate a converged network that utilizes the centralized security model to areas that otherwise are difficult to cover with a typical outdoor network extensions.

IP network

RAP5

2.4GHz 4.9GHz

Apartment

Livingcommunity

CAP/RAP

S3500wired AP

Campusbuilding

Arubacontroller

MSR4K/2K

MSR 4K/2K

MSR4K/2K

2.4GHz

4.9GHz

2.4GHz4.9GHz

arun

_043

6

44



Chapter 7: Aruba Software Technologies

In general, as we have seen, the deployment type will usually determine the choice of AP model and operating system. However, some simple topologies such as single hop mesh networks can work equally well with AP175 or MSR2000. This chapter describes some of the unique outdoor networking software features - with a focus on Aruba's AirMesh operating system - to help the wireless architect choose between product families.

Choosing an Outdoor Operating SystemAs you saw in previous chapters, the deployment type generally drives the decision of AP hardware, which in turn determines which OS will be used. Campus extension networks almost always use ArubaOS with APs managed by a controller. Outdoor mesh networks almost always use Aruba AirMesh with a decentralized control plane and full layer 3 routing inside each mesh node. Table 8 summarizes the features that are available on each operating system.

Table 8 Feature comparison of Aruba outdoor APs by operating system

Operating SystemArubaOS AirMesh

AP-175 MSR2000 MSR4000 MST200 MSR1200

AP Type Thin Autonomous

Controller required

Centralized policy enforcement

Layer 2 mesh

Layer 3 mesh with AWR

Role-based access control

Client authentication with 802.1X and WPA2

Active Video Transport

Any radio / Any band

MobileIP Roaming IP Mobility MobileMatrix MobileMatrix MobileMatrix MobileMatrix

Dynamic radio management ARM RFM RFM RFM RFM

4.9 GHz band support

Aruba Networks, Inc. Aruba Software Technologies | 73


ArubaOS for Campus ExtensionArubaOS is the operating system and application engine for Aruba mobility controllers and thin mesh APs. Aruba customers that are extending their network outdoors can simply increase the AP license counts to cover the outdoor APs and maintain existing authentication mechanisms, while retaining full advantage of all controller-based features. The software architecture of ArubaOS is designed for scalable performance, and is built using three core components:

First, a hardened, multicore, multithreaded supervisory kernel manages administration, authentication, logging, and other system operation functions.

Second, an embedded real-time operating system powers dedicated packet-processing hardware, and implements all routing, switching, and firewall functions.

Third, a programmable encryption/decryption engine built on dedicated hardware delivers client-to-core encryption for wireless user data traffic and software VPN clients.

The ArubaOS secure enterprise mesh solution is an effective way to expand network coverage for outdoor and indoor enterprise environments using layer 2. Using mesh, you can bridge multiple Ethernet LANs or you can extend your wireless coverage. As traffic traverses across mesh APs, the mesh network automatically reconfigures around broken or blocked paths. This self-healing feature provides increased reliability and redundancy: the network continues to operate if an AP stops functioning or a connection fails.

Aruba controllers provide centralized configuration and management for APs in a mesh environment; local mesh APs provide encryption and traffic forwarding for mesh links. Aruba thin APs can be configured as mesh portals or mesh points. Mesh portals are wired to the network infrastructure. APs that are connected to the wired network occasionally are referred to as “gateways” or “root nodes.” Their primary function is to accept connections from unwired mesh points to expand coverage even when wired network access is unavailable.

Provisioning mesh APs under ArubaOS is similar to thin APs; however, there are some key differences. Thin APs establish a channel to the controller from which they receive the configuration for each radio interface. Mesh nodes, in contrast, get their radio interfaces up and running before making contact with the controller. This requires a minimum set of parameters from the AP group and mesh cluster that enables the mesh node to discover a neighbor to create a mesh link and subsequent channel with the controller. To do this, you must first define and configure the mesh cluster profile before configuring an AP to operate as a mesh node.

For much more information on the mesh features of ArubaOS, please refer to the ArubaOS 6.0 User's Guide.

N O T E

ArubaOS is typically selected for outdoor networks in two scenarios. First, when providing dual-band client access to outdoor areas around an existing ArubaOS indoor deployment. Second, when there is a desire to use the controller to enforce role-based access control in an outdoor mesh environment.



AirMesh for Outdoor Mesh NetworkingThe challenge of providing high-speed Wi-Fi to users across large outdoor areas requires a different approach than distributed thin APs connecting over reliable wired links to a scalable controller. Each unwired mesh node must participate in an exchange of reachability information and examine path metrics to make intelligent forwarding decisions. This exchange must occur even during periods where it may not be possible to communicate with a centralized controller. In addition, the user community on large outdoor networks is not typically managed by a single centralized entity, so authentication must become more highly distributed and flexible.

To meet these challenges, Aruba developed a purpose-built mesh operating system called AirMesh. AirMesh has many technologies that were specifically created to address the hard problems of delivering high capacity reliably across large geographic areas:

Radio Frequency Management (RFM) - RFM is a patented intelligent technology that successfully brings up an entire mesh network including provisioning of IP addresses and subnets, channel assignments and radio power levels. Once the mesh is up, it ensures extremely fast convergence due to topology changes, and continuously optimizes the network to mitigate RF interference.

Adaptive Wireless Routing™ (AWR) – AWR is a patented routing protocol designed for wireless networks that provides RF-aware, layer 3 network intelligence and fast convergence that optimizes the traffic flow in a mesh network.

Path Distance Factor (PDF) – PDF is the underlying link metric protocol and algorithm used to estimate both RF channel conditions and IP network distances. It is used by both RFM and AWR.

Active Video Transport™ (AVT) – AVT is a patented technology that optimizes and prioritizes video traffic. AVT significantly improves video quality by reducing packet loss and jitter. AirMesh delivers video at 30 frames per second.

Virtual Private LAN over Mesh™ (VPLM) – Customers used to have to choose between the simplicity of layer 2 mesh tree topologies or the robustness of layer 3 any-to-any packet forwarding. With VPLM, the mesh presents a standard L2 VLAN that is tunneled across a transparent L3 mesh, giving the best of both worlds.

MobileMatrix™ – This Aruba capability enables reliable roaming between mesh routers in less than 50 milliseconds so that users’ application sessions are maintained even if they are moving at high speed

In this section, we look in detail at the operation and algorithms used by all six of these technologies.



Radio Frequency Management

Radio Frequency Management (RFM) is the software module that performs neighbor discovery and automatic channel assignment. This technology automatically provisions IP addresses on all routers and links, maintains link status, and reports dynamic changes to ensure that no portion of the network becomes isolated from a mesh portal. RFM is tightly coupled with the AWR protocol and notifies the routing algorithm of channel changes or when new mesh nodes are detected. Though it appears in the middle of the software process stack in Figure 60, we begin with RFM here as it makes both AWR and PDF easier to understand.

Figure 60 Interprocess communication of key AirMesh modules

Neighbor discovery is a good example of how RFM, AWR and PDF work together to operate the mesh transparently. The automatic neighbor discovery feature is an “always on” process in the PDF layer where each radio performs an active scan of configured channels to find a valid neighbor. If one is not found within approximately 5 seconds, a periodic discovery process kicks in that continues to scan for available neighbors until one is found. RFM synchronizes channel use across the cells in an area every time a radio changes channels. If multiple radios send “channel changing” announcements at the same time, the announcement precedence value decides which link each mesh node should select.

Each radio runs the neighbor discovery process independently and each radio reports the discovery outcome to the RFM process. After validation, radios compute path metrics for IP route selection and choose from a list of candidates that are selected by a defined priority list from the pool of neighbors within radio range. IP addresses can be generated automatically or they can be generated statically using DHCP reservations based on the MAC address of the radio. After a candidate is selected by RFM, the link is established and AWR is notified.

arun

_054

8

Table updatesProtocol messages

PDF valueannouncements

Exchange neighborinformation

PDF metrics

Channel changesNew mesh nodes

Layer 3

Layer 2

AWR

RFM

Layer 2

Node A Node B

Via IP

Via beacons

Via 802.11 dataframes

• RF-aware L3 routing• Layer 3 Convergence• Multiple gateways• OSPF ASBR• Local Repair

• Neighbor Validation• Channel Provisioning• IP Address Provisioning

PDF

AWR

RFM

PDF• Neighbor discovery• Layer 2 Convergence• Island Avoidance



To be considered a “valid” neighbor, all of the following must be true about the radio interface configurations on each neighbor:

The radio operation mode must be set to mesh (as opposed to client service) The mesh ID profile must match The radio physical mode must be compatible (e.g. data rate, channel width) The SNR must be above the minimum threshold (defaults to 15dB or greater) The neighbors must have available peer capacity It must not conflict with any existing neighbor-to-neighbor link It must be allowed by the ACLs on each neighbor

Note that the neighbors need not be on the same channel, as channel assignments are made later in the discovery process. Each radio on an AirMesh AP may have a maximum of six neighbors. So an MSR4000 could have 24 neighbors while and MSR2000 could have 12 neighbors. If more candidates are available than slots, neighbors are prioritized based on the following criteria, in order of precedence:

A “preferred neighbor” specified by the network administrator has highest priority. This is used to force specific links if desired.

The candidate neighbor with the higher RSSI/SNR will always have next highest priority. The candidate neighbor with more reachable mesh portals is chosen next. A candidate with a more optimal RF link will be preferred.

If multiple radios are able to see the same neighbor, additional decision making is performed in the AirMesh AP to make the selection.

Channel assignment is one of the most important RFM functions and well worth understanding. It is significantly more complex than the non-mesh indoor AP systems such as Aruba's Adaptive Radio Management (ARM) in ArubaOS. In a mesh network where multiple radios are used on every AP, often with directional antennas, the mesh must intelligently choose the ideal channel for each hop between neighbors. The ideal channel can change over time, forcing periodic ripples in the mesh as nodes reconverge.

For every link between valid neighbors in the mesh, a “reverse tournament” algorithm is used to select the most optimal channel. The algorithm uses the Path Distance Factor subsystem described later in this chapter to conduct the tournament. Here are some of the factors that go into the patented process:

The tournament always begins with the mesh portal choosing its channel(s) first Nodes with lower PDF values have higher priority and complete first (the process proceeds from

the portal outwards) Non-DFS channels are preferred over DFS channels Self-interference within the AP is considered

The result is a channel reuse plan that maximizes end-to-end performance across the entire mesh, not just individual links. RFM runs continuously in all AirMesh nodes. If new nodes are found, RFM adds them into the existing mesh. If a better neighbor link is found, it will replace the old link so long as the old link is not passing traffic. If more redundant links can be formed, they will automatically be incorporated into the link metrics.



The power of RFM is its ability to bring up a large, complex mesh very quickly with minimal administrator configuration. For a 50 node mesh, the entire process can typically be completed in five minutes or less. Reconvergence due to link failures can occur in 10 seconds or less. All the network administrator needs to do is set up a mesh ID profile, and optional channel list or preferred neighbor list. The RF and layer 3 IP routing topology is self-generating and optimized for end-to-end performance rather than simple individual links.

Adaptive Wireless Routing

Aruba Adaptive Wireless Routing (AWR) is a purpose-built layer 3 wireless routing protocol that is the foundation of the reliable AirMesh technology. The patented AWR routing protocol is specifically designed for scalable wireless networks and is suitable for mobile and fixed wireless applications.

To address the need to cover large areas, AWR provides standard layer 3 IP routing advantages, such as fast convergence, carrier-grade scalability, natural load sharing, and support for multiple concurrent gateways.

Unlike legacy IP routing protocols, AWR is based on a distance vector algorithm that incorporates wireless link quality metrics like RSSI and Wi-Fi retransmissions in the optimal L3 path selections. Additionally, AWR includes client loading information for very fast IP route convergence that can maintain mobile sessions even during routing updates. AWR routing includes special enhancements that dampen fluctuations during route table updates, a very important feature to stabilize the wireless network even when the RF environment may be changing rapidly.

AWR works well for both mobile and fixed wireless mesh networks with a combination of desirable features:

Fully distributed, providing resiliency against link and node outages No system-level single point of failure in Aruba AirMesh networks Dynamic, adaptive and proactive routing Self-forming, self-healing mesh topology requires no user administration Fast convergence supports high mobility and high network up-times Flexible adaptation to radio link quality changes delivers users the highest capacity Highly scalable protocol requires low computational and communication overhead Simple and easy layer 2 features ensure a connection at any instant Maintains multiple routes to each destination for fast fail-over and load-balancing Security over mesh links ensure all the routing packets are encrypted Protocol discriminates between temporary wireless fades and other outages

Having established the critical role of AWR in delivering the performance and reliability benefits in an AirMesh system, the wireless architect will wish to thoroughly understand its operation. Therefore, we will consider it in detail in the next few subsections. To set the stage, let's begin with a short primer on key concepts relating to routing protocols in general and mesh routing in particular.

Layer 2 vs. Layer 3 Forwarding in a Wireless Mesh Network

In an outdoor mesh network, all inter-node communications are wireless - which is why most wireless mesh vendors claim to use some form of “routing” among all nodes. But such “routing” is really nothing more than an extension of the spanning tree protocol designed for L2 networks.



Most wireless mesh vendors do not use L3 routing because of its high cost and complexity. Regularly exchanging route information can consume precious wireless bandwidth. In large networks routing tables can grow correspondingly large, requiring a considerable amount of memory in the routers. Constantly updating routes and making real-time packet forwarding decisions demands substantial computational resources. Because network layer routing can undermine the price/performance of a vendor's product, most choose instead to “route” at the link level.

Figure 61 Layer 2 vs. Layer 3 forwarding paths in a wireless mesh network

As you can see in the figure, all L2 mesh architectures funnel all traffic to a single gateway. This poses numerous performance and reliability challenges. While the gateway may in fact be the destination for important traffic such as video archiving servers, what about secondary video feeds that are needed by mobile law enforcement or emergency response personnel? Also, this type of unintelligent forwarding does not allow for multiple gateways for load balancing, much less redundancy. There is a single point of failure in the network. This approach truly ties the hands of the wireless architect for today's increasing performance demands.

Technical Requirements of Mesh Routing

A routed wireless mesh network is highly flexible and inherently fault-tolerant. It simplifies line-of-sight problems and extends the reach and coverage of the network with a minimal amount of network infrastructure and interconnection costs. There are also hybrid wireless mesh networks where some mesh routers are mobile and the others are not.

arun

_055

3

A

B

Wired BackhaulOptimal paths to the gatewayAlternate back-up routes

STP path between node A and B

Gateway

AWR routing path between node A and B



In whatever case is considered (whether mobile or fixed or hybrid), wireless mesh networks have some common essential characteristics:

Highly dynamic Autonomous Peer-to-peer Multi-hop Limited per-hop and end-to-end bandwidth Limited computing power and storage memory

Wireless mesh networks are highly dynamic for several reasons: First, the routers themselves may move (e.g. in mobile or hybrid wireless mesh networks), causing fast topological changes. Second, even if the routers themselves are fixed, radio link qualities can change very quickly because of interference, geographical and environmental factors. Link state in a radio mesh is not binary (up or down); rather a wide range of link qualities are possible depending on RF channel conditions. A link going sub-threshold can produce a topology change even though the path remains valid for other points in the mesh. Third, nodes may enter or leave at any time due to local RF or power conditions. As a result, the rate of topology changes in a mesh network is dramatically higher than in a wired system.

Each node must be completely autonomous in its decision making, with no dependencies either on other mesh nodes or a centralized topology store that could become unreachable. Communication with other mesh nodes will therefore be peer-to-peer by definition and will require multiple hops most of the time. Resource constraints both in terms of channel capacity and the mesh radios themselves must also be considered.

From these characteristics of wireless mesh networks, it is evident that the desired technical requirements for any wireless mesh routing protocol are as follows:

Distributed operation Very fast convergence (for mobility as well as topology changes) Extremely scalable (to hundreds or thousands of nodes) Very small channel bandwidth use for routing control plane Very small CPU and memory footprint Proactive operation (to reduce initial delay) Link quality-aware and link capacity-aware path metrics Free from loops at all times

Given these baseline requirements, it becomes possible to compare the major available routing methods in use to see which is the best starting point for a mesh routing protocol.

Review of Dominant Routing Methods

The traditional routing protocols - such as Open Shortest Path First (OSPF) or Routing Information Protocol (RIP) - are designed for wired networks, and can’t deal with fast topological and link quality changes that are common for wireless mesh networks. They can be classified into two categories according to their design philosophy:

Distance vector (RIP, IGRP) Link state (OSPF)



Distance vector routing protocols (DVRPs) were used in early packet networks such as the ARPANET. The main advantages of the distance vector approach are simplicity and computation efficiency. However, this approach suffers from slow convergence and tendency of creating routing loops. While several approaches were proposed which solve the looping problem, none of them were able to overcome the problem of slow convergence.

This led to the creation of a completely new routing technology called link state. In link state, global network topology information is maintained in all routers by periodic flooding of link state updates by each node. Any link change triggers an immediate update. As a result, the time required for a router to converge to the new topology is much less than in the distance vector approach. Due to global topology knowledge, it is also much simpler to prevent routing loops.

Unfortunately, as the link state method relies on flooding to disseminate the update information, excessive control overhead may be generated, especially when mobility is high (or severe radio interference is present that causes link status flapping) and frequent updates are triggered. As mobility increases, the traditional link state protocols like OSPF become infeasible as they will consume a large portion of network capacity and node processing power just to keep up with the fast topological changes. The overhead of the routing control traffic becomes so large that it simply overwhelms the data traffic.

In addition to the traditional routing protocols designed for wired networks, a number of routing protocols designed for mobile ad hoc networks have been proposed. They are commonly broken down into two broad classes:

1. Reactive routing protocols (i.e. AODV, DSR, TORA), which only discover and maintain routes on demand. By adapting to the traffic pattern on a demand or need basis, they can utilize CPU and bandwidth resources more efficiently, at the cost of increased route discovery delay.

2. Proactive routing protocols (i.e. DSDV, OSLR), which always maintain routes to every possible destination on the assumption that they may be needed. In certain contexts, the additional latency incurred by the reactive routing protocols may be unacceptable. Proactive routing protocols are desirable in these contexts if bandwidth and CPU resources permit.

The ad hoc routing protocols mentioned above have made significant improvements with regard to dealing with fast topological changes in mobile environments. For example, AODV is an on-demand (reactive) distance vector protocol. The basic idea behind these reactive protocols is that a node discovers a route in an on-demand fashion. It computes a route only when needed. In on-demand schemes, query/response packets are used to discover route to given destination. However, since a route has to be entirely discovered prior to the actual data packet transmission, the initial latency may degrade the performance.

Like the link-state method, on-demand routing schemes also suffer tremendous protocol overhead when traffic load and mobility are high. Simulation results have shown unacceptable level of packet loss and long delays during high mobility. Moreover, most of the existing protocols, traditional and ad hoc alike, have severe scalability and stability issues in terms of adapting to fast radio link quality changes that are common for both mobile and fixed wireless mesh networks.

Distance Vector Routing Properties

We have seen that none of the dominant wired routing protocols may be used directly for a wireless mesh. It is also clear from this analysis that the link state method is inherently incompatible with the technical requirements set forth above due to its high resource requirements for bandwidth, CPU power and memory footprint. Conversely, the DVRP method is a very good fit for these requirements



due to its simplicity and small resource requirements, providing that the convergence and looping challenges can be overcome. Let's consider these problems in more detail.

There are two types of loops that can occur in a DVRP-based system: short term and infinite. Both can be visualized in the diagram below. The number listed next to each link is the path cost.

Figure 62 Possible loops with distance-vector routing

A short-term loop can occur if link B->C fails. In this case, a routing loop between A and B is created if A tries to send a packet to C. The loop will clear itself once the cost of path A->B is equal to 10.

An infinite loop will occur if link A C also fails before a short-term loop between A and B is cleared. In this case, traffic will loop forever between A and B.

In addition to loops, convergence time has always been a shortcoming of DVRP-based systems. From a routing perspective, convergence is defined as the process of the entire network reaching a new stable loop-free state following any event that changes the topology. The convergence time is simply the elapsed amount of time required to complete this process. In a large network with many nodes and exponentially more possible paths, it is not uncommon too see convergence times on the order of minutes or even hours. This occurs because no individual node has any idea of the entire network topology. As a result, convergence requires many individual changes to ripple through the network until a new optimal solution is reached.

The AirMesh AWR Implementation

Aruba developed the Adaptive Wireless Routing protocol to solve the problems mentioned above. The AWR protocol is an adaptive, distributed, proactive routing protocol designed specifically for wireless mesh networking. It is based on a distance-vector foundation, with enhancements that eliminate both types of looping while radically reducing convergence time (typically five seconds or less for a large mesh of several hundred nodes). AWR also includes a unique mechanism to dampen fluctuations during routing table updates, as well as the patented Local Repair mechanism for fast route convergence and network overhead reduction.

In AWR, each router maintains an enhanced DVRP routing table that contains all the information necessary to forward a data packet toward its destination. Each routing table entry is tagged with a destination sequence number originated by the destination node. This helps on identifying the stale routes from the new ones, thus avoiding loops.

In AWR, each node keeps track of its continued connectivity to its neighbors. A broken link may be detected by the layer 2 PDF protocol discussed later in this chapter, or it may be detected by using

arun

_054

9

BA

C

Cost = 1

Cost = 1

Cost = 10



AWR's layer 3 enhanced wireless hello protocol. To maintain the consistency of routing tables in a dynamically varying network, each node periodically transmits scheduled updates, and also transmits triggered updates when significant new information is available.

To address the convergence time problem, AWR introduced two patented innovations to the classic DVRP method: Route Diversity and Local Repair. With route diversity, AWR maintains multiple routes to each destination that could be used for fast fail-over and load balancing. This is similar to the feasible successor concept from link-state protocols. In other words, every AWR node effectively has a priori local link state knowledge for all valid paths to its immediate neighbors. It is considerably different from DVRP in which only one valid path exists to each destination. While simple in concept, Aruba's implementation required significant engineering to perfect.

Local Repair is a process to discover a new route locally without resorting to an end-to-end route discovery when an intermediate link breaks. AWR includes a proprietary messaging mechanism to further speed up the convergence and to further reduce the routing overhead by making the re-convergence as local as possible. The benefits of Local Repair are visualized in Figure 63 and Figure 64 below.

Figure 63 Topology events propagate network wide without local repair

Consider a source node S attempting to send traffic to destination node D across a mesh network, as show in Figure 63 above. Consider further that the link from A to B fails. Without local repair, A will broadcast to the entire network that it has lost the connection to B. This routing update will flood to all of A’s upstream and downstream nodes, including S, and cause them to update their routing tables.

Figure 64 Local repair contains topology events and accelerates convergence

arun

_055

0

A

C

HD

B

S

arun

_056

0

C

HD

B

A S



With local repair, a special message is sent by A only to its immediate neighbors. Neighbor C will respond to that message that it has a valid route to D. Node A can then repair its routing table without affecting any other nodes. In this way, local repair can greatly reduce the routing overhead in unreliable network.

Figure 65 AWR allows the mesh to route around failed or degraded links

This innovation delivers extremely fast convergence - even faster than wired link state protocols. This is because AWR effectively localizes topology changes to the nodes in the immediate vicinity of the change (from an L3 perspective). This reduces both the geographic scope of topology change control traffic, as well as the convergence period. This approach is best fit for the typical topology event triggers in a wireless mesh.

AWR works well for both mobile and fixed wireless mesh networks. It offers a very attractive combination of desirable features for wireless mesh routing protocols:

Fully distributed, providing resiliency against link and node outages, ensuring there is no system-level single point of failure

Dynamic, adaptive, proactive routing: self-forming, self-healing, reducing initial delays Fast convergence: enabling high mobility and greatly improving serviceability Flexible adaptation to not only topological but also radio link quality changes Maximize user throughput by taking radio link quality into consideration (extremely important for

wireless mesh networks) Highly scalable (low computational & communicational overhead). Especially important for large

wireless mesh network deployment Simple and easy to implement Multiple loop-free routes to each destination for fast fail-over and load-balancing Security (all the routing packets are encrypted and authenticated) Support multi-radio, multi-hop wireless mesh networks



Unique capability of discriminate between temporary wireless fades and an actual loss of wireless links due to mobility, router failure etc.

AWR is running in tens of thousands of wireless mesh nodes, including some of the largest outdoor mesh networks on the planet. Many simulation and experimental results have shown that AWR works well for both mobile and fixed wireless mesh networks, and AWR showed superior performance with respect to its peers for a wide range of user applications.

Inter-band and Intra-band Routing

A full layer 3 mesh routing protocol provides the wireless architect with unprecedented degrees of freedom. Especially when combined with multiple backhaul radios in a single AP chassis. Therefore, it is worth calling out some of the specific new topologies that are possible with AWR.

Each radio in an AirMesh access point that is configured for backhaul mode is treated by AWR as a discrete IP interface. It has most of the important interface configuration options you would expect on a wired router. This makes it possible to forward traffic directly between any two backhaul radios in the same chassis. Furthermore, because AirMesh allows any radio to be configured for any band, AWR does not require or care that traffic flow on a specific band. The diagram below illustrates this flexibility:

Figure 66 Inter-band and intra-band routing in an AirMesh network

You can see both intra-band and inter-band forwarding routes in the diagram. Intra-band forwarding means sending traffic between two radios on the same frequency band. For example, ingress on channel 149 and egress on channel 153. Unlike some competing layer 2 mesh implementations, AirMesh does not require you to designate an “upstream” and “downstream” direction. Frames are frames, to be forwarded according to standard routing algorithms.

Even more interesting is inter-band forwarding. In this case, traffic may travel on any combination of bands - 5GHz, 4.9GHz and 2.4GHz - to reach its destination. These are all simply hops with a given cost as far as AWR is concerned.

This provides the wireless architect with the ability to use an alternate band in a particular geographic area that may be challenged in the primary mesh band due to local interference or other conditions. Another application is in very large meshes with two-tier RF backbones using different frequency bands for each tier. Let's say you are using the UNII-3 band for the first mesh tier, and the UNII-2 band for the second tier to inject capacity into the first tier. In the injection nodes, one radio would be designated as an “uplink” on UNII-2, while the other backhaul radio(s) would be on UNII-3.

arun

_055

2

5 GHz5 GHz

2.4 GHz2.4 GHz

4.9 GHz5 GHz

5 GHz4.9 GHz5 GHzInternet



OSPF Autonomous System Integration

Wireless mesh networks from Aruba using AWR also support OSPF routing. OSPF is a hierarchical interior gateway protocol that employs link state to create an optimal hierarchy among routers in a network.

The OSPF autonomous system boundary router (ASBR) function is utilized so that external routers include the wireless mesh in their route tables. The AWR implementation in AirMesh supports both backbone and non-backbone areas as defined in the OSPF standards.

OSPF periodically publishes link state advertisements (LSAs), which are required by the external gateway routers. Support for these OSPF features enables network operators to optimize routing across multiple autonomous systems — wired and wireless — in any single network domain.

Multiple Gateways

Finally, AWR makes it possible to automatically balance traffic loads across all available gateways to the Internet or other external networks. In fact, the entire AirMesh network constantly balances the total load to optimize traffic flow, even under adverse conditions with high levels of RF interference.

This is particularly useful for wireless architects who may be accustomed to “hardwiring” specific traffic flows to get around L2 mesh traffic flow limitations. For example, consider a video surveillance deployment with 10 cameras with a bitrate of 8Mbps each. If only 5 cameras can be accommodated on a single gateway, the wireless designer may choose to statically assign 5 cameras to each of two different gateways on two different channels. However, with AWR this is completely unnecessary. Simply add the necessary number of uplinks and AWR will manage the rest.

AWR Scaling & Configuration Best Practices

Follow these general recommendations when AWR is used: The number of nodes in a single mesh area or cluster must be less than 50. This is more than

sufficient for many networks, as this works out to approximately 5 km2 (2 mi2). Larger meshes can be easily constructed out of multiple mesh areas.

The network should be planned to limit the number of hops to six or less. One mesh portal should aggregate no more than 16 downstream nodes at typical ingress loads.

The number of nodes in a single mesh network depends on the capacity and the traffic of each node. Assume all four radios of a MSR4000 work in backhaul mode, then the total sustainable UDP throughput of the MSR4000 is about 200 Mb/s X 4 = 800 Mb/s. If each child node needs an average 50 Mb/s of bandwidth, the number of nodes that can be served by one MSR4000 gateway is 16. This value may increase or decrease based on your specific offered loads, as well as link-specific throughput reductions due to RF impairments such as free space path loss, attenuation due to ground clutter, and the like.

Path Distance Factor

Both AWR and RFM depend on a patented Aruba technology called the Path Distance Factor (PDF). PDF is both the algorithm and method by which link metrics are computed and shared between peers in an AirMesh network. For example, the AWR protocol utilizes PDF to compute RF-aware routing tables. In an AirMesh network, each node announces its own PDF information in a beacon frame to reduce the need for dedicated transmissions for propagating RF state. Each receiving node calculates and updates its PDF value and propagates it to its neighbor in the next beacon. In the most simple



terms, the PDF value is the router’s distance to the nearest portal. A portal’s PDF value can be 0 or any value larger than zero if we want to give different weight to different portals.

A PDF is a 32-bit integer. A vendor specific IE extension is added into the 802.11 beacon and probe response management frame. The PDF value is then announced periodically inside the beacon frame. Other management frames such as probe response frame may also contain PDF values.

By taking advantage of existing 802.11 beacon and probe response management frames, PDF adds no extra burden to the existing 802.11 mesh network. Since beacons and probe response frames can be received by all neighbors in the coverage area, each one is effectively a local broadcast. Every AirMesh node’s latest PDF value is announced to all neighbors nearby. Contrast this approach with spanning-tree PDUs or routing protocol announcements that add much more burden to the limited wireless bandwidth.

PDF Initialization and Update Logic

Each router initializes its own PDF value to infinite, which implies that each mesh node assumes it does not have a valid link to any portal when powered on.

If the router is a portal, which is known either from the configuration or from user CLI, the router then changes its own PDF value to the user specified value or a default pre-configured portal PDF values (zero in the basic case).

Since each router periodically broadcasts its PDF via beacons, all neighbors will receive the updated PDF value sooner or later. Upon receiving a beacon, the router performs following steps to process a PDF update.

Figure 67 Logic flow for PDF update processing

arun

_044

2

Receiving beacon/probe responseon one interface

Newneighbor?

Neighbor hasfinite PDF & my own

PDF is infinite?

Add the new neighbor intothe topology database,

PDF[if] = neighbor_PDF +cost_to_neighbor,

Start new neighbor creation process

Update the neighbor inthe topology database,

PDF[if] = min(existing PDF,neighbor_PDF+cost_to_neighbor)

Set the new PDFvalue into thebeacon/probe

response

Add the neighborinto the topology

database and return

Calculate own new PDFRouter_pdf = min{PDF[if] for

all ifs on the router}

Update the neighborin the topology

database and return

Do I have avalid link with this

neighbor?

No

No No

Yes

Yes Yes



On a given mesh router, each radio maintains a topology database that contains all scanned neighbors. This is called the candidate list. The radio selects valid neighbors from the candidate list to form links. For those neighbors which have valid links with the router, they form the neighbor list.

As shown in Figure 67, upon receiving a PDF update from a neighbor, the router first checks if it is a new neighbor or not. If the neighbor is new, it is added into the topology database. If the neighbor is not new, the router checks to see if it has an existing link to the neighbor. If it does, the router will see whether its own PDF needs to be updated. If it does not, it means the router does not have valid link with the neighbor, then only the topology database is updated.

One special case is that the router ‘s own PDF value is still infinite while the neighbor’s PDF value is a finite number. This means that the router itself is still in the island state, meaning that it does not have any path to reach a mesh portal but the neighbor does. In this special circumstance, in order to get out of the island state quickly, the router immediately sends out link establishment request to the neighbor.

As Figure 67 shows, the whole PDF is similar to a routing update, but with much simpler logic. Each interface on the router has a PDF value, and the router’s unique PDF value is the minimum PDF value among all of its interfaces.

Multiple portal support is easily achieved utilizing PDF. If the wireless engineer assigns a different PDF range to different portals, one can clearly distinguish different portal paths for each node. By giving different initial PDF values to different portals, portal load balancing is also supported.

PDF Timeouts

Each neighbor’s PDF value has a timeout timer associated with it. If the router has not received any beacon/probe response message from the neighbor within the timeout period, the neighbor’s PDF value will be set to infinite again.

Figure 68 Logic flow for the timeout of a PDF value

Figure 68 shows the logic flow associated with a timeout event. As with the PDF update, the router first has to check if it has a valid link to the neighbor. If it does not, then it just needs to update the topology database. If it does have a valid neighbor, that node has to be moved from the neighbor list back to the topology database (candidate list). The link has also to be either removed or set to down state.

arun

_044

3

Calculate own new PDFRouter_pdf = min{PDF[if]for all ifs on the router}

A neighbor’sPDF update

times out

Do I havea link with this

neighbor?

Update the neighborin the topology

database and return

Change the neighbor’sPDF to infinite,

remove it from neighbor list

Set the new PDFvalue into the

beacon/probe response

No

Yes



Loop Avoidance

The PDF update, together with the topology database builds a loop-free tree topology in the mesh network. This loop-free tree is similar to a spanning tree, but with following differences:

1. Link redundancy is supported2. Each link's weight is taken into the consideration3. Multiple portals are supported, and each one can have different weights

The mechanism the PDF algorithm uses to ensure a loop-free tree is as follows. Each node builds its topology with PDF information received from all of its neighbors. The topology database contains each neighbor’s PDF value, plus its own PDF value to the neighbor. Note that the neighbor’s PDF value is the neighbor’s cost to the nearest portal, similar to the Reported Distance (RD) in the diffusing update algorithm (DUAL). The router’s PDF value, which is similar to the Feasible Distance (FD) in the DUAL, will be one of the neighbor’s PDF values (RD) plus the router’s cost to that neighbor. Such neighbors are referred as Successors in DUAL.

Figure 69 Loop-free tree using PDF

Figure 69 is an example of how PDF ensures a loop-free tree. Assume the PDF value of router A to destination X is 100 via router B. In another words, A’s feasible distance to X is 100, and B is A’s successor. Router C also has a route to the destination X, which has a PDF value of 90. Since the cost from router A to router C is higher than the cost from router A to router B, router A chooses router B as the successor, and sets its own PDF value to X as 100.

Although router A does not use router C as the successor, router A still saves router C’s information, together with its reported distance to X in its topology database.

If the link from router A to router B fails, router A can then choose router C as the new successor. In this case, it will update its PDF to destination X to 110. However, how does the router know whether the path from router C to destination X is loop-free?

Router A can safely decide the path is loop-free by comparing router C’s reported distance (90) with router A’s original feasible distance (100). If the RD is less than the original FD, it means that router C satisfies the feasibility condition (FC), and router C becomes the Feasible Successor (FS). In our case, router C’s reported distance (90) is indeed less than the original FD (100). This means that router C’s path to destination X does not involve router A, therefore router A can safely choose router C as the new Successor.

arun

_044

4

PDF to X = 90

PDF to X = 90

A’s successor is B

PDF to X = 100

Cost = 20

Cost = 10

DestinationXA

C

B



If router C’s RD value is higher than router A’s original FD, it implies that router C’s path to destination X might be via router A, which might lead to a loop.

It can be seen from the example that the PDF algorithm completely guarantees that the constructed tree is loop-free. On the other hand, not all loop-free paths are included in the tree.

Each router calculates its PDF to the destination via all feasible successors, and picks the lowest FS as the PDF value. If multiple FSs yield the same minimum PDF value, they all become the router’s successors. In the case of topology changes, if any given FS is locally available in the topology database, the convergence will be very fast. If no FS is available, the router will have to send out queries to its neighbors. The query propagates (“diffuses”) until a reply is received. Routers that do not find a FS will return an unreachable message.

Directional Link Buildup

After a neighbor is selected, the one with larger PDF value needs to explicitly send out a link establishment request to the other end. If a router receives a link establishment request from a neighbor which has even smaller PDF, the request will be denied.

The principle is that the link buildup process is directional. The one with larger PDF value has to be the link establishment initiator. This requirement brings more benefits than ensuring a loop free tree. When the receiver gets the request, it can evaluate the overall channel utilization and RF conditions before specifying which channel should be used for the connection establishment.

Radio State Machine

Each radio interface runs an independent state machine. The state machine consists of three different states: Discovering, Connecting and Connected.

Discovering state: The radio performs either passive or active scanning. This is the initial state when radio is enabled to backhaul mode and its mesh configuration mode is set to auto.

Connecting state: if any neighbor is found in the Discovering state, the radio goes into the Connecting state to establish a valid link with that node.

Connected state: If a valid link is established, and the router’s own PDF becomes finite, the radio stays in the Connected state.

In any of these states, if the radio interface receives a connecting request from a neighbor with larger PDF value, it will respond to it.

Since each radio will periodically get into passive scanning mode even after valid links are established, the radio will eventually find all possible neighbors on all channels.

Mesh Network Convergence

PDF propagation starts from the mesh portals and reaches every node in the mesh network. This is called convergence. After the convergence is complete, a loop-free tree exists and every node has achieved a stable state.

Since only mesh portals will start with finite PDF values, other nodes can only change their default infinite PDF values to finite values after finding a valid path to one of the portals.

This implies that if a router still has the default infinite PDF value, it is in the island state. In this case, it will keep doing neighbor discovery until its PDF becomes finite.



Multichannel Mesh Channel Assignment

PDF not only ensures a converged mesh network, but also provides a mechanism for automatic best channel selection during the mesh forming process.

Mesh portals always pick the best channel first. They then propagate their choice to child routers within one hop of the portal. When multichannel backhaul is being used, these child routers will then pick the best channel from remaining available channel pools, and so on.

Basically, the portal has higher priority over mesh points to choose the best channel available at the time of neighbor link formation. And each router makes its own local decision on which subsequent channel to pick, based on channel conditions such as interference, signal strength and noise. This procedure ensures an optimal channel selection scheme for the entire mesh network. With Aruba AirMesh, no work needs to be done by the wireless engineer to provision the backhaul network, nor is there any requirement for inflexible static channel assignments.

Active Video Transport

Aruba’s Active Video Transport™ (AVT™) traffic-shaping system delivers progressive, non-interlaced HD-quality video at up to 30 frames per second. With AVT, users perceive a significant improvement in quality, while behind the scenes AVT makes intelligent tradeoffs between latency and impairments to video quality.

Challenges In Carrying Video Traffic Over IP Networks

Fully appreciating AVT’s ability to improve video performance requires understanding the causes and effects of the most common impairments to video quality – packet loss, packet reordering and packet jitter.

Voice and video traffic is normally transmitted in wired and wireless networks using the connectionless UDP protocol. Real-time applications like voice normally cannot benefit from the retransmission feature of the connection-oriented TCP protocol. When packets are lost or corrupted in a UDP data stream, they are simply lost and are never recovered.

Figure 70 Challenges in carrying isochronous video traffic across wireless mesh networks

arun

_055

4

Constant rate video packets

Out of order packets

Duplicated packets

Lost packets

Various delayed packets

A B C D E

B A E C D

B B C C D

B D E

A B ....... C D E

IPnetwork

Source

Destination



Figure 70 above illustrates some of the most common issues encountered with video traffic traversing IP networks:

Out-of-Order Packets: Traffic loss in wireless mesh networks can be significant due to periodic congestion. Dropped packets or transmission errors that corrupt packets can occur due to a link data rate being too high, external RF noise or interference, antenna misalignment, moving obstacles, multi-path fading, user mobility, or a low or variable RSSI. Consequently, packets often arrive at their destination in a different order than they are sent from the source. TCP reorders these packets into their original sequence, but UDP does not. With UDP applications, packets are consumed in the order they are received. With compressed digital video, the effect of packet reordering can be worse than the equivalent amount of packet loss because an out-of-sequence packet disrupts the decoding process. For this reason, video equipment is often designed to simply drop the packet, or with high-end systems, build in some delay in the decoder to create a brief window of opportunity for reordering out-of-sequence packets.

Lost Packets: Uncompressed video signals are tolerant of moderate packet loss, but any amount of packet loss for compressed video signals becomes noticeable – often in annoying ways. And because wireless mesh networks have limited bandwidth, some form of compression is typically used.

Delayed Packets / Jitter: Without some provision in the video decoder, jitter causes quality to degrade with noticeable pixilation or blurred images. The same delay built into sophisticated video equipment to create a window of opportunity for packet reordering also facilitates the removal of jitter from the incoming packet stream. The sources of packet jitter in a wireless mesh network include variations in delay at the source, variable link data rates along the path, changing traffic conditions in QoS queues, changes in end-to-end routes, the non-deterministic effects of the CSMA/CA protocol, and roaming.

Compression algorithms commonly used in digital video applications are stateful. In stateful communications, the arriving bit stream is used to make changes in the existing image rather than construct a new image during each frame interval. Stateful compression algorithms have the advantage of being highly efficient, which is desirable in a wireless mesh network. However, they have the disadvantage of not being tolerant of packet loss, and often require a disruptive resynchronization between the encoder and decoder when packet loss, reordering and jitter become severe enough.

How AVT Works

As stated earlier, Aruba’s AVT traffic-shaping system delivers high-definition video by making an intelligent tradeoff between latency and the impairments to video quality. The increased latency required to compensate for packet loss, reordering and jitter is imperceptible to users. What is perceptible is the significant improvement in video quality.

AVT uses four technologies to deliver cinema-quality 30-frame progressive video across the wireless mesh network – deep packet inspection, MAC protocol optimization, an in-network retransmission protocol, and adaptive video jitter removal.

An AVT link consists of a tunnel with two components: ingress and egress. AVT improves the video traffic by retransmitting lost packets, removing in-packet jitter, and by reordering packets automatically inside the mesh without any action by the upper layer application.



At the ingress point, AVT uses deep packet inspection to find the video stream in the data flowing across the network. Then AVT differentiates the video frames within in the video stream and marks key video attributes like the compression ratio, frame type, alignment, and AV synchronization of each frame.

Deep packet inspection also identifies and extracts the compression algorithm, video decoding buffer model, video frame-type boundary and video timing used by the packet stream. This information is needed to properly share the traffic through the Aruba wireless mesh network.

At the egress point, AVT restores the video frame into the original format based on the information marked by the AVT ingress.

The MAC protocol optimization and in-network retransmission protocol work together to minimize and, when necessary, recover from packet loss. This combined prevent-and-recover approach is especially effective in noisy RF environments where packet loss can occur most often.

To further improve the video quality, AVT introduces a jitter buffer to overcome any unexpected jitter and delay caused by the wireless network. At the egress, AVT uses the jitter buffer to collect, store, and reorder the video frames and then sends them onto the wired network in their original sequence, synchronized with the video decoder.

Figure 71 shows a typical mesh topology with video surveillance. MSR #1, MSR #2, and MSR #3 create a wireless mesh network:

MSR #1 is a leaf router connected with a video camera. MSR #2 is a middle router that connects MSR1 and MSR2 with two mesh links. MSR #3 is the mesh gateway router that connects to the video surveillance center (PC).

The MSR closest to the camera or video source is considered the AVT ingress, and the gateway MSR generally is the AVT egress point.

Figure 71 Example of AVT ingress and egress

AVT leverages the multicasting capability in AWR to provide concurrent and efficient multi-path transmission of HD-quality video to multiple destinations. Multicasting is especially useful for video surveillance applications that require monitoring and recording at multiple locations or for IPTV applications that broadcast video to multiple viewers.

Virtual Private LAN over Mesh

Virtual Private LAN over Mesh (VPLM) is proprietary Aruba AirMesh technology that is used to provide the simplicity of native layer 2 services to customers while retaining all of the routed network advantages delivered by AWR, operating at layer 3 in the mesh network.

arun

_039

0MSR1

AVTingress

AVTegress

MSR2 MSR3 Videostorageserver

Camera Mesh link Mesh link

Video traffic



VPLM is essentially an overlay technology that enables native plug-and-play L2 VLAN connections that are served by the underlying AWR routing protocol. The concept and architecture of VPLM is very similar to Virtual Private LAN Services (VPLS) technology, which is the solution that is widely used by worldwide ISPs to provide L2 services to enterprise customers. But VPLM is adapted to ensure loop-free topologies over wide area mesh networks.

VPLM Overview

Figure 72 demonstrates the typical VPLM architecture, where VLAN 100 is used by a customer to provide video surveillance services over the mesh network and VLAN 102 is used to provide standard 2.4 Ghz client access. MSR1 on the left runs VPLM on VLAN 100 and Camera1 on the left side can be plugged into the MSR without any special configuration. Data from that port is trunked over the mesh network transparently. All video frames captured by Camera1 are tagged by MSR1 with VLAN 100, and then sent to the wired network over the mesh network where they then are forwarded to the wired network also using VLAN 100. Video frames are routed by the AWR routing protocol so they always use the most optimal path.

Figure 72 Virtual private LAN over Mesh

The same VLAN tagging and AWR routing mechanism applies to client access traffic in VLAN 102, but in this case all devices associated with a specific ESSID and connected with MSR1 are tagged and then sent over the mesh network.

VPLM Implementation

The VPLM feature is similar in concept to how VLANs are implemented on just about any ethernet switch, with the exception that each VLAN must be individually defined on each AirMesh node. Each node maintains a VPLM membership database (MDB). The MDB is a table that describes the

arun

_042

9

VPLM instance

Mesh cloud

VLAN Router ID100 MSR3

VLAN Router ID100

MACCamera2 MSR3

VPLM instanceVLAN Router ID102 MSR3

VLAN Router ID102

MACSTA2 MSR3

VPLM instanceVLAN Router ID

102 MSR2100 MSR1

VLAN Router ID

102

MAC

STA1 MSR2100Camera1 MSR1

STA2

STA1

MSR2

BSS onVLAN 102

BSS onVLAN 102

MSR1

Camera1on

VLAN 100Camera2

onVLAN 100MSR3

CE

CE

PE PE

VLAN 100

VLAN 102



relationship between VLAN numbers and router IDs. Individual VLANs are completely isolated from one another as expected. And VPLM supports 802.11e to 802.1Q mapping.

VPLM utilizes a control plane service to coordinate between AirMesh nodes. This service is responsible for collecting VLAN and site information from all interfaces on a given node. It manages the MDB table, and periodically exchanges MDB data with other nodes participating the VPLM instance.

For each VLAN configured for VPLM on a given node, the classic forwarding and learning processes work exactly as expected. As new source MAC addresses are learned, the VPLM forward information database (VFIB) will be updated. The VFIB is a table that describes the relationship between client MAC address and router IDs. For unknown destination MAC addresses, we will flood the frame(s) to all routers in the same VLAN.

All frames to be bridged on the VLAN will be forwarded to the proper VPLM tunnel. Frames are encapsulated inside tunnel protocol headers and forwarded across the mesh by AWR like any other traffic.

VPLM is meant to be transparent so it does not run STP. Because VPLM fully supports the AWR multiple gateways feature for load balancing, it is necessary to implement an internal loop avoidance mechanism within VPLM. This is accomplished with the Site ID. This is a self-defined number for each L2 wired network that can forward and receive traffic with a VPLM tunnel. All of the AirMesh nodes that connect to the same Layer 2 network must be configured with the same Site ID. Different L2 networks must have different Site IDs, as shown in Figure 73.

Figure 73 VPLM uses site IDs to avoid loops

In summary, VPLM provides native plug-and-play and easy-to-use L2 services to customers, retains all L3 core benefits in the AirMesh network, and is the default mode of operation of Aruba AirMesh products.

arun

_055

5

Site ID 1

Site ID 2

BA

D C



MobileMatrix and Seamless Session Persistent Roaming

Aruba's MobileMatrix™ provides the ability to roam seamlessly, potentially at very high speeds, throughout the Aruba wireless mesh infrastructure. The integration of the network and link layers in AWR provides the cross-IP subnet roaming capability needed to allow clients to move from wireless mesh AP to mesh AP in less than 50 milliseconds while maintaining session persistence and keeping the same IP address. Fast roaming maintains a continuous application connection, which is critical for latency-sensitive applications like voice and video.

The IEEE 802.11 standard does not specify a robust, interoperable mechanism for roaming. In fact, it requires each client have only a single connection. There is also no provision for an AP to identify clients within its range, which places the burden on the clients to detect available connections and initiate a roaming request.

The inter-AP protocol (IAPP) specified in IEEE 802.11f provides a means for nomadic movement between APs. But the delay involved is generally longer than required to support real-time voice and video applications. Plus, IAPP is problematic for connections that use WEP, WPA or WPA2 security. Without any enhancements, IAPP is suitable only for data applications that are insensitive to latency and require no security.

Roaming in Wi-Fi networks is possible at layer 3 with the Mobile IP standard described in IETF RFC3344. Optional for IPv4 and required for IPv6, Mobile IP enables packets to be forwarded in tunnels from a system with a fixed IP address to mobile devices that roam among multiple networks. These mobile devices can roam across multiple subnets, where it becomes necessary to assign a new and different IP address.



Significantly, Mobile IP was designed to be transparent to both the mobile node itself and the correspondent node at the remote end, which may be either mobile or stationary. But Mobile IP is very complicated and is not widely adopted. Aruba’s MobileMatrix leverages the capabilities of IAPP and adopts a simplified version of Mobile IP to make roaming fast and seamless without high overhead. MobileMatrix maintains full interoperability with ordinary Wi-Fi clients. It does not require any special software on servers, clients or internetworking systems external to the wireless mesh. The contrast between Mobile IP and MobileMatrix may be visualized in Figure 74.

Figure 74 MobileMatrix overcomes challenges using MobileIP on wireless mesh networks

MobileMatrix uses four methods to achieve fast, cross-IP subnet roaming: The Global MobileMatrix Service process maintains all mobile user information required by the

other MobileMatrix processes. The Access Point MobileMatrix is responsible for initiating and completing the roaming requests

on behalf of clients and within the wireless mesh infrastructure. It is equivalent to the Mobile IP foreign agent, and runs on all routers.

Any router that serves as a gateway to an external network uses two additional processes that are equivalent to the Mobile IP home agent. The Local MobileMatrix Service maintains the mobile user information required by the local gateway. It is an extension of the Global MobileMatrix Service.

The MobileMatrix Traffic Gateway is the companion process in gateway routers. It is responsible for establishing the route to the mobile client’s current AP, which includes advertising any new route to the AWR protocol.

arun

_055

6

Centralized• Long handoff control path over multi-hops to AC• Slow switch to new AP

Distributed• Short handoff control path between local mobility services• Fast switch to new AP

RFC3344 Mobile IP

Client10.1.2.100

Client10.1.3.100

Client moves

Handoff setupcontrol path

PermanentHome Agent

ForeignAgent

Old AP10.1.2.1

ProxyHomeAgent

MobileMatrix

Client10.1.2.100

Client10.1.3.100

Client moves

PermanentHome Agent

ForeignAgent



MobileMatrix roaming begins by inheriting the standard MAC layer trigger mechanism initiated by the Wi-Fi client. But IAPP only supports roaming within a single IP subnet, which is problematic for IP applications. In fact, if a client roams with IAPP to an AP on a different IP subnet, its IP address can no longer be used in a current session. Consequently, it will require a new IP address and re-initiate a new session.

To support cross-IP subnet roaming, MobileMatrix uses a special gateway function, which is analogous to Mobile IP’s home agent, to recognize that a client IP address is now using a different AP. MobileMatrix immediately updates the routing tables in AWR to route packets via the new AP.

Figure 75 Roaming within a mesh - control plane

arun

_055

7

Gateway 2

Foreignagent

Homeagent

Client info

Client info

Client

Gateway 1

LMS: Local MobileMatrix ServiceMTGW: MobileMatrix Traffic GatewayGMS: Global MobileMatrix ServiceAPM: Access Point MobileMatrix

GMSDHCPDNSAAA

LMSMTGW

LMSMTGW



While MobileMatrix and Mobile IP take a similar approach to roaming, the important difference is speed. The ability to complete the transition in less than 50 milliseconds gives MobileMatrix seamless session persistence for virtually any IP application, including voice. The 50-millisecond transition period is from beginning to end for both the client and the wireless mesh network. Wi-Fi clients constantly scan for available wireless mesh APs. When a stronger signal is detected, the client can initiate an IAPP roaming request. In fewer than 50 milliseconds, MobileMatrix recognizes the request and initiates an update to the route tables in AWR, and then propagates the changes to affected routers in the wireless mesh while the client simultaneously re-associates with the new wireless mesh AP.

Figure 76 Roaming within a mesh - data plane

arun

_055

8

Gateway 2

Homeagent

Foreignagent

Client

Gateway 1

GMSDHCPDNSAAA

LMSMTGW

APM

APM

APM

LMSMTGW

3

4 6

7 8

5

1

2

Control transfer

LMS: Local MobileMatrix ServiceMTGW: MobileMatrix Traffic GatewayGMS: Global MobileMatrix ServiceAPM: Access Point MobileMatrix



Chapter 8: Planning the Access Layer

When outdoor mesh coverage for fixed and mobile client devices is planned, many important things must be considered in a well-defined sequence, beginning with the connection between the client devices and the closest AP. Client devices come in a rich variety of types—from mobile Wi-Fi devices such as smartphones or mobile routers installed on vehicles to fixed cameras wired directly into mesh nodes or wireless client bridges. As a rule, always plan the access layer first and then plan the backhaul layer.

Contrary to the approach recommended by some vendors, mesh access layers cannot simply be planned by taking a fixed number of APs in a square kilometer or square mile and laying them out in a grid. While this type of approach may help for rough order-of-magnitude budgeting, it will not deliver a working mesh. There is no “one size fits all” mesh access layer design - every one is unique. From the available mounting assets to the ground clutter environment, and from the client device mix to the typical building construction materials, no two mesh access layers can use the same design.

The access layer design methodology recommended by Aruba can be visualized in five major steps as follows:

Figure 77 Mesh access layer planning methodology

By following this process, the wireless architect can be assured that the node densities are appropriate to that specific mesh and that it should generally operate as expected. The balance of this chapter explores the technical concepts required to move through the steps successfully.

DiscoveryA thorough discovery process is the key to getting the survey right the first time, and avoiding unexpected trips to revisit the site. Most if not all of the following areas should be considered during this process.

arun

_054

6

Define coveragefootprint

Identify sitingconstraints

Identify QoS / SLA zones

Specifykey designparameters

Identify offeredload by device

Assign minimumthroughputs

Choose oversubscription ratio

Identify usablechannel count Iterate until all requirements are met

Discovery Cell SizeCapacity Plan RF Design Model

Matchclient to AP power

Choose minimumdata rates

Estimatepath losses

Identify validmounting assets

Choose antennasand bearings

Create 3DCoverage Model

Create bill ofmaterials

Choose bestmounting sites

Aruba Networks, Inc. Planning the Access Layer | 101


Define the Coverage Footprint

All outdoor networks start with one or more coverage areas marked on a map. For outdoor hotspot networks, this may simply be an outline showing the service area. For video surveillance networks, it could be the field-of-view map from the video integrator showing all camera locations and bearings. For campus extensions, it could be the perimeter around selected buildings on the campus.

Figure 78 Municipal mesh network coverage area

Figure 79 Intermodal railyard coverage area

Whatever the footprint size or shape, this is the first major constraint on the project. Only the mounting assets in the footprint are generally of interest. The topography and ground clutter characteristics of the area will have to be managed. The overall cost of the project is directly related to the size of the



area. Later in the planning process, if the budget will not cover the entire footprint, one of the most common solutions is to reduce the coverage area permanently or to break it into phases. (Reducing AP density to meet budgetary requirements is almost always a bad idea - it is much better to maintain density and reduce the coverage area.)

Identify Siting Constraints

The design team needs to know what limits they have to choose mounting assets within the network footprint. For mesh portals, it is critical that a complete list of preferred mounting structures is provided before the survey team deploys. There can be thousands of buildings in a few square kilometers or miles. Survey teams on a tight schedule cannot afford to drive around randomly choosing structures. They need an initial set to work with.

For municipal deployments, it is common that city-owned and even county-owned buildings are preferred because the leasing arrangements are minor or nonexistent. Schools are another common mounting point for municipal networks.

If existing poles are to be used, you need to find out what entities own which poles, and how you tell the difference. Some areas are served by multiple power, telephone and even cable companies. Each may have their own poles, and on the same street even. Find out which electrical utility and telco providers are willing to contribute their poles, and how to recognize and read their pole tags. Confirm that acceptable electrical power exists.

It's also crucial to know what structures cannot be used. For campus deployments, identify which buildings have the least aesthetics or approvals requirements. For historic areas, even city-owned structures are likely off limits or have serious limits.

Identify Quality-of-Service or Special Service Level Agreement Zones

It is common that some areas within a given coverage footprint have special requirements. Examples of these include:

High bandwidth / Minimum QoS zones: These areas may have a higher minimum data rate specified than the average required for the rest of the footprint. There may be more rigorous indoor penetration requirements. Different zones could have different minimums.

Low user density zones: For areas with sparse populations or low expected offered loads, perhaps it is desirable to have reduced node density to save overall network cost.

Fee-based / High availability zones: These areas may have paying subscribers on the network. Extra care may be desired in the engineering in these areas to maximize the customer experience.

Specify Key Network Design Parameters

The customer should provide top-level network engineering criteria that will significantly impact the choices available to the design team. Examples include:

Minimum average data rate in the overall coverage area Expected concurrent users per access node Average and peak bit rate from video cameras selected for the project Minimum and maximum available uplink circuit bandwidth (physically available and affordable

within the available budget)



Total budget for capital equipment Total budget for physical installation

These are just a few of the possible examples. As you begin the discovery process, think carefully about the nature of the project to come up with a list that is as comprehensive as possible.

Capacity PlanningWith discovery completed, the next step in planning any access layer is to understand the expected offered load that the network must support. This information is required before an RF design can be completed because capacity requirements determine AP counts and densities.

Offered Loads of Typical Network Services

Different services have different bandwidth and delay requirements from the outdoor network. These services can be generalized into three typical profiles:

real-time voice, which requires very low delay and jitter real-time video, which varies by quality, frame rate, and encoding type other data services, like Internet access

For planning purposes, non-real-time video generally can be regarded as a data service and we generically use a range of 512 Kb/s to 2048 Kb/s for real-time video. While the theoretical peak 2 spatial stream data rate of an 802.11n outdoor network is 300 Mb/s, this throughput metric represents only the physical RF connection rate, not the application layer throughput. As in all data connections, application communications must pass through several other communications layers with each layer adding overhead to the original application layer data. The true application layer throughput is measurably smaller than the physical RF link data rate.

If an AP uses encryption, this added fixed-length header also adds overhead. For normal data messages this is typically <5%, which can generally be ignored. However, for a network with voice applications, where many small packets are sent very frequently, the throughput consumed by the encryption header can become relatively large. Network designers must factor this loss of throughput into capacity planning in areas that are expected to support a high volume of VoIP.

Table 9 lists the throughput requirements for typical Wi-Fi network services.

Bandwidth vs. Throughput

The 802.11 physical layer operates at half-duplex because only one station may transmit on a wireless channel at the same time. 802.11 data rates such as 54 Mb/s or 300 Mb/s refer to one-way, “raw” physical layer bandwidth. No consideration is given for latency from an application perspective and no

Table 9 Service capacity and priority

Service Type Quantity and Unit

Occupied Throughput Direction Delay

Sensitivity Remarks

Voice Single caller 64 Kb/s Bidirectional Very Using common codecs

Video surveillance Single camera 512 - 2048 Kb/s One way (up or down)

Very Depends on video quality

Internet access service Per active subscriber 512 - 1024 Kb/s Asymmetric No Typically rate-limited



consideration is given for transmission “overhead”. In addition, such data rates represent single client to single AP speeds under best-case conditions; actual client bandwidth will be lower under real-world conditions and client counts.

By contrast, application performance is heavily dependent on round-trip, full-duplex network performance. Even applications that have asymmetric traffic profiles, such as video cameras, must still employ delivery guarantee mechanisms which require acknowledgements, windowing, and sequencing at upper layers of the protocol stack. Latency in all layers and in any network element can significantly reduce performance.

As a result, application developers typically express their network requirements in terms of throughput. Throughput is defined as the effective data transfer rate at the application layer, and can be measured as an average or a peak.

Therefore, we need a mechanism to convert between the bandwidth values that will be used for WLAN design and the throughput values that must be guaranteed to applications. The simplest and most conservative technique is the following formula:

802.11n Throughput = Bandwidth * 0.50 802.11a/b/g Throughput = Bandwidth * 0.40

In other words, 300Mb/s of half-duplex bandwidth provides 150 Mb/s of full-duplex throughput with 802.11n. If a client is connecting at a lower data rate, say 78Mb/s (also known as MCS12) it could expect a maximum throughput of 39Mb/s. Note that these data rate examples assume the use of 40 MHz channels. With outdoor deployments, it is more likely that 20 MHz channels will be used on both the access and backhaul layers.

Client Throughput Requirements

Traffic modeling is a useful tool to set expectations for service levels on the network. The idea is to clearly outline coverage areas for each mix of applications, based on application demands and customer usage behaviors. The result of traffic modeling is the average and peak calculated offered load during busy times, which the system should then be designed to handle.

A vital step in designing the access layer of an outdoor network is to determine the minimum client PHY-layer data rates to an AP based on the traffic model. This in turn determines the maximum allowable distance to an AP during the RF design phase, delivering adequate Wi-Fi coverage so that high throughput connections are possible from the client to the AP or AirMesh router. Network planners then move on to ensure that backhaul calculations connections deliver sufficient capacity where it is needed.

Data modeling includes: Coverage area and client devices by priority Data applications in priority and subsequent throughput calculations Number of associated users and the expected number of concurrent transmitters (e.g. duty

cycle) Average PHY data rate and total application throughput for a typical AP radio

A common recommended user number of each AP is 15 to 25 for data-only users. For users with VoIP wireless handsets, the user number is reduced to 7 to 8 voice users per AP, when data is present. In large public venues, 50 users per AP is not uncommon as a planning value, however, this represents a bandwidth limitation and may not always be achievable based on the ability or inability to re-use



channels in each venue. Though hundreds of users can be associated to an AP, the goal is to determine the number of active users that can be supported by available RF capacity. This number should be used as a guideline and may vary widely when considering outdoor networks. Different regions and concentrations of users require more client and backhaul capacity than less used areas.

Oversubscription Ratio

When designing an outdoor network, begin by considering the number of fixed permanent terminals, such as IP video cameras for surveillance, Wi-Fi voice telephones, and RFID or scanner applications. Then consider other types of client access devices that are expected to be used, typically enterprise or public Internet access users, with an expectation of seamless roaming across the coverage area.

Multiply the bandwidth requirements of each of these applications by the number of concurrent devices using each application. Then add all of the application requirements together to determine the total expected offered load for each subarea within the network.

However, this is not a complete picture. If only one of four online users is actually communicating at a time, the duty cycle of each client can be considered to be 25%. The oversubscription ratio is the inverse of the duty cycle, or 4:1 in this example.

For voice services, the oversubscription ratio usually can be large, such as 8:1, but data services often trend up and down with a peak time, based on user behavior. Enterprise networks see a mid-morning and mid-afternoon peak in activity, which planners must prepare for. For this reason, pure data services may require a subscription ratio as small as 2:1 or lower. Network planners must set realistic objectives, but network operators often trade these usage trends and subscription ratios against the need for adding unnecessary or expensive capacity. Large public networks may use different oversubscriptions ratios and time of day, but the key point is that the network must accommodate the highest load anticipated within each subarea of the outdoor network.

Strategic Throughput Reservation

Consider the long-term growth requirements when the wireless network capacity is specified. When the capacity of a wired network is planned, operators usually reserve 20% or more for network expansion and optimization. For best results, when planning your outdoor Wi-Fi network, you should do the same to accommodate unanticipated coverage gaps or the need to add mesh portals to busy areas that may need more capacity as more devices and new applications emerge.



Determining Cell SizeThe distance between clients and APs is one of the main factors that determines the throughput of an access layer. Therefore, the minimum throughput requirement from the capacity plan directly determines the number of access layer radios that are required per km2 (mi2). As the distance from the client to the AP increases, the wireless signal strength decreases. Lower signal strengths produce lower connection data rates. Adding more access layer radios can reduce the AP-client distance and increase performance. In this section, you'll learn about some of the complex factors affecting PHY-layer data rates.

Matching Client and AP Power

The biggest factor that limits the size of a access layer Wi-Fi cell is the transmit power of the client devices that will be supported in each area. Wireless communications are not one-way; they require a solid round-trip connection with equal performance to and from the client device. Professional wireless designers routinely conduct surveys to measure the received signal strength indicator (RSSI) or SNR at various distances from each AP. However, this measurement is only one-half of the round-trip communications link because the client device must also be capable of transmitting to the AP successfully.

Common sense suggests that handheld devices with batteries and small antennas that operate at street level will transmit over a much shorter distance than an AP mounted up in the air using high-gain directional antennas.

When planning access layer coverage, the transmit power of the expected client devices determines how far apart the mesh routers can be placed. The devices must remain in range of these weaker client devices. Coverage must overlap to ensure no signal loss to maintain seamless, session-persistent roaming.

For example the AP-175 is capable of 25 dBm of conducted power with both chains, but the latest models of voice handsets and ruggedized mobile terminals are limited to 12 dBm plus a small amount (~3 dB) of gain from the antenna. If this important imbalance is overlooked, Wi-Fi users would say that they have “5 bars” on the client display, but they have a very poor connection. This problem occurs because the client can hear the AP signal, but the AP cannot hear the client well enough to establish a reliable link.

Figure 80 Mismatched client and AP power output

Ret

ail_

130

20 dBm Output

20 dBmOutput

12 dBmOutput



Free-Space RF Propagation

The RF signals emitted by an antenna go through significant attenuation, even in free space (i.e., no obstructions between the transmitter and the receiver), before they reach the intended recipient. The free-space propagation loss in dB is given by the formula:

Lp = 32.4 + 20 log10(f)+ 10nlog10(d)

The frequency of transmission f is specified in MHz and the distance d is specified in kilometers. The higher the transmission frequency, the higher the propagation loss is for the same distance.

The parameter n is known as the path loss exponent (indicating how fast the signal attenuates with distance), whose value is 2 for free-space communication. In non-line-of-sight communication and in indoor environments, many other factors such as attenuation due to absorption, reflections and multipath come into this equation. If the types of material and the exact amount of the attenuation are known, these losses may be added to the propagation loss formula to help you calculate the actual loss. In a mixed environment, such as a warehouse, a different path loss exponent value may be used instead to approximate the path loss. For example, a value of 2.5 to 4 may be typical of most indoor environments, though the path loss exponent can be as high as 8 in some RF unfriendly environments.

Figure 81 shows how path loss increases with distance in the 2.4 GHz and 5.8 GHz bands.

Figure 81 Link loss across a range of common outdoor WLAN distances



You may have noticed when using a connection utility or site survey software that the signal strength of an AP drops off very fast near the AP, and then much more slowly. The figure clearly shows why this is the case, due to the exponential nature of RF signal propagation. In fact, for every doubling of distance, an RF signal will experience 6dB of path loss.

Figure 82 Effect of distance on mesh network capacity

Effect of Path Loss on Data Rate and Throughput

802.11 protocols have different physical layer connection rates, which are usually realized based on signal quality and other environmental factors specific to each location. Radio manufacturers characterize the PHY-layer data rates that can be achieved based on minimum levels of required SNR for each individual rate. Figure 83 shows data from one combination of radio and antenna that show how the predicted data rate decreases as path loss increases.

Figure 83 TCP throughput with increasing path loss (theoretical)

arun

_034

8

Lowthroughput

Radius

Mediumthroughput

Highthroughput



If the mesh link will be used to backhaul wireless clients (such as laptops or wireless terminals) as opposed to wired IP clients (such as a video camera or server), it is also useful to have an estimate of the AP to client throughput, again by modeling or measurement of the specific AP and client combinations. For MIMO connections, it is important to include the antennas that are planned to be used at both ends of the client connection in the test or analysis. The following Figure 84 shows an example test result using a MIMO client with 5 dBi antennas to the Aruba AP175P with ANT-2x2-2005 (5 dBi omnidirectional) antenna array. This is a common configuration, such as might be found between a mobile vehicle with a roof mounted antenna, and a mesh point providing access layer service. The results presented are using a 20 MHz, 2.4 GHz channel in the upstream direction. Depending on the application, testing or modeling of both the upstream and downstream directions independently may be an important.

Figure 84 Outdoor MIMO client rate vs. range test (2.4GHz, HT20, 5dBi each end)

Of course, the aggregate capacity of a cell depends on the percentage of total clients that are farther away (higher path loss). When a far client is transmitting at a low rate, nearby clients must wait. It is an ironic challenge that the bigger the coverage area of an AP, which can reach many more people than a small cell, the fewer users it can serve based on bandwidth availability.

In cases where more capacity is required, the cell size can be further decreased, which lessens this effect because fewer users will be connecting to each AP and through each mesh portal. This standard step is designed to accommodate growth, and network operators should plan for this to happen as the network experiences increasing numbers of users in the different areas. AirMesh products are designed with flexible capacity growth in mind and mesh nodes can be added easily.

Estimate Path Losses

A major part of coverage planning is accurate RF link budget calculations and propagation models that are specific to the area to be served. Each link budget accommodates the maximum path loss through



the air and walls and trees, and includes loss from antennas and connectors. To form a link budget, calculate the maximum path loss between each transmitter and receiver in a wireless system to arrive at the limiting link budget:

The forward path link budget from the AP to the client The reverse link budget from the client radio to the AP

In other words, the calculations in the formula in the Link Budget Calculation and Link Balance section must be applied in both directions. To achieve a balanced link budget for planning, the lowest value should be used when outlining where nodes should be placed to ensure that they are always in range. Typically, the limiting link budget will be from the client to the AP, which we learned earlier determines what bandwidth and modulation type are available to deliver the required TCP throughput.

Link Budget Calculation and Link Balance

Because each bit rate requires a specific minimum receiver sensitivity for a given radio, any wireless network (simply referred to as link for the purpose of this discussion) design must estimate the available link budget in dB to make sure that the link budget is at least 0 dB for the highest bit rate desired. It is also a good practice to leave some reasonable margin (e.g., 10 dB) in the link budget to accommodate any variations in signal strength caused by interferers or reflectors and to increase the reliability of the link. Use the link budget analysis to estimate the range or capacity or to select an antenna.

The first step in the calculation of the link budget is to calculate the received power at the receiver.

The received power is given as:

Received Power = Radiated Power (or EIRP) — Path Loss + Receiver Gain

The equivalent isotropic radiated power (EIRP) is the correct technical term) in dBm is given as:

EIRP (dBm) = Radio Transmit Power (dBm) — Cable/Connector/Switch Loss (dB) at Transmitter + Transmit Antenna Gain (dBi)

The path loss can be calculated using the appropriate path loss formula, as discussed earlier, and may include attenuations caused by other objects in the path, if known. The Receiver Gain is given as:

Receiver Gain = Receive Antenna Gain (dBi) — Cable/Connector/Switch Loss (dB) at Receiver

When the received power (or signal strength) is known, the link budget can be calculated by subtracting the receive sensitivity of the receiver from the received power:

Link Budget = Received Power — Receive Sensitivity

The noise floor at the receiver can be subtracted from the received power to calculate the SNR. If the noise is lower than the Rx sensitivity, the link will be limited by the Rx sensitivity. Otherwise, the link will be limited by the noise floor.

For example, with 30 dBm EIRP (e.g., 23 dBm transmit power, 10 dBi antenna gain, and 3 dB cable/connector loss) in 2.4 GHz, the signal attenuates to -50 dBm at 100 meters in free space. For a receiver with receive gain of 0 dB (e.g., 2 dBi Receiver antenna and 2 dB cable/connector loss), the



received power is -50 dBm. If the receive sensitivity is -91 dBm for 1 Mbps, then the link margin is 41 dB. However, if the noise floor is -85 dBm, then the SNR is 35 dB. In either case, the signal is more than enough to decode the 1 Mbps data rate. However, as the distance increases the noise floor will be the limiting factor in this specific example.

The choice of an antenna and transmit power are dictated by the specific requirements of the wireless system. For example, in order to create symmetric links (i.e., each end of the wireless link can talk to the other end with same bit rate at the same reliability), the transmit power at both ends should be kept the same, assuming the RX sensitivity and noise floor are identical at both ends. The range of the system for such symmetric networks should be increased by selecting the appropriate antennas on both ends, rather than increasing the transmit power at one end (which increases the range in only one direction). It is also important to calculate the link budget in both directions separately to make sure that the bidirectional system requirements are met, given the system parameters in each direction.

By solving the equation above for a desired value such as power at the receiver, free space propagation models can be used to calculate the RF levels of access and backhaul links, including a fade margin to address expected environmental variances. Backhaul link calculations are more predictable, because most have clear LOS due to their higher mounting heights. Less accuracy is found in the access links because of NLOS expectations and unstable signal strength due to complex wave environments and more interference and unpredictable noise at street level. Aruba provides planning tools to help automate this process and minimize the mathematics. Nevertheless, outdoor engineers are encouraged to understand the fundamentals behind the planning tools to assist in troubleshooting network issues.

Path Loss Due to Cumulative RF Absorption

Natural features and the materials used to construct structures produce variable loss in RF signals. The attenuation numbers for concrete walls are somewhat controversial. This is because there are different types of concrete materials in use in different parts of the world, and the thickness and coating of each type differs depending on whether it is used in floors or in interior or exterior walls. Brick walls usually have attenuation at the lower end of the range shown in Table 10.

Table 10 Attenuation of common building materials

2.4 GHz 5 GHz

Interior drywall 3-4 dB 3-5 dB

Cubicle wall 2-5 dB 4-9 dB

Wood door (Hollow - Solid) 3-4 dB 6-7 dB

Brick/Concrete wall 6-18 dB 10-30 dB

Glass/Window (not tinted) 2-3 dB 6-8 dB

Double-pane coated glass 13 dB 20 dB

Steel/Fire exit door 13-19 dB 25-32 dB



In outdoor areas, trees can have significant impact on RF propagation. The attenuation caused by trees varies significantly depending on the shape and thickness of the foliage. The rule of thumb is about 1.2 dB of attenuation per meter for 5 GHz and about 0.5 dB per meter for 2.4 GHz. However, rain, snow, and fog attenuation outdoors is very small for frequencies under 10 GHz. The rain attenuation at 5 GHz is barely noticeable (< 1 dB per kilometer).

Figure 85 Cumulative losses from free space and structural absorption

The diagram above shows how such a model would be applied to an outdoor radio system. Building penetration and walls are usually assigned figures of 6 - 20dB, whereas different types of terrain, such as trees, can be modeled by modifying the path loss coefficient. All the losses along the path are additive, of course. This type of model is used outdoors, where cellular operators have found that the large amount data available (for example, from satellite surveys) allows them to accurately model terrain and individual structures with considerable success, although results are statistical rather than precise.

arun

_044

8

Signalstrength

(dB scale)

Distance

5 mindoors

5 mindoors

Transmitter Receiver

10 dBwall

6 dBwall

100 mfree space

100 mfree space

50 mwooded region



Path Loss Modeling for Indoor Coverage by Outdoor APs

In access layer RF design you can use the recursive methods and the Keenan-Motley models1 for calculating path loss. These methods can be used wireless backhaul links and AP network-to-client links. In all cases, consider penetration loss when the RF signals must pass through walls, windows, and floors. Frequently, outdoor networks are designed to reach the back wall of the street-facing rooms. The designs assume that some type of window or other RF entrance to the building must be considered.

In these formulae: d is the distance λ is the wavelength N is the number of walls Lw is the penetration loss of that wall (many references available on-line) M is the number of floors Lf is the penetration loss of each floor.

Propagation models used in macro cellular telephone models are not applicable in Wi-Fi mesh systems. Telephony base stations are most frequently installed on rooftops and operate at lower frequencies with different loss characteristics. In mesh systems, the antennas of the APs are typically installed closer to street level, to avoid the penetration loss associated with buildings, terrain, foliage, walls, and other obstacles.

Figure 86 Estimating path loss in multistory structures

For nonmetalized windows or other dielectric layers in the window, the loss is typically ~3 - 6 dB. The loss does not necessarily increase for multiple layers and a thin layer of nonmetallic material imparts little or no loss. In this context, “thin” is defined as less than λ/20, where λ = wavelength and λ is 12 cm (4.72 in.) at 2.45 GHz.

1. J. M. Keenan, A. J. Motley, “Radio Coverage in Buildings”, British Telecom Technology Journal, Vol. 8, No. 1, Jan 1990.

PL =

arun

_042

8

Gant-AP

PAP-outRsen-AP

PSTA-outRsen-STA

PL

LfeederLwall

Lfloor



Summary

As we have seen, the range and coverage of each client access AP will depend on a number of RF considerations:

1. The lesser of AP or client device power2. The sum of the AP and client antenna gains3. The pattern of the AP antenna and client antennas4. The receive sensitivity of the AP and client radios5. The target data rate and associated 802.11 required SNR6. Consideration of absorption and losses

Using the Aruba 3D Outdoor RF PlannerLink budget calculation for horizontal coverage is a laborious process, especially for large mesh networks. In Chapter 4, we learned how important it is to also calculate the vertical coverage of every antenna. In the past, wireless engineers typically used complex spreadsheets to estimate horizontal coverage and relied on their hard-won experience to judge the vertical coverage.

Aruba introduced the patented 3D Outdoor RF Planner in 2007 to make this process faster, simpler and more accurate. This 3D visualization tool is fully integrated with Google Earth and provides the ability to accurately predict outdoor WLAN coverage based on RF calculations (bandwidth, distances, gain, coverage, etc.) using real-world data including actual access point and antenna polar pattern data. This free online tool accepts Google Earth .kmz input files and then assists the wireless engineer in visualizing large outdoor networks in 3D, with various combinations of antennas and transmit power.

Figure 87 Example of Google Earth mapping functions for planning



Access to the 3D Outdoor RF Planner is online and no copies of your work are saved. You input a Google.kmz file and enter the APs and antennas you want to use. Then the tool calculates a propagation heat-map for several data rates and signal strengths and saves that as a file that you can download and play back in Google Earth Pro. You can work many variations with the tool, to find the right balance between coverage and capacity. It is located online at:

http://outdoorplanner.arubanetworks.com

While the Aruba 3D Outdoor RF Planner is free, you must purchase a Google Earth Pro license when using it for commercial purposes. These online tools can help you to create a network plan that is very close to the actual design before you travel onsite, which minimizes field survey work later.

Finished RF Plan Examples

The following figures show examples of a completed RF plan.

Figure 88 Coverage example: Simultaneous mesh/coverage view with color-coded channels in two groups


http://outdoorplanner.arubanetworks.com


Figure 89 Client throughput display

Figure 90 3D view coverage



Chapter 9: Planning the Mesh Backhaul Layer

In the previous chapter, we learned to calculate link budgets based on the expected client device radio capabilities to ensure that the access layer of an outdoor network will support the minimum required client connection rates. These calculations lead to a nominal distance between mesh points or, said differently, the node density of each area stated as nodes per square kilometer or nodes per square mile. In this chapter, the discussion changes to planning network capacity of the mesh itself using both individual RF links and end-to-end paths in the backhaul layer.

The general methodology Aruba recommends for backhaul layer planning is shown in the figure below:

Figure 91 Planning methodology for mesh backhaul layer

As you can see, this is an iterative experience of continual adjustments based on technical and nontechnical factors.

Identify Portal CandidatesEvery backhaul design must begin by identifying a list of potential mesh portal locations, with a preference for buildings that are owned by individuals or companies that are participating in the network operation, which may be the local school district, city government, public safety agencies, or port or transportation facility operators. Create this list to reduce the time needed for the civil permitting processes and increase the probability of obtaining the mounting assets with optimal LOS conditions for the network. Always start with the obvious beneficiaries of the network and move on to other possibilities only if necessary.

N O T E

The term “mesh” denotes a specific network topology. However, both ArubaOS and AirMesh use the same mesh software regardless of the topology or number of links. Therefore, we use the term “mesh” generally to describe any wireless backhaul connection.

arun

_054

7Iterate until allrequirements are met

Capacity Plan RF Design

Identify usablechannel count

Choose HT40or HT20

Choose RFbackhaul topology

Identify PortalCandidates

Choose capacityinjection topology

Computeingress load

Computeegress load

Modelend-to-end flows

Finalizeportal locations

Add relaynodes if needed

Choose antennasand bearings

Discovery Topologies Model

Create 3DCoverage Model

Create bill ofmaterials

Aruba Networks, Inc. Planning the Mesh Backhaul Layer | 119


Choose RF Backhaul TopologyWhen mesh networks are built, several very common topologies are used over and over. These are the “building blocks” of any campus extension or outdoor mesh network. This section describes the most common scenarios that can be combined to deliver the required coverage and capacity across small or large geographic areas, using the Aruba MSR product line.

Serial Point-to-Point Connections

Figure 92 shows a simple deployment scenario where two locations are connected using a point-to-point link. Based on a strong link budget, these links can achieve up to 300 Mb/s of physical layer throughput with minimal loss due to management overhead data.

Figure 92 Serial point-to-point link

When many serial PtP links are extended over large distances with many hops, it is commonly referred to as a linear or serial mesh. Figure 93 is a multichannel design in this topology to preserve and maximize end-to-end throughput. We saw in Chapter 3: Outdoor Access Points and Multichannel Backhaul the profound performance difference of a multichannel design over a single-channel system.

Figure 93 Serial multi-channel mesh links in a linear topology

Parallel Point-to-Multipoint Connections

Figure 94 shows three nodes that connect to the wired mesh portal directly.

Figure 94 Parallel point-to-multipoint single channel links

arun

_039

5

Node 1 Node 2

149

arun

_034

1

Point ofpresence

Ch. 149 Ch. 157 Ch. 153 Ch. 161

Point ofpresence

149149

149

arun

_033

9



Each unwired mesh point uses one radio to build a mesh link that is connected to the mesh portal in parallel. In this topology, full throughput can be assumed between the mesh portal and each mesh point. However, topology also requires that each mesh portal fall within the antenna pattern of the mesh portal and share a common channel.

Full Mesh in a Multi-Gateway Design

Serial point-to-point and parallel point-to-multipoint topologies can be combined with multiple wired gateways to construct a robust mesh. If a network contains multiple mesh portals, don’t forget to consider the throughput balance in the case of a failed or unreachable portal. Wired mesh portals are often backup connections for other large mesh areas and they become critical components in delivering backup network capacity when a critical link fails. Both load scenarios should be considered if the mesh network directs traffic to available wired egress points. If engineered too conservatively, a failure of one mesh portal could cause users in adjacent areas to experience slow service. If the primary portal fails, all users in the affected area are automatically routed to the remaining portals that do have access. So the pain of the outage is spread by reduced capacity for all affected users, but no users completely lose connectivity. Figure 95 illustrates this concept in the mesh network topology.

Figure 95 Mesh network of any scale, campus to metropolitan area networks

One powerful feature of fourth-generation mesh networks is the use of MIMO-based directional antennas to deliver much higher capacities. As we will see in Chapter 10: Site Surveys for Large Outdoor Networks on page 137, this changes the outdoor site survey process in some important ways. For instance, when surveying for first- and second-generation meshes with omnidirectional antennas, it was always a struggle to find a pole that had ideal LOS to all of its neighbors. By using high-gain directionals on each leg, the wireless architect has dramatically more flexibility to choose poles. Increased transmit and receive gain also improves performance through trees and other ground clutter.

149

149

149

153

153

153

157

157

161

161

161

165

165

165

arun

_034

0

Portal 2

Portal 1

Internet

Internet



Choose Capacity Injection TopologyWith the optimal RF backhaul topology selected, the wireless architect should turn next to the capacity injection topology. You may think of this as the L3 traffic topology for flows that leave or enter the mesh. RF backhaul is primarily concerned with L1 physical paths without regard for the mesh role of any node is a portal or a point. By contrast, the capacity injection topology designates which nodes are to be mesh portals, and what the size of their uplinks needs to be.

There are two basic injection topologies in a mesh network that utilizes L3 routing: end-fed and center-fed. Each type is further differentiated based on whether a single-channel or multi-channel RF topology underlies the layer 3 path. Advanced designs may combine both types into hybrid topologies that provide the greatest possible IP throughput while conserving channel space.

The choice of which nodes to make into portals can have profound consequences for the downstream capacity of a mesh network. As a result, injection planning is always an iterative process. The wireless architect will create a first draft and then run end-to-end calculations. Many more drafts follow as the architect does what-if scenario analysis to arrive at the maximum ingress & egress capacity for the lowest cost.

End-Fed Injection Topologies

As the name implies, an end-fed mesh places the uplink or capacity injection point at the end of a string of mesh nodes. Figure 96 shows a typical end-fed capacity injection topology.

Figure 96 End-fed capacity injection topologies

Both single-channel and multi-channel RF topologies can be end-fed. The specific combination of RF and injection topologies chosen determines the capacity of that system (or branch). In the next section, we will review how to model the actual throughput of these combinations.

arun

_056

4

Singlechannel

Multi-channel

Ch. 161Ch. 153Ch. 157Ch. 149

3 hops2 hops1 hop

3 hops2 hops1 hop

Internet

Ch. 149Ch. 149Ch. 149Ch. 149



While the injection topology is effectively linear in terms of IP packet flow, the underlying layer 1 RF backhaul topology could be any of the choices covered in the last section. For example, the nested point-to-multipoint network shown in Figure 97 is an end-fed design.

Figure 97 End-fed injection topology overlaid on hierarchical point-to-multipoint RF topology

Most legacy L2 mesh networks are by definition end-fed designs. In Figure 97 you can plainly see that the single mesh portal serves as an end-fed injection point for the entire system. This is because layer 2 meshes funnel all traffic up to a single common uplink. Even when a secondary mesh portal is deployed for redundancy, in an L2 mesh will generally use STP to eliminate loops within the mesh so there is only one active gateway at any time.

149

153153

157

161

157

161

157165

arun

_056

5

Internet



Center-Fed Injection Topologies

In a center-fed design, the uplink node is chosen to be approximately in the middle of a string of mesh nodes. Figure 98 shows a center-fed topology in both the single-channel and multi-channel case.

Figure 98 Center-fed capacity injection topologies

Center-fed designs are intended to reduce hopcount, therefore maximizing capacity from the leaf nodes and minimizing uplink latency. While they can be used with either single- or multi-channel RF topologies, they are especially useful in single-channel systems where keeping hopcounts very low is vital to achieve the capacity objectives of the mesh.

One very powerful form of center fed injection is the dual-uplink ring topology. It is only available in fourth generation mesh architectures that employ native layer 3 routing inside the mesh. To illustrate why, consider a ring of mesh nodes as shown in Figure 99. Imagine that we have deployed mesh nodes on the corners of a city with streets in a standard grid configuration. Each node is only able to see two other nodes due to the height of buildings in between.

Figure 99 Dual center-fed injection topology overlaid on a ring RF topology

arun

_056

6

Single channel

Multi-channel

Ch. 161Ch. 153Ch. 157Ch. 149

2 hops1 hop2 hops 1 hop

2 hops1 hop2 hops 1 hop

Internet

Ch. 149Ch. 149Ch. 149Ch. 149

Ch. 161

Ch. 149

Ch. 153

Ch. 149

161 161

153157

149165

arun

_056

7

Internet

Internet

Portal B

Portal B



In a layer 3 ring with two mesh portals, traffic can flow either direction inside the ring to the portal with the least cost path. With AWR, both injection nodes can be used simultaneously and load can be balanced across them. If any of the nodes opposite the portal were to fail, the system falls back to being a linear mesh with two center-fed segments. If either of the uplinks were to fail, the system retains the ring topology with a reduced overall uplink capacity.

Hybrid Topologies

It is possible to combine the end-fed and center-fed topologies into a hybrid topology. One of the most common such applications is to use an end-fed, multi-channel backbone to anchor a series of center-fed single-channel branches as shown in Figure 100.

Figure 100 Hybrid topology that combines end-fed and center-fed

There are multiple scenarios where the wireless architect may wish to employ this approach: Geography: The physical layout of the coverage area lends itself to this design. For example, in

a metropolitan area where a main highway or ring road is used as a backbone to feed branches down individual streets.

Channel Count: The number of available backhaul channels is very limited, such that there are not enough to use multi-channel everywhere inside the coverage area.

Capacity: To create a very high-capacity mesh “core” with multiple load balanced uplinks, capable of serving much lower offered loads from individual branches

Redundancy: To limit hopcount increase from node failure, by leveraging the ring topology for the mesh “core” with multiple uplinks

arun

_056

8Internet



Maximum Hop Count

If a mesh nodes use multiple backhaul radios with and the uplink and downlink using well separated channels in the same band (40MHz minimum center frequency spacing), the throughput between any given pair of nodes changes very little as the number of hops increases. However, the uplink bandwidth must still be shared by all of the different traffic streams flowing through it. The total accumulated offered load from “tributary” mesh nodes along the path is the main factor to consider because wider coverage areas often bring more users and the need for more capacity.

For multichannel mesh backhaul, Aruba recommends a maximum hopcount of 10, with a preferred value closer to 6. If the mesh nodes will use a single backhaul radio, then the mesh becomes subject to the classic 1/N performance limit, where N is the number of hops. Aruba recommends limiting hop counts in single radio backhaul to no more than 3.

Maximum Number of Children

In an Aruba AirMesh mesh network, some number of mesh points associates with each mesh portal and create a mesh area. The AWR routing protocol is used within the mesh area to determine the optimum path. L3 operation of the infrastructure is critical to increasing the scale of the network, even if only L2 services are presented at the mesh edge using VPLM.

When large outdoor areas are covered, Aruba recommends that one outdoor mesh area contain no more than 50 child nodes. At a common node density of 10 APs per km2 (25 nodes per m2), this is sufficient to cover 5 km2 or 2 m2. If more nodes are required for a very large mesh, divide that coverage area into multiple smaller mesh areas by adding mesh portals. Realistically, the capacity demands on these Wi-Fi networks seldom allow such sparse deployments and the number of nodes per portal is generally much smaller than 50.

Ratio of Mesh Portals to Mesh Points

The number of mesh portals determines the maximum possible uplink throughput for all users connected in a given mesh area. In sparse networks (with relatively few nodes per square kilometer or square mile) many mesh points can cover a large area without a lot of mesh portals. Conversely, high-usage areas usually have many users or use applications like video surveillance, which has a high duty-cycle. In high-usage areas, reduce the ratio of unwired mesh points to each mesh portal, to ensure that capacity is available to deliver high-speed access to the wired network without the mesh portals becoming a network bottleneck.



For example, Figure 101 shows a mesh that has five nodes: one mesh portal and four mesh points. It is easy to see that if the wired network capacity to the mesh portal were limited to 45 Mb/s, the entire network would be forced to operate within that available capacity. To scale capacity to support future applications, add another portal location. Or convert any mesh point to a mesh portal by putting a DSL or other modem on a pole where service is available. This adds more capacity when it is needed and exactly where it is needed, without the requirement to predict where the usage may occur. A common strategy in outdoor broadband networking is to provide a good enough level of coverage during the initial build-out, then selectively add capacity as the network grows. At a high level, the strategy is to start with large areas, but make them smaller and smaller over time by adding mesh portals.

Figure 101 Wired uplink capacity shared by child mesh nodes

Capacity PlanningAs with planning any wireless network, a mesh backhaul layer requires many tradeoffs. For instance, a suboptimal portal location must be used because the owner of the preferred location will not agree to a lease. Or a relay node is required on a very long connection to maintain end-to-end throughput. This section will summarize some of these factors.

Determine Number of Usable Backhaul Channels

The 5-GHz band(s) allow many more nonoverlapping channels than 2.4 GHz. In the United States before 2007, the UNII-I, -II, and –III bands allowed the use of a total of thirteen 20-MHz channels (or six 40-MHz channels). The number of available 5-GHz channels varies significantly from country to country.

In 2007 the radio regulatory bodies in many countries allowed the use of the “UNII-II extended” band from 5470 MHz to 5725 MHz as long as UNII-II equipment was capable of Dynamic Frequency Selection (DFS). DFS requires that the AP monitor all RF channels for the presence of radar pulses and switch to a different channel if a radar system is located. Wi-Fi equipment that is DFS-certified can use the extended band, which adds up to another eleven 20-MHz channels or five 40-MHz channels (depending on the radio regulatory rules in each country).

The MSR series of wireless mesh routers also supports the 4.9 GHz licensed band. In the United States, up to two nonoverlapping 20-MHz channels are available between 4940MHz and 4990MHz.

149

153 157

161

165

arun

_039

3

Node 1Mesh portal

Node 5

Node 3Node 4

Node 2

Node 1

45Mbps

Internet



Japan allows up to four nonoverlapping 20MHz channels, or two non-overlapping 40MHz channels. See Appendix A: Allowed Wi-Fi Channels on page 173 for a complete list of available channels.

Remember that with AWR, licensed 4.9GHz channels can simply add to the total channel pool. They can also be reserved and assigned on a static per-link basis in high-interference or other challenging areas.

Choose Between 40MHz or 20MHz Channels

One of the most complex questions facing the wireless architect with an 802.11n outdoor mesh is whether to use bonded channels for backhaul links.

In general, Aruba recommends that one approach this question based on the network topology.

Campus Extension: Outdoor continuity or building perimeter coverage will generally use bonded channels for consistency with indoor networks, especially in University environments where laptops with full-featured NICs are used outside. However, in many countries the lower UNII channels are restricted for indoor-only use. Therefore, it may be necessary to use DFS channels to achieve channel bonding if desired by the wireless architect.

Outdoor Mesh: Most outdoor MIMO wireless mesh networks should only use 20-MHz channel widths, also known as HT20. Using high-throughput 40-MHz (HT40) channels reduces the number of radio channels by bonding them together. As we learned in Chapter 5, channel bonding is required to achieve the highest possible MCS rates of 300Mbps or better.

From an access-layer perspective, it is better to have 50 users each on two different HT20 channels than 100 users on one HT40 channel. Also, most handheld devices are not capable of taking full advantage of 40-MHz channels due to their limited processing power single spatial stream radios. HT40 channels are never expected to be used on the 2.4-GHz band for reasons that are beyond the scope of this guide.

The main benefit to using HT40 channels for access is the ability for individual stations to burst at the maximum PHY rate when only a portion of the users are trying to use the WLAN. However, in the auditorium scenario, we must support so many users in a single room that we need every possible channel. In this case, we accept a reduction in the maximum per-station burst rate during light loads in exchange for a greater total user capacity at all times.

To DFS or Not to DFS?

With as many as twenty 20-MHz channels, the 5-GHz band with DFS now has sufficient channels to employ channel bonding on a mesh backhaul in dozens of countries. Without DFS channels, channel bonding is really not a viable strategy because it consumes the channel space too rapidly. So why wouldn’t everyone use DFS?

Actual or false positive radar events can be extremely disruptive to a WLAN that attempts to use DFS channels. Users on DFS channels can potentially experience lengthy service interruptions from radar events. Because radar frequencies do not align with 802.11 channelization, such events can impact multiple Wi-Fi channels simultaneously. See Appendix B: DFS Operation on page 177 for a more detailed discussion of radar operation and DFS compatibility. The wireless architect must assess proximity of the network to radar sources in the 5250-MHz to 5725-MHz band. A DFS survey is strongly recommended as the best way to answer this question.



Compute Ingress Load

With the channel space determined, the next step is to quantify the expected offered load that will be presented to the mesh layer for transit. To simplify the problem, Aruba recommends dividing the client population into categories and looking at each separately. Common client categories in outdoor mesh networks include:

Fixed clients: Any client device which is stationary and generates continuous, predictable load. Typical examples include IP video cameras, SCADA or similar telemetry sensors, and remote buildings connected wirelessly by bridging across the mesh. For metropolitan mesh networks, you would also include providing access to indoor users via high-power customer premise routers.

Temporary clients: Devices which come and go from the access layer. Examples include smartphones, tablets and laptops used to access the mesh from homes or businesses. Other common temporary clients in metropolitan networks include mobile vehicles such as police or fire with rugged computers installed and higher gain antennas fixed to the roof. For enterprise networks such as railyards or manufacturing plants the mobile vehicles could be cranes, locomotives, trucks or other service vehicles.

You should have already computed most or all of this information if you followed the access layer planning methodology presented in Chapter 8: Planning the Access Layer. For each type of client, it will be necessary to assign an average offered load in kilobits or megabits per second. This is generally simplest for fixed clients, where an IP camera or sensor bitrate can be easily determined. For temporary clients, the wireless architect must make assumptions about the average number of clients per cell, and the average offered load of each.

Once the loads are assigned by client type, Aruba recommends constructing a spreadsheet with every node in the network, and totaling the fixed and temporary loads for each one.

Compute Egress Load

With all the attention to channel planning, the uplink throughput or wired egress of a Wi-Fi mesh is often overlooked. It is common to see costly and sophisticated mesh networks connected to low-capacity DSL uplinks, especially in metropolitan networks outside the United States.

Understanding the full offered load from the mesh to each portal is vital to any successful outdoor network. Be sure that with lower capacity wired uplink networks, the wired network itself does not become a bottleneck. Follow the complete process described in this chapter to accurately assess the required uplink bandwidth.



Estimate Bandwidth of Individual Mesh Links

With the ingress and egress loads computed, the wireless architect needs one more input before analysis of end-to-end traffic flows can be performed. That is an estimate of the IP bandwidth of each mesh link. Short links at high data rates will be able to carry much more traffic than long and/or impaired links at low data rates. This in turn directly affects the end-to-end analysis.

The Aruba 3D Outdoor RF Planner will automatically compute estimated bandwidths for every link in a mesh. This saves the architect significant time and effort in hand-computing link budgets for every possible link. Also, link budgets do not directly translate into speeds in MIMO, as spatial stream decorellation is not directly related to SNR. To help the architect gain a general feel for how a MIMO backhaul link can perform with various antenna combinations, Aruba provides real-world data below from our open test range.

Figure 102 5 GHz MIMO backhaul

Figure 102 provides the mesh test results for a single hop mesh using a 40 MHz channel, short guard interval (SGI) in the 5 GHz frequency band. The results shown were obtained with the Aruba AP175P and ANT-2x2-5614 (directional, 14 dBi) or ANT-2x2-5010 (omnidirectional, 10 dBi) in various combinations. The differences in throughput achieved are primarily a function of the different antenna gains used in pairs for this test. Different RF and capacity injection topologies may require combining directional and omnidirectional antennas on different links, so it is often useful to have a throughput estimate, based on either testing or modeling of the specific antenna pairs, for each antenna combination planned to be used in the network design.



Mesh Capacity Math for Single Channel Backhaul Systems

In “Comparing End-to-End Performance” in Chapter 3, we explained the throughput penalty that is experienced by single channel mesh backhaul topologies. Because wireless architects will want to employ hybrid injection topologies, it is worth understanding in more detail how to quantify the performance penalty on single channel backhauls. This will prepare you for the next step in the methodology where capacity planning is done on the entire backhaul layer.

For purposes of this discussion, we will make the following assumptions: All APs are using the same 5 GHz channel All APs can “hear” all of the other devices on the same channel All traffic uses all hops to get off of the network The APs are distributed in a linear fashion as would be the case for a street deployment The AP which is hardwired is at one of the mesh The propagation environment does not generate a lot of reflection

With Aruba's unique multi-polar antennas the solution provides the full 802.11n dual stream throughput rates

Allowing for some derating this analysis assumes 80 Mbps goodput in a 20 MHz channel No other channel impairments are present

With these assumptions in mind, let us consider single channel backhaul systems with varying hop counts.

Figure 103 One-hop system

In this case there is one hop. The leftmost mesh portal will exchange its traffic directly on the wire. The first mesh point on the right will exchange its traffic over the 80 Mbps connection to the mesh portal.

Figure 104 Two-hop system

In this case there is one hop to the first mesh point and two hops to the second mesh point. If only one unit is busy at a given time bandwidth available to the first mesh point is 80 Mbps ÷ 1 and the bandwidth available to the second mesh point is 80 Mbps ÷ 2. This is referred to as the peak rate available at each unit.

arun

_056

9

Ch. 149

1 Mbps

arun

_057

0

Ch. 149

1 Mbps1 Mbps

Ch. 149



When both units are busy and loaded equally the wireless architect now has to consider the total number of hops in determining the bandwidth that both devices can carry. For example assume that the first and second mesh points have a sustained ingress load of 1 Mbps.

The traffic from the first meshed unit has to use the 5 GHz channel one time to send the 1 Mbps back to the portal.

The traffic from the second meshed unit has to use the 5 GHz channel two times to send the 1 Mbps back to the portal.

So, 1 Mbps of ingress traffic on each unit generates 2 Mbps of egress traffic on the mesh portal as expected. However, this process generates a total of 3 Mbps of 5 GHz traffic on the channel. In this fashion we say that the average sustained date rate at each AP is equivalent to 80 Mbps ÷ 3 or 26.6 Mbps. Total system capacity of the meshed units is 53.2 Mbps so the single-channel mesh penalty for a two hop system is 33%.

Figure 105 Three-hop system

In this case there is one hop to the first mesh point, two hops to the second mesh point, and three hops to the third mesh point. If only one unit is busy at a given time, bandwidth available to the first meshed unit is 80Mbps ÷ 1, the bandwidth available to the second meshed unit is 80 Mbps ÷ 2, and the bandwidth available to the third meshed unit is 80Mbps ÷ 3.

For example assume that the first, second, and third mesh points have a sustained ingress load of 1 Mbps.

The traffic from the first mesh point has to use the 5 Ghz channel one time to send the 1 Mbps back to the portal.

The traffic from the second mesh point has to use the 5 GHz channel two times to send the 1 Mbps back to the portal.

The traffic from the third mesh point has to use the 5 GHz channel three times to send the 1 Mbps back to the portal.

So, 3 Mbps of combined ingress traffic generates 6 Mbps of 5 GHz traffic on the channel. In this fashion we say that the average sustained date rate at each AP is equivalent to 80 Mbps ÷ 6 or 13.3 Mbps. Total system capacity of the nodes is 39.9 Mbps so the single-channel mesh penalty for a three hop system is 50%.

Figure 106 Four-hop system

arun

_057

1

Ch. 149

1 Mbps1 Mbps1 Mbps

Ch. 149Ch. 149

arun

_057

2

Ch. 149

1 Mbps1 Mbps1 Mbps1 Mbps

Ch. 149Ch. 149Ch. 149



The logic just presented can be extended to higher order systems with arbitrary numbers of hops. We use N to denote the number of hops. We represent channel capacity with the variable C.

The peak rate that can ideally be delivered to the end of a single channel linear mesh decreases as C/N. The sustained average that can be delivered to each AP can be shown to be equal to 2C/(N*(N+1))

The single-channel mesh penalty can similarly be calculated as (100 - 200/(N+1))%.

Model End-to-End Traffic Flows

We can use our understanding of mesh capacity math to model the aggregate end-to-end offered load that is presented to wired uplinks at mesh portals, and to perform “what if?” analysis on various injection topology options. After you have calculated the throughput requirement for access connections and ensured that end user devices are close enough to each AP to achieve typical offered load onto the backhaul layer from each mesh node. Don't forget to include any wired devices such as video cameras or remote bridge links that also place load on the backhaul layer.

For example, in Figure 107 the Wi-Fi mesh network and the point-to-point throughput of each mesh link is 80 Mb/s, based on link budget planning for the mesh links. Let’s assume that a number of cameras (that operate at 4 Mb/s each) will be deployed in this area and let’s explore the impact of hop counts and aggregated data flows.

Figure 107 Point-to-multipoint multihop example

In Figure 107 if the portal is located at AP1 in an end-fed design, the maximum backhaul throughput of the whole network is only 80 Mb/s. The throughput is 80 Mb/s because all users that are not directly connected to AP1 must traverse the mesh to reach AP1 and the wired network. If a camera with a 4Mb/s bitrate connects directly to AP2, there is now 76 Mb/s of access throughput for all downstream nodes in the rest of the mesh. If each AP has 12 concurrent cameras, this throughput may not be sufficient. Therefore, AP3, AP4, and AP5 must limit their offered loads to 26 Mb/s each, even though their individual mesh links are capable of carrying more traffic. If each AP has 12 concurrent cameras, this uplink throughput will not be sufficient.

arun

_035

5

AP 3

AP 2

AP 1

AP 5

80 M80 M

80 M

AP 4

80 M Uplinkoption 2

Uplinkoption 1



Now consider the same topology, but use a center-fed topology with AP3 as the gateway or portal to the wired network. The backhaul capacity of the whole network is greatly increased. AP1 and AP2 share 80 Mb/s of backhaul and can provide about 40 Mb/s of access-layer throughput each. AP4, and AP5 each can offer a full 80 Mb/s of access throughput. The total supported egress load would therefore be 240 Mb/s for the same monthly uplink circuit cost.

RF DesignMultichannel mesh backhaul layers deliver very high end-to-end performance, but this comes at a price. The RF design of such a mesh requires more work, and careful planning of the bearing and mechanical tilts of the high-gain antennas that will be used. First- and second-generation meshes with their omni antennas were simpler in some respects, but suffered big performance penalties. So this is a small price to pay for the performance gain, but it is important to recognize the additional time required during the survey and planning process.

Let us consider a sample backhaul RF topology, which uses different channels on each backhaul radio link. In Figure 108 each radio talks to at least two neighbors that use the same channel. This is important to consider that fact when selecting the antenna type for each location. It is no longer preferable for an AP to be in range of a single upstream device, although this always remains an option when forming mesh links to distant endpoints using linear “spurs”. In general, the antenna pattern should accommodate strong links to at least two diverse upstream mesh points for system reliability.

Figure 108 Frequency reuse on access and backhaul networks

36

36

36

40

36

40

40

149

44

157

153 149

44

149

44

157

153

arun

_042

7

165

165

157

153Channel 1

1

11

6

1

6

1

11

6

11

6

1

11

1

11

66

6

11

11

1

6

1

6

11

1 1

11

Channel 6

Channel 11



Avoid engineering a link through obvious obstacles like buildings and trees, which can absorb the 802.11 radio signals. When possible, the best case for a strong link outdoors is a clear line of sight (LOS) or as close to that as you can achieve.

Planning Mesh Layers with the Aruba 3D Outdoor RF PlannerIn addition to providing three-dimensional visualization of client coverage, the Aruba 3D Outdoor RF Planner also predicts the performance of individual mesh links automatically.

When entering data into the online tool, the wireless architect can specify backhaul antennas, cable losses, mechanical tilts and much more. The tool performs horizontal and vertical antenna alignment checking, and estimates received signal strength and throughput for each link.

One particularly handy feature is the auto mesh computation function. The tool will automatically compute and visualize only the links that exceed a specified RSSI threshold. Of course, manual link specification is also supported.

This means that the tool is equally valuable for backhaul-only deployments that do not include client coverage. Video surveillance deployments are a good example of such a use case.

Here are some examples of mesh links that have been modeled with the tool:

Figure 109 Coverage example: Simultaneous mesh/coverage view with color-coded channels in two groups

N O T E




Figure 110 Elevation profile visible in Google Earth

Figure 111 Overhead standard plan view coverage with mesh links



Chapter 10: Site Surveys for Large Outdoor Networks

The survey process for an outdoor mesh network can be extremely labor intensive. Outdoor networking necessarily is more complicated and has many more external dependencies and inputs than a traditional indoor deployment has. An outdoor site survey is also critical to a successful deployment. No amount of modeling in Google Earth or studying photographs can replace the need for an engineer to physically stand at each and every proposed mounting location to verify that it will actually work. In fact, the engineer should ideally be in an aerial manlift at the exact height where the AP is expected to be mounted!

Aruba recommends the following basic process for an outdoor survey. You will customize the process on a site-by-site and project-by-project basis depending on the particulars of each one.

Figure 112 Outdoor survey process

This guide assumes that an experienced, insured, and licensed (if required) outdoor integrator has been selected to perform the critical survey, design, and installation activities. Specialized skills and test equipment are needed to successfully plan, install, and test large-scale outdoor mesh networks. Your Aruba representative can recommend a qualified outdoor integrator who can assist you in this process.

Create a “Soft” RF PlanBefore an outdoor wireless network is deployed, the environment must be evaluated. A successful evaluation of the environment enables the proper selection of Aruba APs and antennas and assists in their placement for optimal RF coverage.

The potentially immense scale of outdoor deployments requires consideration of factors that do not come into play in a typical indoor deployment:

Range or distance between APs must be considered during the planning phase, taking into consideration the AP to client link calculations. Available AP mounting locations are often far less flexible in an outdoor environment than an indoor environment. Regardless of these outdoor restrictions, the desired goal is increasingly to achieve throughput results similar to an indoor deployment.

Considering vertical antenna patterns and planning for elevation differences between APs and from AP to client can be critical to success. To plan for these differences in elevation, you must understand the full three-dimensional coverage pattern provided by the antennas that will be

arun

_056

1

Finalnetworkdesign

Civilapprovalprocess

Surveymesh

portals

Surveymeshpoints

Create“soft”

RF plan

Aruba Networks, Inc. Site Surveys for Large Outdoor Networks | 137


deployed in the environment. You must make sure that backhaul antennas are aligned in both horizontal and vertical planes.

The RF environment might change on a day to day basis, so consider nonfixed items, such as shipping containers, vehicles, and future building construction, when planning for an outdoor deployment.

The predesign or “soft RF plan” is critical to building the confidence to move ahead with further planning. For planning purposes, Aruba recommends that large coverage areas be broken into smaller areas of 1 km2 or 1 mi2 as shown in Figure 113. When breaking areas into subareas, that they should follow the terrain or the user communities. Areas can and will overlap in practice, but this approach allows smaller RF network surveys to be more easily divided among multiple survey teams. Develop a preliminary soft RF plan using a GIS tool such as GoogleEarth. Complete a first-order approximate placement of all mesh nodes in each grid square. Factor AP density by zone type (e.g. normal, QoS, paid). Choose the ideal mounting locations from an RF perspective, for instance you may wish to preferentially site APs at intersections to maximize the number of sight lines. Be as realistic as possible about site mounting locations, based on the node densities per area and known mesh portal locations.

Figure 113 Separating large coverage areas into 1 square mile / 1 square kilometer squares



Secondarily, pay particular attention to LOS and propagation for each backhaul mesh link. A unique link budget calculation might be required and the AP might need to be moved to achieve the acceptable link budget. Each site should be located where some antennas can cover the client access requirement. Others that have near LOS to other mesh radios may require remote installation of some antennas. When using this technique, be sure to account for cable loss in the link budgets.

Figure 114 Example “soft RF plan”

General Considerations for Choosing Mounting AssetsThe goal of any outdoor network survey is to identify readily accessible radio locations, especially wired network connections. On a campus, an even distribution of AP-175 mesh portals can be mounted on the various buildings around the campus and coverage can be extended using a few AP-175 mesh points. In larger outdoor networks, operators must often secure roof rights or civil permits for installation on street lights or other building structures for AirMesh routers.

Consider several factors when selecting mounting locations for mesh portals and mesh points. The locations must be physically strong enough to support the weight and wind-loading of the proposed radios and they must have access to electrical power. Do not select lights that may be bank-switched. Bank switches use a centralized photo-sensor to turn power on for entire blocks of lights, but otherwise power at each light remains off during daylight hours. This detail is easy to overlook, because most site surveys are done in full daylight and power to the lamps may be assumed. To guarantee that a series



of lights are independently powered, observe whether they turn on individually or as an entire bank at dusk.

Identifying RF Absorbers, Reflectors, and Interferers

It is critical to identify RF absorbers, reflectors, and interference sources while out in the field during the installation phase. Ensure that these sources are taken into consideration when installing and mounting an AP to its fixed outdoor location.

Selecting Mounting Locations for Mesh PointsIn general, link budget planning to support the desired client devices delivers a desirable distance between mesh points, which is also described as the node density of a mesh network. However, in outdoor networks, radios must be situated wherever mounting locations can be found. The goal is to closely meet the node density requirements that are identified for each square mile or square kilometer as physically achievable based on mounting assets that exist in each area.

Street lights get a lot of attention as mounting locations for a few reasons. They can handle the weight of the radio; they may be below the tree line (so the light reaches the street); they are out in the center of the street rather than to one side or another; and they may have electricity in some form already available.

When you consider antenna locations, combine built-in down tilt with mechanical alignments that come from use of features on the mounting bracket. But, remember that not all brackets allow positioning in all dimensions. In general, the installer should attempt to keep antennas vertically or horizontally aligned for many reasons, not the least being that this is the most visually appealing.

Minimize installations that are close to obstacles like metal signs, walls, and floors and instead try to create cells that offer good LOS to subscribers.

N O T E

If bank switched lights cannot be avoided, it will be necessary to budget for battery-assist cabinets for every pole that is bank-switched. Your Aruba representative can help you identify a vendor that can provide these products.

RF Absorbers Cement and concrete structures Trees and vegetation Brick walls

RF Reflectors Metal objects, such as roof-installed air-conditioning equipment, chain link fences, wire fences, or water pipes - shipping containers

RF Interference Sources Other 802.11 or non-WiFi access equipment operating nearbynvers Industrial RF welding equipment or other Industrial, Scientific and Medical (ISM) equipment that

utilizes RF to heat or alter the physical properties of materials Military, commercial aviation, or weather radar systems



Performing the Survey

With all the preparation complete, it is finally time to begin the survey. If there are multiple grid squares, assign them to separate survey teams. Then each team will visit every location in the soft RF plan in their square.

At each location, these tasks will be performed: Select a physical pole using the mandatory pole selection criteria described below Perform 2.4/5GHz spectrum clearing using a bucket truck at intended mount height For troublesome locations, note the nature of the problem, such as:

All 2.4GHz/5GHz useable channels are taken The noise floor, in either band, is higher than expected (usually higher than -85dBm) Local narrowband interference

Look for backup / alternate mounting locations with desirable spectrum characteristics Discard pole if spectrum characteristics are undesirable, and an alternative can be provided Record pole tag(s) and GPS position(s) Photograph pole(s) from multiple angles & identify filename Take any further notes on how pole can be identified, such as

“in front of building front door with address…” “3rd pole south from NE corner of Main street and 1st Avenue”

To simplify the choice of choosing poles and ensure consistency among multiple survey teams, it is helpful to define mandatory and optional selection criteria during the discovery phase. Mandatory criteria are non-negotiable. Every pole must meet those requirements to make it from the soft RF plan into the final design. Examples of such factors include, but are not limited to:

Pole is owned by an eligible provider participating in the project The pole has no physical damage and is standing exactly vertical The concrete to which the pole is mounted has no physical damage and the pole does not sway

unexpectedly when force is applied Lamp has a photo sensor with an installed photocell (ensures constant power) Lamp voltage is <=240V (using power company supplied criteria) Mesh node has <40% Fresnel-obstructed LOS to all adjacent neighbors If the mesh point is adjacent to a portal, it must have unobstructed LOS If mounting on a lamp arm, the arm is of sufficient length for Aruba pole-mount kit



Choosing a Pole

Each location where you choose to mount a radio should have these characteristics: Meets all of the mandatory selection criteria; the optimal height that has the best LOS to all

adjacent portals and mesh points A good view of the street level that is to be covered with Wi-Fi access at 2.4 or 4.9 GHz - minimal

trees or other ground clutter obstructing sight lines

Figure 115 Good mounting location versus a bad mounting location

Intersections are preferred over the middle of a block to maximize coverage. As Figure 115 shows, it may be better to mount the radio out over the street rather than close to the pole itself if direct-mount omnis are being used. A location that is close to the pole may obstruct coverage behind the pole and may have less LOS to other mesh nodes. However, if directional antennas are being used, you may mount directly to the vertical section of any pole without concern.

Figure 116 Flexible mounting of Aruba APs to vertical poles and lamp arms

When mounting locations are selected, a primary consideration for the operational expenses of the network is who owns the asset. Whether it is a rooftop or a city-owned street light, the owner must

No ground clutter Heavy ground clutter

Arm mount with omnis Pole mount with directionals



grant access. Often the owner will offer a lease agreement allowing the mesh routers or AP-175 to be mounted and powered if they are given compensation and indemnification for any possible accidental damage.

Evaluating Pole Power From the Ground

It is not always possible to determine in advance whether the poles in a soft RF plan have constant, unswitched power. Or whether the AC voltage is compatible with the radio equipment.

Experienced outdoor survey teams can often make an accurate, informed guess by inspecting a pole for the presence of a photocell. This can be done from ground level in many cases. The figures below show poles with and without photocells.

Figure 117 Identifying pole power taps from the ground

For a pole to have constant power, it must have a photocell. The photocell acts as a switch that allows power to flow to the lamp when it is dark. If there is no photocell, then the pole must be remotely switched.

Photocells are quite large - nearly 4 inches across. They are quite easy to see from the ground. Figure 117 shows lamp heads with and without a photocell. On the right, you can clearly see the cylindrical shape of the photocell sticking up on the right. On the left, you can see a much smaller “bump”. This is the plastic cover that protects the socket from weather when a photocell is not installed.

Lamp Head - Remotely Switched Lamp Head - Constant Power



Reading Pole Tags

In order for a pole owner to review and approve your list of selections, they have to be able to identify each pole. This is done through the pole tag.

When you have chosen a pole, look for a numbered plate in the lower section of the pole, within 2 meters of ground level. It may be plastic or metal. As you go down a street, the numbers on the tags should increase or decrease. On very old poles, the entire pole may have been repainted many times causing the tag to almost disappear into the paint. There are an almost infinite variety of pole tags. A few examples are depicted below.

Figure 118 Pole identification tags and marks come in many forms - and are vital to obtaining attachment rights

Sometimes, strings of bank switched poles are identified by a single pole somewhere on the string. If the pole you really want does not have a tag, follow the AC power up or down the street until you find a pole on that circuit that does have a tag. Without the tag, the pole owner will not be able to include it in a lease agreement.

Measuring Pole Dimensions

One of the most easy-to-overlook tasks on an outdoor survey is to measure the dimensions of the poles to which equipment will be mounted. Many deployments have been interrupted when the



installation crew did not have the proper size mounts. Having insufficient length of cables - whether for power, ethernet or RF - is also an instant showstopper. Taking good measurements during the survey will ensure that expensive install crews do not have downtime for these reasons.

There are two methods to measure the diameter of a pole. As shown on the leftmost photo, one can hold up a measuring tape across the pole and simply estimate the diameter. For a more precise measurement, one can wrap the tape around the pole as shown in the righthand photo. Then divide the measured circumference by the value of pi, or 3.1415927.

Figure 119 Two methods to measure diameter of pole

Most poles taper with height. So if you are measuring at ground level, you will want to estimate how much less the diameter will be at the intended mounting height. A street pole that is 20cm (7.8 in) in diameter at chest height may be only half of that diameter at the top. However, a wooden style telephone pole will not taper as much. The most accurate measurements of course are made at the mounting height. If the taper is not considered, the install crew could easily bring mounting parts that are too big.

Pole height is also critical. Most people are not good at judging pole height from ground level with any accuracy. Poles are rarely marked as to their height. In a pinch, if you have two engineers in your survey team one can stand next to the pole and the other person can estimate how many multiples the pole appears to be. Aruba recommends carrying a high power laser distance finder that is rated to at least 40m to take an accurate fix from the ground.



Radio LOS Path PlanningAll wireless links require “radio line of sight” between the two antennas for optimum performance. The concept of radio LOS involves the area along a link through which the bulk of the radio signal power travels. This area is known as the first Fresnel Zone of the radio link. Ideally, no object (including the ground) must intrude within 60% of the first Fresnel Zone. Figure 120 illustrates the concept of a good radio LOS.

Figure 120 Visual and radio LOS

If obstacles are in the radio path, there may still be a radio link but the quality and strength of the signal will be affected. Calculating the maximum clearance from objects on a path is important because it directly affects decisions about AP and antenna placement and height. Clearing the Fresnel zone is especially critical for long-distance links, where the radio signal could easily be lost if not accounted for when planning a link budget.

When planning the radio path for a wireless bridge or mesh link, consider these factors: Avoid any partial line of sight between the antennas. Be cautious of trees or other foliage that may be near the path or may grow and obstruct the

path. Be sure there is enough clearance from buildings and that no building construction may

eventually block the path. For very long distance links, the curvature of the earth (20 cm per km) may need to be

considered in the calculation of relative heights, but this is done automatically in the Aruba 3D Outdoor RF Planner.

Check the topology of the land between the antennas using topographical maps, aerial photos, or even satellite image data. Google Earth features offer a visible link path with obstacles as a tool.

Avoid a path that may incur temporary blockage due to the movement of cars, trains, or aircraft.

Antenna Height

A reliable wireless bridge or mesh link usually is achieved best by mounting the antennas at each end high enough for a clear radio LOS between them. The minimum height required depends on the distance of the link, obstacles that may be in the path, topology of the terrain, and the curvature of the earth (for links over 3 miles).



For long-distance links, the AP may have to be mounted on masts or poles that are tall enough to attain the minimum required clearance. Use Table 11 to estimate the required minimum clearance above the ground or path obstruction for 5 GHz mesh links.

In Figure 121, a wireless bridge or mesh link is deployed to connect building A to building B, which is located three miles (4.8 km) away. Mid-way between the two buildings is a small tree-covered hill. From Table 11 it can be seen that for a 3-mile link, the clearance required at the mid-point is 5.3 m (17.4 ft). The tree tops on the hill are at an elevation of 17 m (56 ft), so the antennas at each end of the link need to be at least 22.3 m (73 ft.) high. Building A is six stories high, or 20 m (66 ft.), so a 2.3 m (7.5 ft.) mast or pole must be constructed on its roof to achieve the required antenna height. Building B is only three stories high, or 9 m (30 ft.), but is located at an elevation that is 12 m (39 ft.) higher than building A. To mount an antenna at the required height on building B, a mast or pole of 1.3 m (4.3 ft.) is needed.

Figure 121 Link budgets include Fresnel Zone clearance

Table 11 Maximum clearance of First Fresnel Zone by altitude

Total Link Distance Max Clearance for 60% of First Fresnel Zone at 5.8 GHz

Approximate Clearance for Earth Curvature

Total Clearance Required at Mid-point of Link

0.25 mile (0.402 km) 4.6 ft (1.4 m) 0.007 ft (0.002 m) 4.6 ft (1.4 m)

0.5 mile (0.805 km) 6.2 ft (1.9 m) 0.03 ft (0.010 m) 6.2 ft (1.9 m)

1 mile (1.6 km) 8.9 ft (2.7 m) 0.13 ft (0.04 m) 8.9 ft (2.7 m)

2 miles (3.2 km) 12.5 ft (3.8 m) 0.5 ft (0.15 m) 13.1 ft (4.0 m)



5 miles (8 km) 20 ft (6.1 m) 3.0 ft (0.9 m) 23 ft (7.0 m)

7 miles (11.3 km) 23.6 ft (7.2 m) 6.2 ft (1.9 m) 30 ft (9.1 m)

9 miles (14.5 km) 27 ft (8.2 m) 10.2 ft (3.1 m) 37 ft (11.3 m)

12 miles (19.3 km) 30.8 ft (9.4 m) 18.0 ft (5.5 m) 49 ft (14.9 m)




Surveys for Mesh Portal Mounting LocationsDevelop a list of potential mesh portal locations, with a preference for buildings that are owned by individuals or companies that are participating in the network operation, which may be the local school district, city government, public safety agencies, or port or transportation facility operators. Create this list to reduce the time needed for the civil permitting processes and increase the probability of obtaining the mounting assets with optimal LOS conditions for the network. Always start with the obvious beneficiaries of the network and move on to other possibilities only if necessary.

The process of examining sites usually involves walking and documenting the physical mesh portal mounting locations.

Visit each preferred portal location in the soft RF plan that was used in Google Earth and the Aruba Outdoor Planning tool to confirm the availability or constraints of mounting on the desired structure. Make sure each location can be powered and properly grounded and that outdoor Ethernet Category 5E cable to the wired network is available for mesh portals. Always be prepared to find alternate locations and carefully consider future use of some radios that may require point-to-point mesh backhaul to increase capacity in an area. These upgrades may be critical to an economically successful long-term network deployment.

Wired Backhaul Assessment

For locations with desirable locations, complete a backhaul survey with the following steps: Inspect the MDF on the property for space and power to support wired connectivity between

Mesh Portals and Aruba controllers Determine the cable path (risers plus horizontal runs) between the MDF and the IDF(s) nearest

the Mesh Points. Ensure that adequate conduit space exists, that difficulty/cost of running connectivity is within acceptable limits.

Determine the cable path (risers plus horizontal runs plus penetrations) between the Mesh Points and their nearest IDFs.

The mounting location for the AP is within 100 meters of an IDF, using an existing roof penetration; OR

The mounting location for the AP is within 100 meters feet of an installable all-weather enclosure with power that can house Aruba power injectors and Ethernet lightning arrestors, AND the enclosure location is within 100 meters of a network interconnect.

Determine how any necessary grounds will be accomplished.

Antenna Position and Orientation

After the required antenna height has been determined, other factors that affect the precise position of the wireless bridge or mesh link must be considered:

Be sure no other radio antennas are within 2 m (6 ft.) of the wireless antennas, including other Wi-Fi radio antennas.

Place the wireless bridge or mesh link away from power and telephone lines. Avoid placing the wireless bridge or mesh link too close to any metallic reflective surfaces, such

as roof-installed air-conditioning equipment, tinted windows, wire fences, or water pipes. Ensure that at least 5 feet clearance exists from such objects.



The wireless bridge or mesh link antennas at both ends of the link must be positioned with the same polarization direction, either horizontal or vertical. Proper alignment helps to maximize throughput. Remember that these antennas must be aligned vertically. It is very common to forget this requirement when installing mesh backhaul links.

Radio Interference

The avoidance of radio interference is an important part of wireless link planning. Interference is caused by other radio transmissions using the same or an adjacent channel frequency. First scan your proposed site using a spectrum analyzer to determine if any strong radio signals use the 802.11a/b/g/n channel frequencies. Then plan to use a channel frequency that is far away from any other signal.

If radio interference is still a problem with your wireless bridge or mesh link, change the antenna direction to see if the situation improves.

Weather Conditions

When planning wireless bridge or mesh links, consider any extreme weather conditions that are known to affect the location. Consider these factors:

The wireless bridge or mesh link is tested for normal operation in temperatures from -30°C to +55°C. Operating in temperatures outside of this range may cause the unit to fail.

The wireless bridge or mesh link can operate in winds up to 90 miles per hour and survive higher wind speeds up to 125 miles per hour. You must consider the known maximum wind velocity and direction at the site and be sure that any supporting structure, such as a pole, mast, or tower, is built to withstand this force.

To protect against lightning-induced surges, the AP requires lightning protection on the radio interface ports.

The wireless bridge or mesh link is weatherproofed against rain. However, it is recommended that weatherproof sealing tape be applied around the Ethernet port and antenna connectors for extra protection. If moisture enters a connector, it may cause degradation in performance or even a complete failure of the link. For long-distance links, plan 0.7 dB of additional margin per km to allow for RF losses that occur during periods of heavy rain or snowfall. See Chapter 12: Installation, Validation, and Optimization on page 159 for detailed instructions.

Snow and ice: Falling snow, like rain, has no significant effect on the radio signal. However, a buildup of snow or ice on antennas may cause the link to fail. In this case, clear the snow or ice from the antennas to restore operation of the link.

Ethernet Cabling

When a suitable antenna location has been determined, you must plan a cable route from the wireless bridge or AirMesh router to a suitable power source and, in the case of mesh portals, to an Ethernet network.

Consider these points: The Ethernet cable length should never be longer than 90 m (295 ft). Determine a building entry point for the cable (if applicable) and how to weatherize the building

ingress/egress.



Determine if conduits, bracing, or other structures are required for safety or protection of the cable.

For lightning protection at the power injector end of the cable, consider using a lightning arrestor immediately before the cable enters the building.

Grounding

Weatherproofing and grounding of the radio is critical to protect the sensitive electronics. The Aruba AP-175 and outdoor AirMesh routers all use industrialized housings that are designed to protect the radio from the weather. It is important that the wireless bridge or mesh link cables and any supporting structures are properly grounded. In general, every RF connector that is attached to an antenna should be protected with a lightning arrestor that is individually grounded. Each AP-175 or AirMesh router also includes a grounding screw for attaching a ground wire to the AP housing. Be sure that grounding is available and that it meets local and national electrical codes.

Figure 122 AP85 with lightning arrestors and chassis ground

Civils ApprovalsMajor Tasks:

Customer submits Mesh Point selections to utilities & manages approvals/rights process Customer submits Portal location selections to land/building owners and manages approval/

rights process Customer generates a list of all encroachment, minor-use, major-use, miscellaneous-use and

other permits required by all government agencies with jurisdiction inside the coverage zone.



Approvals: One or more city governments Utilities Landholders / Building owners PUC / Regulatory agencies

Final Network DesignWith the survey complete, the wireless architect now has all of the facts needed to adjust the soft RF plan as needed to produce a final network design. This may be further broken down into a high-level design and a low-level design. The high-level design contains macro elements such as how the overall RF and capacity injection topologies map to specific structures, uplink capacities, access layer cell sizes, client types supported and the like. The low-level design often consists of the site-by-site build plans for individual rooftops, or physical installation plans for each of the various types of poles to be used. If battery-assisted or solar-assisted power solutions are required, these may also be detailed in the low-level design.

Best Practices for Conducting Outdoor Surveys

Personal Safety & Security

Outdoor surveys are some of the most interesting and technically rewarding projects a wireless engineer can undertake. They also expose engineers to a variety of personal safety and security risks which should be thoroughly understood and prepared for before setting off into the field.

The most obvious risks have to do with driving, walking and climbing. While driving, the survey crew is studying the environment scouting for mounting assets and it is very easy to become distracted. Survey vehicles tend to drive more slowly and make unplanned turns and stops, which can also create unsafe situations. Survey crews regularly exit their vehicle to study sight lines, take photographs or GPS readings, and perform tests such as spectrum analysis or throughput measurements. In a metropolitan survey, these activities usually take place right next to an active roadway. In a railyard or manufacturing plant, they may occur next to active train tracks or moving cranes or other vehicles with limited visibility to people standing on the ground, or ability to stop suddenly. And of course climbing can involve walking on rooftops, getting up in bucket trucks or actually climbing poles to inspect sight lines and mounting locations.

Other risks are less obvious but happen more often than you might think. Here are a list of some of the situations for which you should be prepared:

Theft or mugging – Survey crews routinely carry valuable items including laptops, PDAs, cameras and GPS units. Because they usually travel and are unfamiliar with the survey area, it is easy to enter less safe parts of town without realizing it. A lone engineer is more of a target than a team. Always secure the survey vehicle when stopping for lunch or going inside a building to talk to property managers.

N O T E

Aruba strongly recommends that outdoor survey crews work in teams of at least two engineers.



Police Stops – Police officers and private security guards are trained to notice unusual activity in their patrol areas, and survey crews create a lot of unusual activity! Many experienced survey engineers have been stopped and questioned about what they were doing at various points in their careers.

Citizen Stops – Security personnel are not the only ones to stop you. It’s pretty common to have locals come up and ask what you’re doing. Looking official with a company shirt and ID badge makes a big difference here.

Photography – Experienced survey engineers try to be very aware of what direction they are shooting photos. If one is in a neighborhood that feels risky, it probably is. Be very careful not to aim at people engaging in activities that might become upset. The same rule applies to military bases or other high security facilities.

Safety and security are the responsibility of each survey team member that goes into the field. Here are some best practices to consider:

ALWAYS work in teams of at least two individuals. This allows one person to focus on driving or situational awareness, while the other person can focus on the survey itself. Should something occur, the “buddy” system can also ensure that timely assistance is forthcoming.

ALWAYS be aware of your environment and what is going on nearby. ALWAYS wear personal protective equipment (PPE) including at least an approved high-

visibility traffic safety vest and steel-toed boots or other appropriate footwear. Additional PPE such as hardhats and safety glasses may be required in specific facilities. Most hardware stores around the world carry a selection of safety glasses, hard hats, vests and other PPE.

NEVER climb any type of structure without successfully completing an appropriate safety class and ensuring adequate spotters

NEVER operate lifts, booms, cranes or other machinery without successfully completing an appropriate qualification and training class

It’s a standard procedure both inside the United States and especially in the rest of the world to notify the local authorities before heading out for survey work that involves driving or testing in public areas. It’s a good idea to have a business letter on official letterhead from the agency or enterprise that is sponsoring the survey with a contact number for security personnel to call for more information.

It’s also a good idea to look as “official” as possible – Wear a company shirt, an ID badge, and have extra business cards with you.

Never leave any hotel without your passport or driver’s license.

Building a Complete Outdoor Survey Kit

Knowing what to pack in advance before boarding a plane to go conduct an outdoor survey can mean the difference between success and failure. A complete outdoor survey kit enables a wireless engineer to squeeze every bit of productivity from each minute in the field. This means getting done faster, with more accurate data, and less stress trying to make the plane back home.

Aruba recognizes that the full kit as described here is an investment for many partners and customers. But for those engineers who expect to perform outdoor surveys even just a few times per year, the investment is easily recoverable in productivity and reduced travel costs from quicker onsite trips.



Road time is some of the most expensive time on a per-hour basis, especially when travel to other countries is considered.

The basic field survey kit that Aruba outdoor engineers carry consists of the following: Soft RF Plan: Never go into the field without having completed a soft RF plan! One of the fastest

ways to waste time in the field is to show up without a plan and try to design the network on the fly. The few hours needed to do a thorough soft plan can literally shave days or weeks off the time required to build a network up link by link on the ground.

Laptop with GoogleEarth Pro and GPS-enabled 3G/4G card: Chances are the soft RF plan will be completed in Google Earth (http://earth.google.com). The Pro version includes support for realtime GPS tracking. When combined with a 3G/4G modem, this allows you to track your exact position in realtime. This allows the survey team to reach each node on the soft plan very quickly and accurately. Without this tool, it is incredibly easy to waste time driving in unfamiliar areas trying to find the exact spot on the paper map. This tool alone can cut survey time by as much as half. Many 3G/4G modems include a built in GPS that can be read by Google Earth on a COM port. If yours does not, you can use an external handheld GPS with a USB cable.

GPS-enabled Camera with 10X Optical Zoom: A good camera has always been a basic requirement of any survey kit, indoor or outdoor. Aruba strongly recommends a 10X optical zoom for outdoor surveys, which allows the engineer on the ground to read labels high in the air on pole lamps and radio equipment from as far as 40 meters (130 ft) below. Many inexpensive GPS-enabled cameras were introduced recently. The GPS feature saves enormous amounts of engineer time. Here’s why. Without GPS tagging, the engineer has to spend many hours manually sorting and filing all of the photographs taken on a per-pole or per-node basis so they can be easily recovered later for a report. With GPS tagging, there is no need to do any sorting at all. When using a GPS-aware photo application such as Google Picasa (http://picasa.google.com) on Windows or iPhoto 2011 on the Mac (http://www.apple.com/ilife/iphoto/), you can simply browse the photos by location. All pictures taken at the same street corner will be grouped together, making photo recovery a snap. Not only that, but engineers who were not even present on the survey can quickly get a sense of the network layout.

Handheld GPS with USB Port: Buy the cheapest handheld GPS you can find that includes a USB port. Use the GPS to capture waypoints of potential radio locations and other structures of interest. The GPS can also be used while driving for realtime maps with Google Earth. And it is necessary to perform drive testing with AirMagnet Survey Pro or Ekahau Survey Pro. You’ll also find the “tracks” function useful as it records your entire drive path which can very useful weeks after you get back to jog memories. Many smartphones also include basic GPS functionality like capturing waypoints. However, they are nowhere near as accurate as a dedicated GPS, and they burn up the battery very fast.

GPSGate Software: While GoogleEarth can directly communicate with a GPS over the USB port, other applications such as AirMagnet Survey Pro only know how to read from COM ports. GPSGate is a USD $40 software utility that provides a realtime USB-to-COM port software bridge (http://www.gpsgate.com/).

N O T E

Aruba strongly recommends that outdoor survey crews work in teams of at least two engineers.


http://earth.google.com

http://picasa.google.com

http://www.apple.com/ilife/iphoto/

http://www.apple.com/ilife/iphoto/

http://www.gpsgate.com/


High-power Laser Distance Meter: Most people are bad at estimating heights of poles and structures from the ground. Very bad. Since accurate heights are critical for modeling in the Aruba 3D RF Outdoor Planner, why guess? A high power model good to 40m (130 ft) will give you precise information and typically costs less than USD $100.

Mobile Spectrum Analyzer: You cannot use a given AP location if there is too much interference. A basic task at each proposed radio site is to perform a few minutes of spectrum clearing, and record it for subsequent report creation. There are multiple options, from USB cards that connect to laptops, to custom cards that plug into an iPad, to dedicated handheld units. Aruba currently prefers AirMagnet SpectrumXT which costs a little more but has advanced time-saving features and good accuracy. We also happily recommend the Wi-Spy series of USB analyzers from MetaGeek which start at as little as USD $199 as of this writing (http://www.metageek.net/).

GPS-Enabled Site Survey Software: For post-installation drive testing to prove coverage, it is absolutely necessary to have a professional site survey software tool that is GPS-enabled. The two principal choices as of the publication date are AirMagnet Survey Pro (http://www.airmagnet.com/) and Ekahau Site Survey Pro (http://www.ekahau.com). Expect to pay a couple of thousand US dollars. In the past, outdoor engineers were known to hand draw coverage zones by taking periodic readings with a laptop NIC client adapter utility. Such methods are not only highly inaccurate, but also very incomplete and look unprofessional. Aruba uses AirMagnet due to its integration with Google Earth, which allows the wireless engineer to superimpose models from our 3D RF Outdoor Planner on top of the drive test results to quickly prove to the customer that we met their expectations.

Magmount Antennas with RF Extension Cables: Readings taken inside a vehicle are not accurate due to absorption, reflection and scattering from the metal and glass on the vehicle. Signals inside may be anywhere from 5dB to as much as 20dB off. When drive testing, it’s vital to have antennas mounted outside the vehicle. Any major online vendor of RF products will carry magmount dual-band antennas covering both 2.4GHz and 5GHz. Make sure to get at least 2 meters of cable length, and we recommend having an RP-SMA connector on the end. You should also buy a short pigtail to adapt the RP-SMA connector to an MMCX connector, which is used on WLAN NICs and USB spectrum analyzers that support external antennas. Finally, don’t forget to buy as many antennas as you have radio chains in your test equipment!

Handheld Radios (or “walkie talkies”): Aruba recommends that each member of the survey team have a handheld radio for low-cost and fast communication while outside the vehicle. When performing active testing with an engineer at each end of a link, these come in very handy. When surveying in another country, it’s much cheaper than burning minutes on a SIM card.

Power: Power management is a critical survey skill. Nothing wrecks a survey faster than running out of juice. Aruba engineers carry multiple power sources and backups to deal with almost any contingency. These include: 400Watt AC inverter and power strip. This allows direct charging of equipment from the

vehicle alternator using standard AC power cables. It’s a good idea to buy an inverter that includes separate leads for a cigarette lighter and alligator clips for direct battery attachment.

Laptop Cigarette lighter attachment. Many laptop makers and third parties offer DC adapters that power a laptop directly from a cigarette lighter. These are quite useful.


(http://www.airmagnet.com/)

http://www.ekahau.com

http://www.metageek.net/


Spare laptop batteries. In case of a power issue or inverter failure, Aruba engineers carry multiple high-capacity laptop batteries to keep things going. This is also handy if multiple laptops are in use and there aren’t enough slots in the power strip. Be sure to recharge every night in the hotel.

Spare camera battery and charger. Rare is the survey that does not exhaust a camera battery midway through the day. We have noticed that the GPS-enabled cameras also burn more battery than usual to maintain a GPS position fix. Be sure to recharge every night.

Lots of AA and AAA batteries. The handheld GPS, walkie talkies, laser distance meter and other devices use standard disposable batteries. Have plenty of spares of each type you require.

Personal Protective Equipment: Approved traffic safety vest, appropriate footwear, and any other required safety items for the environment in which you will be working.

All of the above items are required for basic predeployment visual survey work and post-deployment coverage validation testing. However, if you expect to perform active RF testing using Aruba APs and antennas then you will need additional equipment.

Mobile Crane or Towers: Outdoor surveys are useless unless the test AP is mounted at roughly the actual height at which it will be installed. Homebrew towers made of PVC pipe are inherently unstable above just a few meters which is not nearly enough. Aruba strongly recommends mobile masts from Blue Sky Mast (http://www.blueskymast.com/), which are lightweight, man-portable units originally designed for military communications. They are available in heights from 5m to 15m, with a rich assortment of accessories for rigging APs and antennas. Alternatively, Aruba engineers have rented towable booms or lifts. These have the advantage of allowing an engineer to inspect the exact sight line from the perspective of the antenna. However, they require proper safety flagging on the ground and can require a trained operator.

Access Points: Generally you will pack two APs for mesh testing, to evaluate link speed at various ranges or environmental conditions. This also gives you a spare in case something happens to one of the units.

Antennas: Pack a full set of antennas to adequately measure the expected use cases from the soft RF plan. It’s a good idea to throw in some extra antennas in case your testing shows that an alternate model would be a better fit.

Throughput Testing Software Utility: With MIMO systems, SNR and data rate alone do not paint a complete picture of link performance. Aruba recommends pushing traffic across a test link using a test application. The freeware utility iPerf is widely used and quite useful, or if you have access to a professional tool such as Ixia IxChariot (http://www.ixia.com) or VeriWave WaveDeploy (http://www.veriwave.com) this is even better.

Cables: Be sure to include plenty of each major type of cable. If you are mounting the AP at tower height, be sure that your ethernet cables can reach the ground with length to spare to reach the power injector. Make sure to have RF cables to connect antennas. If your AP is at ground level, make sure you have long enough RF cables to reach the antenna height for each RF connector on the AP. And don’t forget to pack your console cable plus a spare!


http://www.blueskymast.com/

http://www.veriwave.com

http://www.ixia.com


Power: All this additional equipment needs to be powered. The AP power source must be stationary at the tower location, so it is separate from the vehicle power listed earlier. You will need the following for each end of the link: 802.3at power injector. Injectors allow you to both power the AP as well as connect to it from

the ground to do configuration and gather statistics. Be sure to have the proper injector for your AP. The MSR4000 requires 4-pair, 60W 802.3at while the MSR2000 uses 2-pair 30W 802.3at.

Fully charted 12V deep cycle battery. Aruba recommends a standard automotive or marine battery in a 100 amp-hour size. 100 Ah is generally enough to get in a full day of testing with extra equipment plugged in.

12V battery charger. Keep this in the hotel room to recharge the deep cycle batteries at night. You’ll need one per battery.

Tools: A field toolkit with all the tools needed to assemble mounting brackets, hang APs, fix antennas, and the like needs to be included.



Chapter 11: IP Planning for Aruba AirMesh

A well-designed IP network is critical to building and managing many networking devices, including each AirMesh router. Keep accurate records of geographic installation details to aid in troubleshooting network problems and maintaining network service-level agreements. IP network planning identifies IP network infrastructure components and addressing schemes that must be considered and planned.

Configure a Router IDThe router ID is a unique 32-bit IP address that represents the name of the router, which is used often in AWR routing.

Mesh Backhaul LinksA typical backhaul link in Aruba mesh networks can be thought of as a logical point-to-point connection. To avoid unnecessary issues, Aruba highly recommends that IP subnets that use /30 bit IP subnet masks be used to allow for a large number of IP networks, which may occur as the network grows over time. Mesh networking by its very nature implies that many IP networks are reachable, but AWR makes sure the entire path, not just the next hop router, is the highest performing route. Wireless signals often take paths that are completely unpredictable to humans, but intelligent link analysis and network routing keeps up with these changes in real time.

Access Links and Client Devices Access links are basically Wi-Fi access or mesh backhaul interfaces that are used on every mesh portal or mesh point. Onboard DHCP services can be configured manually or allowed to assign IP addresses automatically to infrastructure routers and Wi-Fi clients. Manual DHCP assignment for mesh routers is recommended and requires that the MAC address of the router be recorded and mapped to an IP address in the DHCP server, just as a static user assignment would be. Static addressing of mesh routers ensures that each device can be referred to reliably by its IP address for monitoring and debug, even though the topology may change constantly.

Wired Network Ethernet Link ParametersIn most cases, mesh portal connections to the wired network are automatic, perhaps with an IP address reserved in a central DHCP server. Occasionally, operators may configure the Ethernet interface speed statically, but by default it is auto-sensing, so no configuration is required.

IP Addressing and Networking Large outdoor network plans must include detailed IP addressing and routing information, which may support fixed and mobile wireless clients and devices. Network redundancy and the need to continually add capacity with more mesh portals usually drives the IP network design, making sure that expansion can keep pace with wireless needs.

Aruba Networks, Inc. IP Planning for Aruba AirMesh | 157


As applications are identified and bandwidth estimates acquired, a prioritization scheme should be agreed upon where certain applications like VoIP or a certain group of users such as first responders, are prioritized over other applications or users when the network is busy.

This list summarizes the IP addressing considerations: ESSID to IP network mapping and VLAN identifiers VLAN trunking requirements across mesh backhaul links Wireless encryption for each ESSID and back-office interfaces like RADIUS servers Infrastructure redundancy for wired and wireless infrastructure Infrastructure IP addressing for mesh portal and mesh point router IDs Infrastructure IP addressing for mesh links and clients Requirements for NAT with consideration for legacy and emerging applications DHCP servers, scope assignment, and lease times for each network DHCP static IP assignments for infrastructure devices

Aruba Networks, Inc. IP Planning for Aruba AirMesh | 158


Chapter 12: Installation, Validation, and Optimization

Typically, installers preconfigure each device to minimize the time to physically mount each radio. At each site, avoid other cables and follow local building codes for connecting to power. Aruba outdoor radios come with flexible mounting brackets, suitable for use on many common mounting structures.

Lightning arrestors should always be ordered and installed with outdoor radio equipment. It is critical that the mounting structure and each radio in particular is well grounded by a licensed electrician, following the recommendations in each installation guide.

MeshConfigAruba MeshConfig is an easy-to-use web-based tool that simplifies the deployment, configuration and administration of Aruba AirMesh wireless mesh routers. From any PC running Windows 7 or XP, IT organizations can view the AirMesh topology, monitor clients, manage faults, view historical reports, and distribute software images to all or specific AirMesh routers.

MeshConfig provides comprehensive historical reporting on client, mesh link and device health. MeshConfig can monitor and manage multiple AirMesh networks containing up to 250 wireless mesh routers. AirMesh routers can be automatically discovered from a range of IP addresses or they can be added manually. As the mesh network inventory populates, MeshConfig identifies the properties of AirMesh routers, including router name, IP address, image version, status and alarm conditions. Network-wide settings can be configured within MeshConfig and propagated to all AirMesh routers in the network. Individual device settings are easily accessible from the web interface.

The MeshConfig topology mapping capability shows the links between AirMesh routers as well as link status, link quality and real-time performance statistics for each link. Individual links can be disabled, enabled or designated as a preferred connection. Router-specific parameters – including radio, channel and security settings – can be configured through the MeshConfig web interface. Once committed, the progress of router configuration changes can be tracked in real time.

Aruba Networks, Inc. Installation, Validation, and Optimization | 159


The browser-based MeshConfig tool lets IT upload topography, Google Earth and other digital background maps to view and monitor mesh deployments across geographical areas.

Figure 123 Realtime display of mesh topology and status in Aruba MeshConfig

MeshConfig keeps track of the AirMesh router inventory by serial number, model and firmware version.

Figure 124 Mesh node inventory report in Aruba MeshConfig



Staffing Expectations Table 12 lists the resources that typically are involved in a mesh network deployment. The table should help you understand the steps to building a successful mesh network. Whether you are building a small video surveillance network or a wide area network that covers many square miles, someone needs to consider all aspects of the network deployment.

One way to think about the planning and deployment of a very large mesh network is to recognize that, fundamentally, a large mesh network is many small mesh clusters organized into seamless coverage areas that use a common authentication mechanism.

Outdoor networks are very complex and have many variables that require flexibility that typically is not encountered when designing indoor wireless networks. The following steps are commonly addressed during the planning and installation process, and they are assigned to one or more of the deployment team members:

Application information collection Data modeling of applications Network load calculations Coverage calculation for preliminary sites Capacity calculation for preliminary sites Map-based provisioning Final site information verification, based on site surveys Final coverage calculation Final capacity calculation

Table 12 Resources and their typical responsibilities

Resources Responsibility

Operator Management Services and coverage roll-out Relationships: permits, mounting, services, customers

Application Engineering Manager RF and IP network design: installation validation Systems specifications and product requirements

Project Manager Budget, planning, and control Customer engineering interface with partners

Radio Design Engineer Capacity design and coverage simulation Site installation guidance IP data network design Site configuration data capture and monitoring

Network Design Engineer IP back office interfacing, billing, and overlay services NMS infrastructure and integration with data design

Site Installation Engineer Site acquisition and installation Permitting, inspections, and compliance

Site configuration Engineer and Commissioning Engineer

Site data configuration Function verification testing System capability verification



Final design, dependent on assets Mesh point site design and equipment configuration Mesh portal site equipment configuration IP network design and equipment configuration Hardware pre-configuration and preparation NMS node configuration and installation Functionality verification tests System coverage verification tests System throughput verification tests Final “as-built” system design documentation

Aruba Outdoor AP Antenna Weatherproofing

Installing Antennas

1. Before connecting the antennas, identify which of your antennas are 2.4 GHz and which are 5 GHz.

2. After identifying which antennas will go where, install them by placing the antenna connector over the corresponding connector and the AP and turning the connector clockwise until hand tight. Repeat this process for each antenna.

3. Place the included metal weatherproof caps over any unused antenna interfaces by turning them clockwise until hand tight.

Weatherproofing Connections

Weatherproofing your antenna and/or cable connections on your outdoor AP is essential to reliability and longevity of your product. This process prevents water from entering the AP or antennas through the connectors.

A good weatherproofing job consists of three wrappings: 1. electrical tape 2. butyl rubber3. electrical tape

The first wrapping of tape should be at least two layers, followed by a single wrap of butyl rubber, and four-layer wrap of electrical tape. This provides good protection from water, heat, and other potential hazards that could damage your AP or antennas.

Additionally, wrap your connections such that water is always directed down and away from connections.



Required Items and Tools

3/4” (19 mm) Vinyl Electrical Tape Butyl Rubber Tape Knife or Box Cutter

Types of Connections

The following sections provide guidance on weatherproofing directly connected antennas (Figure 125) and cable connections (Figure 126). The same materials are needed for weatherproofing both types of connections but the procedure is slightly different. For weatherproofing directly connected antennas, see Weatherproofing Directly Connected Antennas on page 165. For weatherproofing cable connections, see Weatherproofing Cable Connections on page 168.

Figure 125 Directly Connected Antennas

AP

175_

11

Weep holes



Figure 126 Cable Connections

Important Points to Remember

Do not cover the weep holes on the antennas. Doing so can restrict the release of condensation from the antennas.

Proper weatherproofing is not a fast process. Set aside ample time to complete the steps outlined below.

When wrapping, make the each layer of tape as flat as possible. Wrinkles and folds in the tape create places for water and moisture to gather.

AP

175_

16

Connectors on bottom of antenna

N-type connectoron an RF cable

N-typeconnectoron a pigtail



Weatherproofing Directly Connected Antennas

First Wrapping of Tape

1. Before wrapping the antennas, locate the weep holes (Figure 125). Weep holes allow condensation that has built up inside the antenna to escape.

2. Prepare the antenna connector by cleaning and drying it.3. Cut a 4” (100 mm) strip of electrical tape from the roll. Pre-cutting the tape into strips makes in

easier to maneuver the tape around the antennas and other components of the AP’s case.4. Beginning just below the weep holes, tightly wrap the connection with a layer of the 3/4” (19mm)

electrical tape. Overlap the tape to a half-width. 5. Repeat steps 3 and 4 until the wrapping extends all the way to the AP’s case.

Figure 127 First Wrapping of Tape (AP-175 shown)

Wrap tape fromjust above knurledsection to base of

antenna mount

Pieces of tape as needed

Leaveweep holesuncovered

AP

175_

12



Wrapping of Butyl Rubber

1. Cut a 3/4” (19 mm) strip of butyl rubber.2. Wrap the strip of rubber around the taped connector (Figure 128)3. Join the two ends by pushing them together until there is no longer a seam (Figure 129).

Figure 128 Butyl Rubber Placement (AP-175 shown)

Figure 129 Butyl Rubber Wrap

Cut 3/4” stripof rubber

Squeeze thinner& wider Wrap rubber

around baseof antenna

mount

AP

175_

13

Wrap rubberaround baseof antenna

mount

Squeeze tobond rubber

to itself

Rubber willbe wrappedwith 4 layers

of tape

AP

175_

14



Second Wrapping of Tape

1. Cut a 4” (100 mm) strip of electrical tape from the roll.2. Where you begin wrapping depends on the orientation of the antenna. Water should flow in the

opposite direction of the wrapping to prevent water from entering the connector between the layers of tape.Therefore, if the antenna is facing up, you should begin wrapping at the AP end of the connector. This will ensure that your fourth and final layer will be layered correctly. Conversely, if your antenna is facing down, you should begin wrapping on the antenna end of the connector.

3. After completing the fourth layer of tape, check your work to ensure there are no places where water can collect. If there are, you must smooth out those areas with additional layers of tape or remove the weatherproofing and begin again.

Figure 130 Completed Wrapping (Antenna on Top of AP, AP-175 shown)

4. Repeat this process for all connectors.

First and third layers wraptop to bottom A

P17

5_15


of tape

Second and final layers wrapbottom to top




Weatherproofing Cable Connections

First Wrapping of Tape

1. Prepare the antenna connector by cleaning and drying it.2. Cut a 4” (100 mm) strip of electrical tape from the roll. Pre-cutting the tape into strips makes in

easier to maneuver the tape around the connectors and other components but is not required.3. Beginning at the top of the connector, tightly wrap the connection with a layer of the 3/4” (19mm)

electrical tape. Overlap the tape to a half-width. 4. Repeat steps 3 and 4 until the wrapping extends all the way to the cable’s insulation.

Figure 131 First Wrapping of Tape

Wrap tapefrom antennaconnector baseto cable


AP

175_

17



Wrapping of Butyl Rubber

1. Cut a piece of butyl rubber large enough to wrap around the connector and extended past the first layer of tape.

2. Wrap the strip of rubber around the taped connector (Figure 132)3. Join the two ends by pushing them together until there is no longer a seam (Figure 133).

Figure 132 Butyl Rubber Placement

Figure 133 Butyl Rubber Wrap

Stretch thinner& wider

AP

175_

18

Wrap rubberaround connector

and cable

Squeeze tobond rubber

to itself


of tape

AP

175_

19

Wrap rubberaround connector

and cable



Second Wrapping of Tape

1. Cut a 4” (100 mm) strip of electrical tape from the roll.2. Using 3/4” (19mm) electrical tape, begin wrapping at the connector and create four layers.3. After completing the fourth layer of tape, check your work to ensure there are no places where

water can collect. If there are, you must smooth out those areas with additional layers of tape or remove the weatherproofing and begin again.

Figure 134 Completed Wrapping

4. Repeat this process for all connectors.

First and third layers wraptop to bottom A

P17

5_20


Rubber will be wrappedwith 4 layers of tape

Second and final layers wrapbottom to top



RF Coverage ValidationIt is common for the network operator and installer to work together to install instrumentation in the network to gather usage data on a routine basis. Instrumentation is installed very early in the deployment so that the results of this data gathering can be used to enhance the build-out by providing near real-time feedback.

Network verification tests include a review of the installation documentation and the expected coverage and capacity in the final system design. Third-party tools such as Air Magnet or Ekahau are then used to perform drive tests in the coverage area and capture heat maps that display signal strength for the coverage area.

During these drive tests, it is also customary to pause periodically and measure the uplink and downlink data rates. Simple tests are performed using utilities like IPerf or FTP file transfers to document the available throughput.

In most networks, validation becomes iterative; as areas are examined, it is common to fill in coverage gaps using reserved mesh points. If throughput bottlenecks are identified, it may be necessary to reduce hop counts in large areas by converting a mesh node to a mesh portal and provide additional wired network access.

Figure 135 Sample drive test results



Reconciling Drive Test Data with Predictive Models

When AirMagnet Survey Professional and the Aruba 3D Outdoor RF Planner are used, it becomes possible to create a “closed loop” outdoor engineering process. As we have seen, the 3D Planner produces three-dimensional RF coverage models that can be viewed in Google Earth.

As of this writing, AirMagnet Survey features a Google Earth export capability where heat maps that are captured using real-time GPS drive-testing can be overlaid on the predictive model. These enhanced heat maps provide the wireless architect with a sophisticated tool to demonstrate to the customer that the predicted coverage has been delivered.

Figure 136 Comparing 3D Outdoor Planner prediction with drive test results

Mesh Network OptimizationAfter a large mesh network has been deployed and is operational, it is important to monitor and analyze the network for optimal performance. Network operators may change several variables as the number of users increases or as new applications and load are added. These parameters are just some that are available to tune the mesh network:

Tune the backhaul channels Tune the access channels Adjust the antenna directions Adjust the antenna AP power output Add mesh portals to increase capacity Add mesh points to increase coverage



Appendix A: Allowed Wi-Fi Channels

This appendix is a reference of Wi-Fi frequency allocations and channel assignments. Notice that there are significant variances based on the country of operation. This information is valid as of the date of publication.

2.4 GHz BandThe IEEE 802.11 standards define 14 channels. Each channel is 22 MHz wide, but the channel separation is only 5 MHz, which leads to channel overlap that allows signals from neighboring channels to interfere with each other. In a 14-channel system (11 usable channels in the US), only three nonoverlapping 25 Mhz channels (and thus, noninterfering) are possible: 1, 6, and 11.

Figure 137 IEEE 802.11 2.4 GHz ISM channel allocations

4.9 GHz BandThe U. S. Federal Communications Commission allocated 50MHz of spectrum from 4940MHz to 4990MHz for public safety radio services in 2003. Any state or local government entity that provides public safety services - defined as being focused on the protection of life, health or property safety - are eligible. Eligible public safety agencies may apply for a license to utilize this band for the purpose of deploying wireless networks for police, fire, medical and similar user communities. In the US, there are two non-overlapping 20MHz channels. Various channelizations are possible, with the most common being channels 22 and 26.

802.11j-2004 or 802.11j is an amendment to the IEEE 802.11 standard designed specially for Japanese market. It allows wireless LAN operation in the 4.9 to 5 GHz band to conform to the Japanese rules for radio operation for indoor, outdoor and mobile applications. The amendment has been incorporated into the published IEEE 802.11-2007 standard. In Japan, there are four non-overlapping 20MHz channels. HT40 operation is also permitted.

Aruba Networks, Inc. Allowed Wi-Fi Channels | 173


The 4.9GHz band is supported on the AirMesh product family from Aruba. ArubaOS does not support 4.9GHz band operation.

Figure 138 4.9 GHz band plan for US and Japan regulatory domains

5 GHz BandThe 5 GHz band(s) allow many more nonoverlapping channels than the 2.4 GHz band. In the United States before 2007, the UNII-I, -II and –III bands allowed the use of a total of thirteen 20 MHz channels (or six 40 MHz channels), but the number of available 5 GHz channels varies significantly from country to country. Figure 139 shows the number of 20 MHz channels and 40 MHz channel pairs available for use in the 5 GHz band.

Figure 139 5 GHz nonoverlapping channels

arun

_055

9

20 MHz channels(US)

20 MHz & 40 MHzchannels (Japan)

Frequency

Japanbandedge

FCCbandedge

Bandedge

20

184+/188- 192+/196-

196192188184

2421 25

22 2623

4910 4915 4920 4920 4930 4935 4940 4950 4955 4960 4965 4970 4975 4980 4985 49904945

Channels defined for 5 GHz band (US regulations), showing common 20 MHz channel plan and 40 MHz o ptions

Channel

Frequency (MHz)

US UNII I and UNII II bandsUNII I: 5150-5250 MHzUNII I I: 5250-5350 MHz8x 20 MHz channels4x 40 MHz channelsUNII I I requires DFS

149 161157153 BandEdge

Channel

Frequency (MHz) 5745 5765 5785 5805 5850

BandEdge5725

US UNII I II / ISM band5725-5850 MHz4x 20 MHz channels2x 40 MHz channels

Channel

Frequency (MHz)

US intermediate band (UNII I I extended)5450-5725 MHz11x 20 MHz channels5x 40 MHz channelsRequires DFS

36 4844 5240 56 6460 BandEdge

5180 5200 5220 5240 5260 5280 5300 5320 5350

BandEdge5150

100 112108 116104 120 128124

5500 5520 5540 5560 5580 5600 5620 5640

BandEdge5450

136 140 BandEdge

5680 5700 5725

132

5660

US Intermediate Band(UNII-II Extended)

5470-5725 MHz11x20 MHz channels5x40 MHz channelsRequires DFS

US UNII-I and UNII-II BandsUNII-I: 5150-5250 MHzUNII-II: 5250-5350 MHz8x20 MHz channels4x40 MHz channelsUNII-II requires DFS

US UNI-III / ISM Band5725-5850 MHz4x20 MHz channels2x40 MHz channels

Channels defined for 5-GHz Band (US Regulations), Showing Common 20-Mhz Channel Plan and 40-Mhz Options

165



In 2007 the radio regulatory bodies in many countries allowed the use of the “UNII-II extended” band from 5450 MHz to 5725 MHz as long as UNII-II equipment was capable of dynamic frequency selection (DFS). DFS requires that the AP monitor all RF channels for the presence of radar pulses and switch to a different channel if a radar system is located. Wi-Fi equipment that is DFS-certified can use the extended band, which adds up to another eleven 20 MHz channels or five 40 MHz channels (depending on the radio regulatory rules in each country). See Table 13 for additional frequency bands and channels for other regulatory domains.

Table 13 Additional frequency bands and channels for other regulatory domains

Channel # Frequency (MHz)

USA Europe Japan Singapore China Israel Korea Brazil

36 5180 Yes Yes Yes Yes No Yes Yes Yes




DFS

Cha

nnel

s





100 5500 Yes Yes Yes No No No Yes Yes





120 5600 No No Yes No No No Yes No



132 5660 No No Yes No No No No No

136 5680 Yes Yes Yes No No No No Yes

140 5700 Yes Yes Yes No No No No Yes

149 5745 Yes No No Yes Yes No Yes Yes





Total without DFS 9 4 4 9 5 4 9 9

Total with DFS 20 15 19 13 5 8 21 20



Appendix B: DFS Operation

With a total of about twenty 20-MHz channels (different countries support slightly different numbers) the 5 GHz band with DFS now has sufficient channels to implement most outdoor MIMO mesh backhaul links. However, DFS comes with a cost that could adversely affect outdoor mesh performance with DFS enabled. If the network is operating in proximity to radar sources in the 5450- to 5725-MHz band, operation of the mesh could be disrupted as required by the terms of the DFS grant.

The next section provides an overview of how DFS works and what you can expect when radar events occur.

Behavior of 5 GHz DFS Radios in the Presence of RadarActual radar events can be extremely disruptive to a mesh network that attempts to use DFS channels. To better understand what this means, we will review the constraints on APs and clients on these channels.

The rules for DFS are different for different countries or regions. The definitions of what kind of signals should be considered radar signals and how each signal is classified also vary (per region and over time). But, at a high level, the following steps provide a generic description of DFS for WLAN:

1. Before initiating any transmission on a DFS channel, the device (can be AP or client) monitors the channel for the presence of radar signals for the channel availability check (CAC) time. In most cases, the CAC time equals a minimum of 60 seconds, but it is increased to a minimum of 10 minutes for channels in the 5,600- to 5,650-MHz subband in Europe (channels 120, 124, 128, 116+, 120-, 124+, and 128-).

2. If any radar signal is detected, the device “blacklists” the channel and selects a different channel. If that channel is also a DFS channel, the process in step 1 is repeated. If a non-DFS channel is selected, this process no longer applies. Any blacklisted channels are considered unavailable for a minimum of 30 minutes (non occupancy period).

3. If no radar signals are detected during the CAC time, the device can start using the channel.4. While using the channel, the device that “owns” the connection (typically the AP) continuously

monitors the channel for radar signals (in-service monitoring). If a radar signal is detected, the AP issues commands to all clients to instruct them to stop any transmissions on the channel, and selects a new available channel. After detection, the AP needs to clear the channel within 10 seconds.

As you can see, APs on DFS channels take longer to come up and users on DFS channels can potentially experience lengthy service interruptions from radar events. Because radar frequencies do not align with 802.11 channels, such events can impact multiple Wi-Fi channels simultaneously. Therefore, the wireless designer is strongly encouraged to conduct a DFS survey during the HD WLAN planning process to validate the availability of these channels. Just because no airport is nearby does not mean that no radar is nearby. Other common sources of radar include marine shipping traffic, military installations, and doppler weather systems at local television stations.

Aruba Networks, Inc. DFS Operation | 177


A DFS survey is relatively simple to perform, and it requires an Aruba controller and AP. A DFS survey has these basic steps:

1. Install the controller with ARM scanning disabled. If the controller is not in the location where the survey will take place, arrange for wide-area connectivity to the AP.

2. Provision the AP to operate on channel 52.3. Allow the AP to dwell on that channel for four hours.4. If a radar event has occurred, it can be noted from the system log, and you will notice that the AP

will be on another channel.5. Repeat steps 2 and 3 on the next highest 20-MHz channel until channel 140 has been

completed.

Unfortunately, radar pulses cannot be detected with any PC-based portable spectrum analyzers on the market as of this writing. The cost of renting and operating a laboratory-quality spectrum analyzer is typically much higher than simply using the Aruba equipment that you intend to deploy.

Aruba Networks, Inc. DFS Operation | 178


Appendix C: Campus Extension Example

This appendix describes a University campus that wants coverage in the common areas by extending their controller-based indoor network.

It is common for customers who operate large campus style facilities to consider migrating from a first-generation wireless network with partial facility coverage to an enterprise-grade, full-coverage, secure system. Such systems have the capacity and redundancy to support mission-critical high bandwidth applications, employee or student broadband connectivity anywhere on the campus, Voice-over-WiFi and Real-Time Location Services (RTLS). Campus extension networks must also supply secure data access to a variety of different roles within the organization, each with its own security attributes. Finally, it may provide courtesy Internet access to in a secure and dependable manner.

For any campus wireless network project, Aruba recommends using a qualified services partner to complete the extensive diligence that is required including campus inspections, detailed analysis of an RF surveys, 3D RF modeling using Aruba proprietary tools, and technical peer review.

The high-level requirements for any campus are to cover areas in between major buildings, with a dense deployment of three to four APs in range for most users and two APs in less busy areas. We use 10 db as the design margin, which is conservative but recommended due to the trees and other obstructions common in this type of facility. For very heavily wooded campuses, you may wish to consider keeping the typical mounting heights below 4 meters, to avoid the dense tree canopy for client access connections. Don't forget about parking lots, loading docks, maintenance bays, and other areas of interest to the customer.

The ANT-2x2-D805 is often recommended for campus deployments, which is a 120 degree sectored antenna. It can be placed on the external overhang of the campus buildings (see Figure 141).

Figure 140 ANT-2x2-D805 is recommended for campus deployments with close-in coverage requirements

To limit cost, Aruba generally prefers to locate APs indoors, and drill a core hole outside where an antenna will be mounted directly. This produces better coverage by eliminating building shadowing, reduces installation labor and materials. Like security cameras, when mounted directly to the building within 24” of a core hole, this approach eliminates the need for lightning protection on these APs by

Vertical Horizontal

AP-ANT-2x2-20055 dBi

Vert. Beamwidth: 30°

Aruba Networks, Inc. Campus Extension Example | 179


sharing the cone of protection of the structure itself. Mounting APs on a roof with non-penetrating mounts is unnecessary, expensive, and will reduce coverage due to building shadowing.

The building mounting strategy proposed by Aruba has excellent aesthetics, is consistent with building codes, and has provided superior RF coverage in other similar campus installations. Here are photographs of the mounting strategy when mounted to the top of a building wall using the Aruba ANT-91 antenna (the non-MIMO predecessor to the ANT-2x2-D805).

Figure 141 Side mounting to rooftop parapet with 10 degrees of downtilt for campus coverage

Aruba recommends creating a 3D RF model of the campus showing all of the expected antenna mounting locations. For each AP position, the Aruba 3D Outdoor RF Planner creates an antenna pattern in 3 dimensional space that predicts the coverage for various data rates using variables provided by an engineer for output power, fade margin, cable loss and other factors. Here is an example showing campuswide coverage in the 5GHz and 2.4GHz frequency bands.

Figure 142 5 GHz campus coverage using the ANT-2x2-D805



Figure 143 2.4 GHz campus coverage using the ANT-2x2-D805

Be sure to consider high-density areas when preparing a campus design, such as ampitheaters or outdoor eating areas or university-style “quad” spaces. In such locations, it will be necessary to have overlapping coverage from multiple APs sufficient to handle the expected user capacity.



Appendix D: Intermodal Transportation Example

Aruba has worked closely with our customers and partners to develop our recommendations for ports, airports, truck, rail-yards, and other intermodal facilities. We are directly invested in helping our customers deliver reliable and responsive business applications at the lowest possible cost.

Application TypesIn this Enterprise network, several categories of wireless applications run on the network in a typical intermodal facility. Most often, mobile terminal applications are deployed that have the following attributes:

Real time Character-based Transmission Control Protocol (TCP)/Internet Protocol (IP) Uses vehicle-mounted or rugged portable Windows platforms Can be mounted in a truck, mobile crane, or locomotive

Operators frequently have deployed wireless analog and digital video cameras that share the 2.4 GHz band. These devices are often not 802.11n-based devices, but because they share the same frequency band, these cameras and 802.11 systems have been shown to interfere with each other and with the Wi-Fi network. In many cases, migrating these systems to 802.11n-based systems can reduce interference and improve system performance and return on investment for the whole network.

Many intermodal facility operators have also expressed interest in using Voice over Wi-Fi handsets to provide voice without having to pay recurring monthly charges to wireless carriers.

Support for standard wireless devices such as laptops is often required so that management and engineering can access home network resources while working in the field. Mobile terminal applications are often revenue-generating devices, and when service is unavailable, yard operations can be negatively affected directly.

One of the most common business drivers behind network refreshes at intermodal facilities, such as ports, is that 900 MHz radios used by first-generation application terminals are, in general, no longer available. Another common driver is that existing vehicle-mounted and handheld rugged data terminals are approaching end-of-life.

Typically, intermodal facility operators select standards-based 802.11a/b/g/n WLAN technology that operates in 2.4 GHz and 5 GHz as the replacement network platform inside their yards.

Aruba Networks, Inc. Intermodal Transportation Example | 183


Dense Overhead Coverage StrategyThe most reliable design for mobile cranes, vehicles and ground personnel is the dense overhead strategy. In this design, Aruba mesh-capable access points are deployed densely (every 200-300m) and antennas are mounted high up, between 15-40m above ground level. Existing light poles, high masts and communications towers are typically used for mounting APs and antennas.

Aruba developed a specialized low-gain (typically 3-5 dBi) squint omnidirectional antenna that faces down to create tight cells. These tight cells are known to work reliably and deliver consistent performance in environments where many large vertical obstructions may exist. For more information, see Antenna Beamwidth, Pattern, and Gain on page 33. Figure 144 illustrates 5 Ghz coverage for an intermodal transportation facility.

Figure 144 Dense overhead coverage for an intermodel transportation facility

This strategy results in excellent coverage deep inside container stacks and other obstructions. Because of the consistent cell spacing, it also delivers superior voice support. The dense radio topology, with LOS between many APs is consistent with voice handset vendor best practices and the high number of alternate paths results in a very reliable system.



Sparse Side Coverage StrategyThis RF coverage strategy uses existing towers and structures that have power and data services. In this deployment area, we use very high-gain (≥13 dBi), sectored, wide horizontal (120 degree), narrow elevation (8 degree) antennas to cover as much range as possible from each radio position.

The main advantage of this strategy is that it reduces installation costs by leveraging a limited number of existing mounting assets that have power or wired backhaul. See Figure 145 for an example of an intermodal transportation facility.

Figure 145 Spare side coverage for an intermodal transportation facility

This solution works well for fixed cameras where predictable link budgets can be worked out in advance. However, automated RF management algorithms such as ARM or RFM cannot be used in this type of environment due to the rapidly changing RF environment.



Appendix E: Open Pit Mine Example

The Wi-Fi mesh network is an integral part of mining operations. The mesh network allows for the locations of the various grades of ore to be closely tracked so that the blending process can be precisely controlled, which results in the required grades of ore for shipping.

Mining operations contain a very wide variety of applications. For example: Video (CCTV) and portable devices (handheld computers, smartphones) Research and support vehicles Mining vehicles, such as dump trucks, excavators, dozers, front-end loaders, graders, water

trucks, service trucks, fuel tankers, workshop trucks, and blast rigs.

Research equipment also requires additional high capacity wireless bandwidth: Ground-probe radar, portable equipment for data logging etc.

In addition, a changing mix of fixed and semifixed locations such as trailers and buildings need coverage. These nodes can be moved throughout the mine pit area as the shape of the mine pit changes to maintain Wi-Fi mesh coverage throughout the mine pit.

An open pit mine is an extremely complex environment that requires a flexible WLAN solution for standard coverage as the parameters of the mine changes.

Figure 146 is an overview of the coverage. The pit coverage is shown in yellow and the access road coverage is shown in orange. A high-capacity multichannel linear mesh circles the pit and provides alternate paths in case of a backhaul radio failure. The client radio APs are located at the edge of the pit with antennas directing the radio signals into the mine. For this depiction, the wave propagation is shown tilted down by 30 degrees into the pit. The backhaul mesh is shown in the blue and yellow lines, and the client access is shown as opaque orange and yellow coverage.

Figure 146 Open pit mine overview with MIMO data rates

Aruba Networks, Inc. Open Pit Mine Example | 187


Figure 147 represents the expected coverage of the mine itself. The distance between nodes is approximately 1 kilometer and this model shows the expected throughput between each node.

Multiradio mesh configuration is supported in AirMesh by using wireless mesh. Interference management is important, and a separation of at least 2 channels is recommended. Antennas should be mounted carefully to minimize interfering signal levels between upstream (wired network) and downstream (client-device) connections.

Directional sectored antennas can be used to extend distances several kilometers between nodes and throughput is preserved.

Figure 147 is a model of the mine road coverage with backhaul nodes identified. This model can be extended to cover other locations within the mine territory such as crew living areas, railheads, and airstrips and is an example of a linear mesh.

Figure 147 Open pit mine road coverage

Aruba Networks, Inc. Open Pit Mine Example | 188


Appendix F: Aruba Stadium Design Summary

Planning a high-capacity, or “dense” WLAN to deliver ubiquitous WiFi coverage with transparent connectivity for POS devices, press and fans in an outdoor stadium environment requires careful RF design to ensure adequate signal is delivered to all areas and to properly manage co-channel interference. Considerations include antenna selection, antenna orientation both horizontally and vertically, elevation (e.g. the distance from the AP-to-client) and the effect of structural materials on signal propagation. In addition, pervasive high-throughput 802.11n coverage with multiple input-multiple output (MIMO) signal propagation behaves differently than traditional 802.11abg signals in this type of facility. In summary, there are many significant differences in planning the WLAN for a stadium vs. a typical carpeted indoor environment.

Aruba has extensive experience designing complex stadium WLAN solutions, including in stadium environments including for the 2008 All Star Game, Angel Stadium in Anaheim, the 2009 FIFA World Cup, the 2009 and 2010 Australian Open, American Airlines Center in Dallas, United Center in Chicago, and many more. In addition, we have ultra-high capacity stadium designs under construction around the world. In this section, Aruba presents our RF coverage strategy for a hypothetical American baseball stadium. Validation data from testing in real stadiums is presented along with the design below.

RF DesignFor a large venue such as a baseball stadium, planning in terms of both coverage and capacity is critically important, although in the most traditional sense these are somewhat opposing parameters. Whereas “coverage” usually refers to the use of high gain antennas and high power settings to achieve maximum range, “capacity” typically indicates use low gain, low power solutions to limit targeted coverage area per AP and provide better per user experience. To balance these two goals, it is important to leverage 3D antenna pattern visualization techniques which aid in antenna selection and placement.

When considering any new stadium deployment, the following should be kept in mind: AP density is determined by a combination of subscriber count and an “oversubscription ratio.”

Aruba APs can support up to 150 associated clients per radio per AP, though it is considered a best practice to keep the number of simultaneously transmitting clients to 50 or fewer with 802.11n APs in order to maximize average throughput for each client. The oversubscription ratio is similar to that used in a WAN circuit or a LAN switch, where the backplane typically has a lower total capacity than the total aggregate capacity of all ports because not all ports are fully utilized at the same time in most environments. A radio with 150 associated users and 50 transmitters would have a 3-to-1 oversubscription ratio.

High user count areas (such as press box or luxury suites) rely on more targeted coverage and as such can be provided coverage through access points concealed above the drop ceiling similar to indoor office space planning.

Overhangs and rafters should NOT be used to mount APs and antennas. While these locations generally provide good line of sight to a large targeted area, the signal dispersion is so great over just 20-30 feet that significant co-channel and adjacent-channel interference results.

Aruba Networks, Inc. Aruba Stadium Design Summary | 189


Therefore, a method to minimize the size of each cell, such as the “picocell” method proposed below, is necessary to achieve capacity objectives.

With many more radio cells to support thousands of users, it is critically important to minimize overlap of radio cells covering different areas, such as the upper level from the lower level. The 2.4GHz band only has 3 non-overlapping channels, as opposed to 5GHz where there are as many as 9 non-overlapping non-DFS channels (and up to 11 more in the DFS range).

The following section synthesizes all of these considerations with the customer requirements and stated assumptions to produce a complete design.

“Picocell” Coverage StrategyBased on extensive real-world testing, Aruba recommends an RF design strategy that utilizes three interleaved “blankets” of APs to cover the lower sections, club level, and upper sections respectively. We call this approach “picocell” and it can be visualized as follows:

Figure 148 Using three separate coverage 'blankets' to minimize inter-blanket interference

In the figure above, you can see one AP from each blanket. To minimize interference between individual blankets, the lower deck antennas are mounted underneath the precast concrete seats and aimed through the concrete up above the field. Similarly, the upper deck antennas are also mounted underneath the concrete and aimed up through the concrete. In both cases, the antenna is aimed to match the slope of the seats to maximize coverage and minimize self-interference.

In the middle, a blanket of APs is installed in the suites above or below the ceiling (below is preferred for optimal RF performance). Blowups of each blanket along with physical installation details are provided below.



Within each blanket, the design calls for APs to be installed every section. On the suite level the APs are placed every other suite. The APs on each suite level are staggered relative to the APs above or below, creating a checkerboard pattern when viewed from the side. A fourth blanket covers the exterior gate areas and the fan walkways on street level. Of course, the AP densities just described may vary from stadium to stadium depending on the precise construction materials and user density requirements.

With all blankets installed, the stadium would look similar to the 3D model below. This is a predictive RF model showing the estimated -65dBm signal boundary for each picocell. The different colors represent different 802.11 channels in the 2.4GHz band.

Figure 149 Visualization of stadium-wide coverage blankets at -67dBm cell edge power

While not depicted in the model, there are also AP blankets on the press level, the mezzanine (office) level, and the lower service level. Also, each of the concourses has its own blank of APs to provide reliable coverage to the concession areas as well as roaming coverage to fans on the move.

Understanding the Structure of a PicocellThis section will help wireless engineers understand the requirements and constraints of channel reuse. Picocells are one of the most interesting and challenging RF designs to undertake. Controlling the collision domain requires careful analysis in three dimensions, not only between APs and clients but also between clients.



An under-floor picocell system has two key RF design differences from a conventional WLAN deployment:

Client device radios increase picocell radius as opposed to shrinking a cell radius as is usually the case

The link budget for clients at the picocell edge must factor in variable amounts of structural and body loss

In conventional WLAN deployments, wireless designers following best practices will consider client radio capabilities to restrict the effective cell size of an AP. This is because wireless links are two-way, and the weakest link is typically from the client to the AP. A good predictive model will use the lesser of the AP or client transmit power, plus the sum of both AP and client antenna gains, to estimate the distance at which a given data rate is available. Client transmissions beyond the cell can be safely ignored due to the spacing of APs in most environments.

With a picocell design, the extreme proximity of overlapping APs and clients means that client radios effectively increase the size of a cell. This is illustrated in the figure below.

Figure 150 Structure of a picocell

The picocell is divided into the following components in the H-plane: Inner AP Radius (r1) – This is the usual cell edge of the AP. It is the target data rate radius, not

the interference radius. It is defined as the maximum distance at which the SNR exceeds the value required to demodulate the desired minimum data rate, typically 150Mbps in an HD-WLAN using 20MHz channels. In a picocell operating at very low transmit power, this distance is often less than 30 feet (10m).



Client Interference Radius (r2) – This is the distance at which the 802.11 preamble of any AP or client radio transmission (transmitted at 6Mbps per the standard) will interfere with a same-channel transmission by another station. Typically, this means that the SNR is 4dB or greater, which is the minimum required to decode an 802.11a/n frame. Therefore, this is much greater than the Inner AP Radius.

Outer Picocell Radius (r3) – This is the outer boundary of all the Client Interference Radii when multiple clients exist at the edge of the Inner AP Radius. This is the effective radius of the picocell.

From this analysis, it is clear that client transmissions have the result of increasing the size of a picocell due to the small power levels and short distances involved.

Picocell Coverage & Performance Validation Test ResultsAruba has carried out detailed RF propagation studies of concrete at over a dozen stadiums. We have studied concrete as old as the 1920s and as recent as 2005. To help customers and partners better understand Aruba’s picocell strategy, we share below the results of some of those tests.

The basic installation of a picocell is fairly straightforward. Aruba recommends the ruggedized AP-175 802.11n MIMO access point which is IP66 rated and therefore does not require enclosures with labor-intensive bulkhead connectors and RF patch cables. For the antenna, we use two of our ANT-2x2-D607 which is shown in the photos below. This is a 7dBi, 60°H x 60°E dual band, dual element antenna with multiple polarizations inside the radome for improve MIMO performance. Each antenna has two pigtails, which connect to the two corresponding antenna interfaces on each radio.

Figure 151 Picocell installation underneath concrete stadium seating

N O T E

A picocell network works best when the seats are full. In this case, the increased lateral human body attenuation will shrink the client interference radius dramatically. This effect is deliberately exploited in picocell design to achieve reuse.



To measure RF signal from this AP/antenna combination, Aruba used AirMagnet Survey Pro 7.0 with AirMagnet 802.11abgn card to walk every 2nd row for two sections on either side of the AP. Here is the AirMagnet heat map for the 2.4GHz band with RSSI filters of -65, -72 and -85dBm, and signal-to-noise ratio (SNR) filters of 30, 20 and 10dB respectively. (With the filter applied, any area that has more signal than the filter appears in color, while any area that does not meet the criteria appears in grey.)

Figure 152 Picocell RF propagation in 2.4GHz and 5GHz with AirMagnet Survey

N O T E

The TX power during this test was set to 12dBm, plus 7dBi of gain, for an EIRP of 20dBm.



Aruba has also studied bidirectional throughput for the picocell design with a mix of common client devices. We use the WaveDeploy Pro tool from VeriWave Corporation, which loads a small software agent on each client. The agents are remotely controlled from a test console that starts and stops each test run, and visualizes the resulting data.

Figure 153 Picocell TCP throughput heatmap with VeriWave WaveDeploy

For these tests, we used a Windows 7 laptop with a 2x3 802.11n wireless network adapter, an iPhone 3GS, and an iPad 1. The square box with the red dashes shows the minimum target cell size of 25 seats by 20 rows, centered directly above the AP under the stands. You can see that the laptop is able to deliver over 25Mbps of UDP throughput using HT20. The iPad with its HT20 radio delivers over 15Mbps down and 10Mbps up, while the iPhone (as expected) achieves around 10Mbps upstream and 5Mbps downstream. The iPhone 3GS is a legacy 802.11g radio, with a very limited network processor.

Overall, from both an RF and an IP throughput perspective, the picocell design developed by Aruba provides excellent near-field performance, while greatly limiting the interference radius of each individual cell.

For additional information on RF and capacity planning for high density environments, Aruba has published a Validated Reference Design for High Density (HD) WLANs. This free document includes important concepts, methodologies, and performance data needed by a wireless architect to plan this type of environment.



Appendix G: Terminal Doppler Weather Radars

In 2009, the FCC, NTIA, FAA and industry began working to resolve interference to Terminal Doppler Weather Radar (TDWR) systems used near airports that has occurred from some outdoor wireless systems operating in the 5470 MHz – 5725 MHz band. These wireless devices are subject to the FCC Section 15.407 of our rules and when operating as a master device they are required to implement radar detection and DFS functions. The FCC in conjunction with industry is continuing to develop long-term improvements to the DFS test procedures that will ensure that devices better protect TDWR operations.

Figure 154 United States Terminal Doppler Weather Radar Sites

In the interim, the FCC has published new requirements and guidance for operation of outdoor wireless devices which are summarized as follows:

1. Operation in the 5570-5680 MHz frequency range is no longer permitted. As a result when running ArubaOS versions (TBD) or higher, channels 116-128 have been disabled on all Aruba US controllers and access points.

2. When installed outdoors professional installation is required. Although EIRP and channel restrictions are enforced by ArubaOS, professional installers must be familiar with the FCC requirements and ensure that the product is installed in compliance with the FCC granted conditions. These additional conditions include:a. Only Aruba approved antennas and accessory hardware may be used for outdoor

installations. Use of non-Aruba outdoor antennas or third party RF components could result in operation outside the FCC granted conditions. Use of non-approved components can cause spurious and impact compliance with the new TDWR protection measures.

Aruba Networks, Inc. Terminal Doppler Weather Radars | 197

http://support.arubanetworks.com


http://www.spectrumbridge.com/udia/home.aspx

http://www.spectrumbridge.com/udia/home.aspx


b. Changes or modifications to the equipment are not permitted. Any unapproved changes can result in non-compliant operation and will void the equipment warranty.

c. There are no user serviceable parts, all repairs and service must be handled by Aruba Support. (1-800-wifilan), http://support.arubanetworks.com.

d. All products using external antennas must be professionally installed and the installer must ensure that the antenna gain is correctly provisioned in software during setup. The antenna gain used for provisioning may be reduced by the cable loss measured at the installation:

Gprovisioned = Gantenna – Cable Loss (dB)

3. For installation within 35 km of a TDWR location, the FCC requests voluntary registration. A voluntary WISPA sponsored database has been developed that allows operators and installers to register the location information of the UNII devices operating outdoors in the 5470 – 5725 MHz band within 35 km of any TDWR location (see http://www.spectrumbridge.com/udia/home.aspx). This database may be used by government agencies in order to expedite resolution of any interference to TDWRs.

Figure 155 35 Kilometer Voluntary Registration Zones in Florida State




Table 14 provides the list of TDWR locations.

Table 14 TDWR Locations and Frequencies

State City Latitude (N), Longitude (W) Frequency (MHz)

AZ PHOENIX 33.420352, -112.16318 5610

CO DENVER 39.72722, -104.52639 5615

FL FT LAUDERDALE 26.142601, -80.34382 5645

FL MIAMI 25.757083, -80.491076 5605

FL ORLANDO 28.343125, -81.324674 5640

FL TAMPA 27.85867, -82.51755 5620

FL WEST PALM BEACH 26.687812, -80.272931 5630

FL WEST PALM BEACH 26.687812, -80.272931 5615

GA ATLANTA 33.646193, -84.262233 5615

IL MCCOOK (ORD) 41.796589, -87.857628 5615

IL CRESTWOOD (MDW) 41.6514, -87.7294 5645

IN INDIANAPOLIS 39.636556, -86.435286 5605

KS WICHITA 37.506844, -97.437228 5603

KY COVINGTON-CINCINNATI 38.89778, -84.58028 5610

KY LOUISVILLE 38.04581, -85.610641 5646

LA NEW ORLEANS 30.021389, -90.402919 5645

MA BOSTON 42.15806, -70.93389 5610

MD BRANDYWINE 38.69528, -76.845 5635

MD BENFIELD (BWI) 39.09056, -76.63 5645

MD CLINTON (DCA) 38.758853, -76.961837 5615

MI DETROIT 42.11111, -83.515 5615

MN MINNEAPOLIS 44.870902, -92.932257 5610

MO KANSAS CITY 39.49861, -94.74167 5605

MO ST LOUIS 38.804691, -90.488558 5610

MS DESOTO COUNTY (MEM) 34.896044, -89.992727 5610

NC CHARLOTTE 35.337269, -80.885006 5608

NC RALEIGH DURHAM 36.001401, -78.697942 5647

NJ WOODBRIDGE (EWR) 40.593397, -74.270164 5620

NJ PENNSAUKEN (PHL) 39.950061, -75.069979 5610

NV LAS VEGAS 36.144, -115.007 5645



NY FLOYD BENNETT FIELD 40.588633, -73.880303 5647

OH DAYTON 40.021376, -84.123077 5640

OH CLEVELAND 41.289372, -82.007419 5645

OH COLUMBUS 40.00611, -82.71556 5605

OK AERONAUTICAL CENTER 35.405, -97.625 5610

OK AERONAUTICAL CENTER 35.393, -97.629 5620

OK TULSA 36.070184, -95.826313 5605

OK OKLAHOMA CITY 35.27611, -97.51 5603

PA HANOVER (PIT) 40.501066, -80.486586 5615

PR SAN JUAN 18.47394, -66.17891 5610

TN NASHVILLE 35.979079, -86.661691 5605

TX HOUSTON INTERCONTL (IAH) 30.06472, -95.5675 5605

TX PEARLAND (HOU) 29.515852, -95.241692 5645

TX DALLAS LOVE FIELD 32.92494, -96.968473 5608

TX LEWISVILLE (DFW) 33.064286, -96.915554 5640

UT SALT LAKE CITY 40.967222, -111.929722 5610

VA LEESBURG (IAD) 39.083667, -77.529224 5605

WI MILWAUKEE 42.81944, -88.04611 5603

Table 14 TDWR Locations and Frequencies (Continued)

State City Latitude (N), Longitude (W) Frequency (MHz)



Appendix H: Aruba Contact Information

Contacting Aruba Networks

Web Site Support

Main Site http://www.arubanetworks.com

Support Site https://support.arubanetworks.com

Software Licensing Site https://licensing.arubanetworks.com/login.php

Wireless Security IncidentResponse Team (WSIRT)

http://www.arubanetworks.com/support/wsirt.php

Support Emails

Americas and APAC [email protected]

EMEA [email protected]

WSIRT EmailPlease email details of any securityproblem found in an Aruba product.

[email protected]

Validated Reference Design Contact and User Forum

Validated Reference Designs http://www.arubanetworks.com/vrd

VRD Contact Email [email protected]

AirHeads Online User Forum http://airheads.arubanetworks.com

Telephone Support

Aruba Corporate +1 (408) 227-4500

FAX +1 (408) 227-4550

Support

United States +1-800-WI-FI-LAN (800-943-4526)

Universal Free Phone Service Numbers (UIFN):

Australia Reach: 11 800 494 34526

United States 1 800 94345261 650 3856589

Canada 1 800 94345261 650 3856589

United Kingdom BT: 0 825 494 34526MCL: 0 825 494 34526

Aruba Networks, Inc. Aruba Contact Information | 201

http://www.arubanetworks.com

https://support.arubanetworks.com

https://licensing.arubanetworks.com/login.php

http://www.arubanetworks.com/support/wsirt.php

mailto:[email protected]



http://www.arubanetworks.com/vrd%20



Universal Free Phone Service Numbers (UIFN):

Japan IDC: 10 810 494 34526 * Select fixed phonesIDC: 0061 010 812 494 34526 * Any fixed, mobile & payphoneKDD: 10 813 494 34526 * Select fixed phonesJT: 10 815 494 34526 * Select fixed phonesJT: 0041 010 816 494 34526 * Any fixed, mobile & payphone

Korea DACOM: 2 819 494 34526KT: 1 820 494 34526ONSE: 8 821 494 34526

Singapore Singapore Telecom: 1 822 494 34526

Taiwan (U) CHT-I: 0 824 494 34526

Belgium Belgacom: 0 827 494 34526

Israel Bezeq: 14 807 494 34526Barack ITC: 13 808 494 34526

Ireland EIRCOM: 0 806 494 34526

Hong Kong HKTI: 1 805 494 34526

Germany Deutsche Telkom: 0 804 494 34526

France France Telecom: 0 803 494 34526

China (P) China Telecom South: 0 801 494 34526China Netcom Group: 0 802 494 34526

Saudi Arabia 800 8445708

UAE 800 04416077

Egypt 2510-0200 8885177267 * within Cairo02-2510-0200 8885177267 * outside Cairo

India 91 044 66768150

Telephone Support

Aruba Networks, Inc. Aruba Contact Information | 202