Self-organization on Cellular Wireless Network and WLAN

Paul LinMarch 20, 2006

Contents Overview of self-organization Self-organizing on Cellular wireless

network Topology generation and dynamic

routing Issues of self-organization Conclusion

Challenges The size and scope of mobile wireless

networks continue to grow with more users and devices distributed from homes, businesses, to city and world-wide. This is adding to spatiotemporal complexity of the network topology and dynamics.

Due to unpredictability of the network, static setting is insufficient.

Features of Self-Organization Adapting to real-time situation

Optimal resource planning

Self-management and cooperation

Features of Self-Organization (II)

Self-organization is not just distributed and localized control; it is about the relationship between the behavior of individual entities and resulting structure and functionality of the overall system.

The application of rather simple behavior at the microscopic level leads to sophisticated organization of the overall system – emergent behavior

Design Paradigms #1.Design local behavior rules that

achieve global properties #2.Do not aim for perfect coordination:

exploit implicit coordination #3.Minimize long-lived state information #4.Design protocols that adapt to changes

Christian Prehofer and Christian Bettstetter, DoCoMo Euro-labs, IEEE Communication, July 2005

Protocols according to levels of locality/coordination

Design Paradigms putting together

Overview of Cellular Wireless Network

Self-Org of Base Stations BS is able to operate in a standalone fashion. BS collaborate with its peers. Probing phase: (1)auto-configures its IP connectivity, subnet and

uplink interface. (2)channel scan to detect other base stations in its

immediate neighborhood. (3)contact neighbor stations through uplink, and

integrates itself into the network-wide information exchange.

Periodically performs channel scan to detect changes in its environment.

Self-organizing technologies

Adaptive cell sizing

By decreasing the cell radius from 500 to 200m, a capacity increase of 33% is achieved for voice service

Revenue-based cell size control: adjusts beacon transmit power. Under congestion the cell will limit its service area and reduce the inter-base-station interference. This will enable it to serve more users closer to the base station. In light traffic conditions, it will expand and improve the coverage with cells overlapping the same area.

Power control: Ensure a certain quality of service is used, as well as improving capacity

Fixed relay node

Complex Behavior of Nodes Small World: refers to a phenomenon where

the average path length between nodes is small, the nodes are highly clustered, and connectivity distribution peaks at an average value and then decays exponentially.

- the hypothesis that everyone in the world can be reached through a short chain of social acquaintances.

Scale-free: connectivity distributions can be represented by power-law form, which is independent of the size or scale of the network.

Random vs. Scale-free

Parameters of complex network Average path length (L) -average number of hops(edges) in the shortest path

between two nodes Clustering coefficient (C) -average fraction of pairs of neighbors of a node that

are also neighbors of each other Degree (K) -number of links connecting that node to the

neighboring nodes

Small world concept A small world: Average path length(L) is small, and

clustering coefficient is high. It is shown that randomly rewiring a few edges reduces

the average distance between nodes, but little effect on the clustering coefficient.

The degree distribution is exponential. Nodes with high connectivity are practically absent, power-law property is not observed.

Scale-free model Real networks expand continuously by addition of new

nodes, and new nodes attach preferentially to nodes that are already well connected.

Figure shows starting with 3 nodes, and each step adding new node with 2 edges.

Applying the model It is anticipated that

infrastructureless, deployable, wireless relay stations will be used the addition to the cellular infrastructure to improve service to mobile users.

The objective is to design a scale-free overlay FRN network for QoS purposes.

Topology generating

Scale-free result

Internet clustering coefficient is measured to be greater then 0.18, and web is 0.1078.

Dynamic Routing Method 1: Load balancing among only

BSs.

Method 2: Load balancing among FRNs and BSs with no change in destination BS

Method 3: Load balancing among FRNs and BSs with change in Destination BS

Routing experiment

The location of mobile users are generated according to a uniform distribution.

BSs:3 , FRNs: 50, K=1

Discussion issues #1. Cell configuration

#2. Efficient Planning

#3. Coordination on different nodes and layers

Cell configuration The backbone of the wireless mobile

network is the entry points to the networking, especially the cell concept with BSs at the center.

Cells should be able to flexibly adjust its topological coverage to facilitate the flow of signals or packets.

For example: Cell-Dimensioning Algorithm

example: Cell-Dimensioning Algorithm

example: Cell-Dimensioning Algorithm

example: Cell-Dimensioning Algorithm(II)

Cell boundaries before and after BSR x removed

Efficient Planning Adaptation to resources with potential

aspects, ex. Cost, capacity, traffic..etc. Dynamic routing Advanced modeling of reinforcement

learning, which configure service coverage and system capacity dynamically to balance traffic loads among cells by being aware of the system situation.

Example: Integrated Cellular and Ad-hoc Relay System

Coordination on different nodes and layers

SOPRANO

A wireless multihop network overlaid with a cellular structure: base station(BS), router(R), and terminals(T)

Self-Organizing Packet Radio Ad-hoc Networks with Overlay (SOPRANO), IEEE Communications June 2002

Conclusion Self-organization reduces costs,

improves robustness, enhances effectiveness and performance, facilitates automatically utilization of cellular wireless networks.

Its overall goal is to enhance QoS.

References Self-Organization in Communication Networks: Principles and Design Paradigms, Christian

Prehofer and Christian Bettstetter, DoCoMo Euro-Labs, IEEE Communications Magazine • July 2005, p 78-85

Self-organization in future mobile communications, by A. G. Spilling, A. R. Nix, M. A. Beach and T. J. Harrold, ELECTRONICS Xr COMMUNICA'IION ENGINEERING JOURNAL JUNE 2000

Self-Management of Wireless Base Stations, Kai Zimmermann, Lars Eggert and Marcus Brunner, www.ambient-networks.org

On the Design of Self-Organized Cellular Wireless Networks, Sudhir Dixit, Evs, en Yanmaz, and Ozan K. Tonguz, IEEE Communications Magazine • July 2005

Self-Organizing packet Radio Ad Hoc Networks with Overlay (SOPRANO), Ali N. Zadeh and Bijan Jabbari, Raymond Pickholtz and Branimir Vojcic, IEEE Communications Magazine • June 2002

Reinforcement-learning-based self-organization for cell configuration in multimedia mobile networks, Ching-Yu Liao, Fei Yu, Victor C. M. Leung and Chung-Ju Chang, EUROPEAN TRANSACTIONS ON TELECOMMUNICATIONS,Euro. Trans. Telecomms. 2005; 16:385–397

Applying Emergent Self-Organizing Behavior for the Coordination of 4G Networks Using Complexity Metrics, Lester T. W. Ho, Louis G. Samuel, Jonathan M. Pitts, Bell Labs Technical Journal 8(1), 5–25 (2003)