on the performance of the ser-sa protocol in tactical edge … · 2018. 1. 22. · in this report,...

42
CAN UNCLASSIFIED CAN UNCLASSIFIED On the performance of the SER-SA protocol in tactical edge networks A comparison among SER-SA, OLSR, and AODV Ronggong Song Ming Li J. David Brown Mazda Salmanian DRDC – Ottawa Research Centre Defence Research and Development Canada Scientific Report DRDC-RDDC-2017-R140 October 2017

Upload: others

Post on 15-Sep-2020

0 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

CAN UNCLASSIFIED

CAN UNCLASSIFIED

On the performance of the SER-SA protocol in tactical edge networks A comparison among SER-SA, OLSR, and AODV

Ronggong Song Ming Li J. David Brown Mazda Salmanian DRDC – Ottawa Research Centre

Defence Research and Development Canada Scientific Report DRDC-RDDC-2017-R140 October 2017

Page 2: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

CAN UNCLASSIFIED

CAN UNCLASSIFIED

IMPORTANT INFORMATIVE STATEMENTS Disclaimer: Her Majesty the Queen in Right of Canada (Department of National Defence) makes no representations or warranties, express or implied, of any kind whatsoever, and assumes no liability for the accuracy, reliability, completeness, currency or usefulness of any information, product, process or material included in this document. Nothing in this document should be interpreted as an endorsement for the specific use of any tool, technique or process examined in it. Any reliance on, or use of, any information, product, process or material included in this document is at the sole risk of the person so using it or relying on it. Canada does not assume any liability in respect of any damages or losses arising out of or in connection with the use of, or reliance on, any information, product, process or material included in this document. This document was reviewed for Controlled Goods by Defence Research and Development Canada (DRDC) using the Schedule to the Defence Production Act. Endorsement statement: This publication has been peer-reviewed and published by the Editorial Office of Defence Research and Development Canada, an agency of the Department of National Defence of Canada. Inquiries can be sent to: [email protected].

Template in use: (2010) SR Advanced Template_EN (051115).dotm

© Her Majesty the Queen in Right of Canada (Department of National Defence), 2017

© Sa Majesté la Reine en droit du Canada (Ministère de la Défence nationale), 2017

Page 3: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

DRDC-RDDC-2017-R140 i

Abstract

This report evaluates the performance of a recently proposed protocol for simultaneously delivering broadcast situational awareness traffic and performing proactive routing in a Tactical Edge Mobile Ad Hoc Network (TEN). The proposed routing strategy—called SER-SA, for secure and efficient routing by leveraging situational awareness messages—takes advantage of the fact that nodes in future tactical networks are expected to periodically broadcast Situational Awareness (SA) messages to other nodes in the network, indicating their position and status.

With simulations examining various scenarios such as different network size, radio transmission range, and mobility patterns, we compare the performance of SER-SA against other popular Mobile Ad Hoc Network (MANET) routing protocols—specifically, Optimized Link State Routing (OLSR) and Ad Hoc On Demand Distance Vector Routing (AODV). When considering channel utilization and traffic delivery ratio as the performance metrics for our study, we observe that there is no “silver bullet,” i.e., there is no single best routing protocol for every considered TEN scenario. This report highlights the cases under which certain protocols offer benefits over others, and when a particular routing protocol is advantageous.

Significance to Defence and Security

Effective routing protocols are important in tactical ad hoc networks, to allow dismounted and mounted network users to share real time broadcast and unicast information. Due to the limited bandwidth available in many tactical scenarios, these routing protocols must operate efficiently, reducing signaling overhead where possible and providing efficient routes for both unicast communication and broadcast data dissemination. For this reason, traffic delivery ratio and channel utilization are important performance metrics in selecting a network routing protocol.

In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages (that already exist in the network) to build routing tables for providing secure routing services for all other Traffic In Tactical Edge Networks (TEN). Through simulation, we compare the performance of SER-SA against well-known existing TEN routing protocols, paying particular attention to scenarios consistent with the vision presented by the Integrated Soldier System Program (ISSP).

This work is important to the defence and security community since it presents practical and detailed guidance in selecting routing strategies for a number of TEN use cases, considering aspects such as mobility speed and transmission range.

Page 4: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

ii DRDC-RDDC-2017-R140

Résumé

Le présent rapport évalue les performances d’un protocole récemment proposé qui permet la distribution simultanée du trafic de connaissance de la situation à destinations multiples et la réalisation du routage proactif dans un réseau mobile ad hoc tactique en périphérie (TEN). La stratégie de routage proposée, appelée SER-SA (acheminement protégé et efficace en exploitant des messages de connaissance de la situation), tire profit du fait que les nœuds dans des réseaux tactiques futurs devraient diffuser périodiquement des messages de connaissances de la situation (CS) aux autres nœuds dans le réseau pour indiquer leur position et leur état.

À l’aide de simulations visant à examiner divers scénarios, comme la taille de différents réseaux, la portée de transmission radioélectrique et les tendances concernant la mobilité, nous comparons les performances de SER-SA aux autres protocoles populaires de routage de réseaux mobiles ad hoc (MANET), en particulier le protocole Optimized Link State Routing (OLSR [routage destiné aux réseaux maillés, sans fil ou mobiles]) et le protocole de routage Ad-hoc On-demand Distance Vector (AODV [vecteur de distance ad hoc sur demande). Lorsque nous examinons le taux d’utilisation des canaux et de distribution du trafic comme paramètres de rendement dans notre étude, nous constatons qu’il n’y a pas de « solution miracle », c.-à-d. qu’il n’y a pas de protocole de routage universel pour tous les scénarios de TEN examinés. Le présent rapport met en évidence les cas où certains protocoles présentent des avantages par rapport aux autres, et ceux où un protocole de routage en particulier est avantageux.

Importance pour la défense et la sécurité

Les protocoles de routage efficaces sont importants dans les réseaux ad hoc tactiques pour permettre aux utilisateurs de réseau débarqués et embarqués d’échanger de l’information à destinataire unique et à destinataires multiples en temps réel. En raison de la largeur de bande limitée disponible dans de nombreux scénarios tactiques, ces protocoles de routage doivent fonctionner efficacement, en limitant le surdébit de signalisation lorsque c’est possible et en fournissant des trajets efficaces pour la diffusion de communications à destination unique et de données à destinations multiples. Pour cette raison, le taux de distribution du trafic et l’utilisation des canaux constituent des paramètres de rendement importants dans le choix d’un protocole de routage de réseau.

Dans le présent rapport, nous présentons une stratégie de routage récemment proposée (SER-SA) en exploitant des messages de CS (déjà présents dans le réseau) pour établir des tables de routage en vue d’offrir des services de routage sécuritaires pour tous les autres types de trafic dans les réseaux tactiques en périphérie (TEN). Au moyen de simulations, nous comparons les performances de SER-SA à des protocoles de routage de TEN actuels et bien connus, en portant une attention particulière aux scénarios qui sont conformes à la vision présentée par le Programme d’équipement intégré du soldat (PEIS).

Ces travaux sont importants pour la communauté de la défense et de la sécurité, car ils présentent des principes pratiques et détaillés dans le choix de stratégies de routage pour divers cas d’utilisation de TEN, en tenant compte d’aspects comme la vitesse de mobilité et la portée d’émission.

Page 5: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

DRDC-RDDC-2017-R140 iii

Table of Contents

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i Significance to Defence and Security . . . . . . . . . . . . . . . . . . . . . . i Résumé . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii Importance pour la défense et la sécurité . . . . . . . . . . . . . . . . . . . . ii Table of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2 The Routing Protocols Under Investigation . . . . . . . . . . . . . . . . . 3

2.1 Operation of SER-SA . . . . . . . . . . . . . . . . . . . . . . . 3 2.2 Operation of OLSR . . . . . . . . . . . . . . . . . . . . . . . . 4 2.3 Operation of AODV . . . . . . . . . . . . . . . . . . . . . . . 5

3 Tactical Edge Network Simulation Scenarios . . . . . . . . . . . . . . . . 6 3.1 Tactical Edge Network Scenario . . . . . . . . . . . . . . . . . . . 6 3.2 Simulation Considerations . . . . . . . . . . . . . . . . . . . . . 6 3.3 MATLAB Simulation Scenario . . . . . . . . . . . . . . . . . . . 8 3.4 EXata Simulation Scenario . . . . . . . . . . . . . . . . . . . . . 8

4 Comparison of SER-SA, OLSR, and AODV . . . . . . . . . . . . . . . . . 10 4.1 MPR Performance . . . . . . . . . . . . . . . . . . . . . . . . 10

4.1.1 Total MPR Performance . . . . . . . . . . . . . . . . . . . 10 4.1.2 Local MPR Performance . . . . . . . . . . . . . . . . . . . 11

4.2 Traffic Delivery and Channel Utilization Performance . . . . . . . . . . . 13 4.2.1 Broadcast Performance Investigation Under Different Jitter . . . . . . 14 4.2.2 Routing Performance under Mobility . . . . . . . . . . . . . . 16 4.2.3 Routing Performance under Different Radio Transmission Ranges . . . . 19

4.2.3.1 Baseline Performance—No Mobility . . . . . . . . . . . 19 4.2.3.2 Performance—Group Mobility at 5 m/s . . . . . . . . . . 23

5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 List of Symbols/Abbreviations/Acronyms/Initialisms . . . . . . . . . . . . . . . 31

Page 6: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

iv DRDC-RDDC-2017-R140

List of Figures

Figure 1: The strategies for providing SA and non-SA traffic services in a TEN: (a) traditional separated routing, (b) traditional single routing, (c) SER-SA. . . . 2

Figure 2: SER-SA MPR reduction percentage over the greedy heuristic algorithm. . . . 10

Figure 3: SER-SA MPR reduction percentage over the greedy heuristic algorithm under the proposed ISSP radio transmission ranges. . . . . . . . . . . . . . . 11

Figure 4: The average number of local MPRs generated by SER-SA and the greedy heuristic algorithm in a TEN network with 40 nodes. . . . . . . . . . . . 12

Figure 5: The average number of total MPR reduction and local MPR increase with SER-SA compared to the greedy heuristic algorithm. . . . . . . . . . . . 13

Figure 6: The broadcast delivery ratios under different jitter, number of broadcast nodes, and routing protocols. . . . . . . . . . . . . . . . . . . . . . . . 14

Figure 7: The average delivery ratios of SA, CO, and unicast traffic disseminated by SER-SA, OLSR, and AODV respectively. . . . . . . . . . . . . . . . 16

Figure 8: The average delivery ratios of the SA, CO, and unicast traffic disseminated by OLSR under 500 meter radio transmission range and different mobility speeds. 17

Figure 9: The average channel utilizations of SER-SA, OLSR, and AODV on disseminating mixed SA, CO, and unicast traffic. . . . . . . . . . . . . 18

Figure 10: The cumulative channel utilizations of SER-SA, OLSR, and AODV on disseminating mixed SA, CO, and unicast traffic. . . . . . . . . . . . . 19

Figure 11: The average delivery ratios of the mixed traffic delivered by SER-SA, OLSR, and AODV without mobility and under different radio transmission ranges. . 20

Figure 12: The average channel utilizations of SER-SA, OLSR, and AODV on disseminating the mixed traffic without mobility and under different average radio transmission ranges. . . . . . . . . . . . . . . . . . . . . . 21

Figure 13: The cumulative channel utilizations of SER-SA, OLSR, and AODV on disseminating the mixed traffic without mobility and under different average radio transmission ranges. . . . . . . . . . . . . . . . . . . . . . 22

Figure 14: The average delivery ratios of the SA traffic disseminated by SER-SA, OLSR, and AODV under different radio transmission ranges and mobility speeds. . . 23

Figure 15: The average delivery ratios of the CO traffic disseminated by SER-SA, OLSR, and AODV under different radio transmission ranges and mobility speeds. . . 24

Figure 16: The average delivery ratios of the unicast traffic disseminated by SER-SA, OLSR, and AODV under different radio transmission ranges and mobility speeds. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

Figure 17: The average channel utilizations of SER-SA, OLSR, and AODV on disseminating the mixed traffic under different radio transmission ranges and mobility speeds. . . . . . . . . . . . . . . . . . . . . . . . . . 25

Page 7: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

DRDC-RDDC-2017-R140 v

Figure 18: The cumulative channel utilizations of SER-SA, OLSR, and AODV on disseminating the mixed traffic under different radio transmission ranges and mobility speeds. . . . . . . . . . . . . . . . . . . . . . . . . . 26

Page 8: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

vi DRDC-RDDC-2017-R140

List of Tables

Table 1: The different mechanisms for delivering broadcast and unicast traffic by each routing implementation. . . . . . . . . . . . . . . . . . . . . . . 7

Table 2: The parameters and their related values used in our EXata simulation. . . . . 13

Page 9: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

DRDC-RDDC-2017-R140 1

1 Introduction

Sharing Situational Awareness (SA) among allied (or “blue force”) units is an important function in tactical operations, helping to enable the availability of a Common Operating Picture (COP) for commanders and soldiers. Blue force SA information—which can consist of data such as a user's location, identity, and status—is expected to be broadcast periodically and securely to ensure the tactical COP is both up-to-date and complete. Another important communication function in a Tactical Edge Mobile Ad Hoc Network (TEN) is the ability for a commander to deliver orders or update messages securely to all members, where these broadcast messages from the commander may be larger than typical blue force SA and may use more bandwidth. According to Hammel [1], these types of broadcast traffic, i.e., blue force SA and commander broadcasts, will comprise the majority of the network traffic in TENs in the near future.

Unfortunately, communication in TENs is complicated by the fact that nodes at the tactical edge face unreliable wireless links, and limited bandwidth and power. It is an open problem how best to disseminate periodic time-sensitive blue force tracking SA messages among all nodes, while at the same time providing a real-time broadcast service for delivering the commander’s broadcast orders/updates to all other nodes, and providing an on-demand unicast service for dynamic end-to-end traffic in TENs.

Although many efficient data collection and disseminating methods have been proposed for Wireless Sensor Networks (WSNs), e.g., [2–4], they do not satisfy the requirements of communication in TENs. For instance, in addition to broadcast services, a TEN needs to provide a secure real-time unicast service for communication between any two nodes, which is usually not supported by WSN dissemination methods. Using multiple routing protocols to identify routes for delivering different traffic types in TENs may not be efficient, due to the overhead of supporting multiple routing protocols and signaling schemes (when considering the scarce resources available, such as bandwidth and battery power).

With the aim of providing a secure and efficient method to deliver all traffic types in TENs, we proposed a new routing protocol in [5], called Secure and Efficient Routing by Leveraging Situational Awareness Messages (SER-SA). The main idea of the SER-SA protocol is to use periodic real-time blue force tracking SA messages as a mechanism to not only disseminate SA, but also to provide global network topology information to all nodes; the nodes in turn use the global topology information for proactive routing of other real-time broadcast and unicast traffic.

Instead of designing different designated routing protocols, e.g., [2–4, 6–9], with extra traffic overhead for delivering different traffic types (such as SA, broadcast, and unicast), SER-SA utilizes the blue force tracking SA messages to provide real-time routing services for delivering all other traffic in TENs. Figure 1 depicts the strategy of SER-SA compared with the traditional methods for providing SA and non-SA traffic services in TENs.

In addition to using the SA messages to carry routing information, SER-SA takes advantage of the fact that network-wide knowledge of SA information can be used to perform global network routing enhancements, which improve broadcast efficiency when compared to locally-made routing decisions as is traditionally done in TENs. SER-SA uses global network knowledge to consolidate relay nodes in the network (so-called multipoint relay, or MPR nodes), reducing the

Page 10: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

2 DRDC-RDDC-2017-R140

number of relay nodes compared to networks that do not make use of this global information. This reduction in the number of MPRs in the network results in a lower resulting broadcast bandwidth usage compared with existing popular proactive MANET routing protocols such as OLSR [8]. This global MPR consolidation method outperforms other existing MPR reduction algorithms such as those proposed in [11, 12], which reduces the number of MPRs by approximately 10%.

Figure 1: The strategies for providing SA and non-SA traffic services in a TEN:

(a) traditional separated routing, (b) traditional single routing, (c) SER-SA.

In this report, we compare the performance of the SER-SA protocol in tactical scenarios with OLSR (Optimized Link State Routing) and AODV (ad hoc on demand distance vector routing), using MATLAB and EXata [18] simulation platforms. In MATLAB, we simulate tactical networks of different network size and radio transmission range, and compare the number of MPRs generated by SER-SA with the number of MPRs expected using the standard greedy heuristic method from OLSR. In our EXata simulations, we simulate how the three protocols perform, using delivery ratio and channel utilization as performance metrics. We simulate three traffic types, i.e., regular blue force tracking SA, Commander’s Order, and unicast among different nodes in TENs, based on different mobility and radio transmission range assumptions. These TEN scenarios are consistent with the vision and use cases of the Integrated Soldier Systems Program (ISSP) [10].

The rest of the report is organized as follows. The SER-SA and other two routing protocols are briefly introduced in the next section. In Section 3, TEN scenarios and simulation platforms are discussed for SER-SA performance evaluation. In Section 4, the performance of SER-SA is presented and compared against OLSR and AODV. Finally, concluding remarks are given in Section 5.

Page 11: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

DRDC-RDDC-2017-R140 3

2 The Routing Protocols Under Investigation

In this report, we examine the performance of SER-SA as it compares with that of OLSR and AODV. The OLSR protocol is a natural choice for comparison, because SER-SA, as a proactive protocol itself, operates in a similar fashion as OLSR in many respects. We are also interested in AODV because it provides an alternative as an on-demand routing protocol. In this section, we provide an overview of the operations of the three protocols.

2.1 Operation of SER-SA

The SER-SA protocol provides a secure and efficient mechanism for jointly delivering blue force tracking SA, performing MPR selection and global consolidation, and performing route discovery for non-SA traffic in a TEN by utilizing periodically broadcast blue force tracking SA messages. A full description of the SER-SA protocol, including proposed packet formats and security architecture, is available in [5]; this section presents a summary describing the operation and capabilities of SER-SA.

SER-SA operates in two stages: the MPR selection stage and the network topology acquisition stage. The MPR selection stage is the initial stage of the network on start-up. In this stage, nodes immediately begin periodically broadcasting blue force SA information using a simple flooding approach. Each node inserts 1-hop neighbour information into any SA messages that it originates. When a node receives SA messages from its 1-hop neighbours, it rebroadcasts the SA message (since the SA is to be flooded through the network at this stage), but the node replaces the 1-hop information with its own neighbour list (essentially representing 2-hop neighbours of the node originating the SA message). Once each node has initiated several1 SA broadcasts, all nodes in the TEN will have a complete list of their own 1-hop and 2-hop neighbours, and will then begin to compute their “local MPR” nodes based on the greedy heuristic method described in [13]. After a pre-defined condition occurs, e.g., a certain time elapses,2 nodes will enter the network topology acquisition stage.

In the network topology acquisition stage, each node inserts its calculated “local MPR” information into its SA messages and broadcasts the SA messages to the whole network. In this stage, only MPR nodes relay the SA messages, i.e., the network no longer performs a simple flood of SA, but uses only the MPR nodes to re-broadcast SA messages. At this point, all SA and broadcast traffic can be delivered to all nodes in the network using only MPR nodes. Once the network topology acquisition stage has run for a few seconds, each node learns the entire network topology based on the MPR information contained in the received SA messages3 and can calculate a routing table for each destination using Dijkstra’s algorithm [14] for unicast traffic.

1 The default value is set to 3 in our simulation under the MPR selection stage. 2 The default value is set to 20 seconds in our simulation. The default SA broadcast interval is 5 seconds. 3 The length of time that the network topology acquisition stage needs to run before nodes will learn the entire network topology depends on the period of the SA messaging. In general, we have found that waiting for three or more periods is ample time for each node to acquire a complete network topology picture. The default period of the SA messaging is set to 5 seconds in our simulation.

Page 12: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

4 DRDC-RDDC-2017-R140

In contrast to other routing protocols such as OLSR and AODV, SER-SA achieves both dissemination of SA and the computation of MPRs and routing tables by using only (augmented) SA messages. In SER-SA, once all nodes learn the network topology (in the network topology acquisition stage), an “MPR global consolidation node” is selected and calculates redundant MPRs in the network.4 The “MPR global consolidation node” broadcasts the list of redundant MPRs to all other nodes. All nodes then update their local MPRs based on the global consolidation result and treat all remaining 1-hop MPR nodes as their “local MPRs.” With MPR global consolidation, SER-SA not only reduces the number of MPRs in the network but also increases the average number of local MPRs seen by each node (note that this counter-intuitive result is explored in detail in Section 4.1).

To account for changing network topologies as a result of mobility, SER-SA uses a network healing method to reconnect a node to the network once the node has lost connectivity to all MPRs in the network.5 Details of this network healing mechanism and additional details surrounding SER-SA message formats and security architecture are presented in [5].

2.2 Operation of OLSR

OLSR is a proactive routing protocol designed for MANETs employing an optimized multipoint relaying flooding techniques to diffuse link state information to all nodes in the network.

In OLSR, two different types of control messages (HELLO and Topology-Control) are emitted periodically and used for routing table maintenance. Each node periodically generates its HELLO messages6 and broadcast them to its 1-hop neighbours. Each HELLO message contains a node’s own address, a list of its 1-hop neighbours, and a list of its MPR neighbours. By exchanging HELLO messages among 1-hop neighbours, each node obtains information about its 1-hop and 2-hop neighbours and selects its MPR nodes from its 1-hop neighbours as relay nodes to reach all its 2-hop neighbours using the greedy heuristic algorithm [13]. Each Topology-Control (TC) message is generated by an MPR node and contains its own address and a list of nodes who have selected this node to serve as an MPR. By periodically flooding TC messages7 to the entire network through MPR nodes, each node receives TC messages from all MPR nodes in the network and obtains information of the entire network topology. Based on the global network topology information, each node calculates the optimal route to a destination using Dijkstra’s algorithm [14] and maintains its routing table. Therefore, routes are immediately available when needed in OLSR. Detailed information about OLSR is available in the IETF RFC 3626 [8].

4 As discussed in greater detail in [5], a redundant MPR is an MPR that can be removed from the network without affecting the ability of the remaining MPRs to cover connectivity to all nodes in the network. We note that redundant MPRs arise because the selection of MPRs is initially performed using local (2-hop) information, as opposed to global network information. 5 Network healing in SER-SA involves periodic calculation and consolidation of MPRs throughout the duration of the scenario. 6 In this report, OLSR uses its default value (i.e., 2 seconds) for HELLO interval. 7 In this report, OLSR uses its default value (i.e., 5 seconds) for TC interval.

Page 13: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

DRDC-RDDC-2017-R140 5

2.3 Operation of AODV

AODV is an on-demand routing protocol designed for MANETs employing route request and reply techniques to establish a route on demand between source and destination nodes.

In AODV, when a source node wishes to communicate with a destination node, it generates a Route Request (RREQ) message and broadcasts the RREQ message to the entire network. Once the RREQ message reaches either the destination node or an intermediate node with a “fresh enough” route to the destination node, the destination node or intermediate node generates a Route Reply (RREP) message and sends the RREP message back to the originator of the RREQ message through unicast—this is because each node receiving the RREQ message caches a route back to the originator of the RREQ message. A unicast route is established with hop count (represents the distance) between the source and destination node once the RREP message back to the source node. Detailed information about AODV is available in the IETF RFC 3561 [9].

For the purposes of this work, we implement a broadcast function in EXata for these three protocols to disseminate broadcast traffic such as SA in a network (see detail in Section 3.4).

Page 14: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

6 DRDC-RDDC-2017-R140

3 Tactical Edge Network Simulation Scenarios

In this section, we discuss the parameters used in developing our tactical edge network simulation scenarios. While some of the parameters are informed by the vision presented in the Integrated Soldier System Project (ISSP) [10], certain simplifying assumptions have been made with respect to node distribution and node mobility. The simulation scenarios and assumptions are presented in this section.

3.1 Tactical Edge Network Scenario

We consider a TEN serving a light infantry platoon. A typical light infantry platoon consists of a platoon leader, a second-in-command, four infantry, some heavy weapons detachments, and up to three sections. Each section consists of a section commander, a second-in-command, and six infantry. This light infantry platoon thus consists of maximum forty members in four groups, i.e., three sections and one platoon headquarters.

In our scenario, local blue force tracking SA messages need to be distributed throughout the TEN such that all members have SA of all other members; we assume that this is achieved through broadcast. To disseminate this information to higher orders of command, e.g., up to the Company or Battalion commanders, it is assumed that higher-power dedicated links are available between commanders (instead of relying on the lower-power broadcast function available to all nodes). This scenario is consistent with the vision presented in [1, 10, 16]. Furthermore, individual sections may need to operate independently (disconnected from the larger platoon), meaning that broadcast efficiencies are of interest in smaller networks as well, e.g., ten members.

3.2 Simulation Considerations

In our simulations we consider three different types of network routing implementations:

• SER-SA-based network

• OLSR-based network

• AODV-based network

Table 1 shows the different mechanisms by which each of these routing implementations delivers broadcast messages and unicast messages. In both OLSR and SER-SA implementations, MPR-flooding is used to deliver broadcast messaging. Note, however, that in the case of SER-SA, the MPR calculation is based on a global optimization (discussed in Section 2), whereas for OLSR the MPRs are calculated using a local greedy algorithm [13]. For AODV networks, broadcast is performed using a naïve flooding mechanism where every node indiscriminately re-broadcasts any received broadcast message. For unicast communication, OLSR and SER-SA networks both rely on proactively computed routing tables, whereas an AODV network determines routes “on demand” using route request (RREQ) and route response (RREP) messaging.

Page 15: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

DRDC-RDDC-2017-R140 7

Table 1: The different mechanisms for delivering broadcast and unicast traffic by each routing implementation.

AODV-based Network

OLSR-based Network

SER-SA-based Network

Broadcast Traffic

Pure Flooding MPR Flooding (Greedy Algorithm)

MPR Flooding (Global Optimization)

Unicast Traffic

On-Demand (RREQ/RREP)

Proactive (Routing Table)

Proactive (Routing Table)

The following additional assumptions have been adopted for our simulations:

• TEN Size: A TEN consists of maximum 50 mobile nodes based on the rationale described above; we will simulate TENs with different sizes in increments of 10, i.e., 20, 30, 40, and 50 nodes, in order to evaluate and compare the performance of SER-SA protocol with other routing protocols;

• TEN Bandwidth: We consider 1 Mbps as bandwidth in our simulations; this is consistent with the vision presented in [16];

• Node Distribution: For simplicity of simulation, the mobile nodes are randomly distributed in a 2 km by 2 km square battlefield;

• Communication Model: A wireless communication model provided by the EXata simulation tool [18] was used to simulate PHY/MAC layer performance. EXata offers a number of user-configurable parameters including transmission power, receiver sensitivity, antenna height, and antenna gain. We adjusted these parameters to achieve a variety of desired average transmission ranges at 1 Mbps throughput. Specifically, we calibrated the parameters to produce average transmission ranges from 300 meters to 1500 meters at 100 meter increments;

• Node Mobility: The mobile nodes move under a group mobility model based on the rationale described above, i.e., platoon headquarters as one group and each section as one group. Each group moves randomly within the square field. Each individual node moves randomly as well within its group boundaries. The group and node moving speeds are set from 1 meter/second (m/s) to 10 m/s. A moving speed of over 10 m/s is deemed not practical for dismounted infantry. In accordance with our “Node Distribution” assumption, all nodes operate within a 2 km by 2 km area; for simplicity, a mobile node that reaches the edge of the square area is assumed to “bounce off” the edge and continue its movement in a new random direction;

• Traffic Type: In our simulation, all user traffic sent in the TEN uses the User Datagram Protocol (UDP) as its transport layer. The choice of UDP was to allow for the accurate measurement of delivery ratio;

• Blue Force Tracking SA Messages: Each node generates and broadcasts its locally-generated blue force tracking SA message every 5 seconds. Each message is 64 bytes, which is

Page 16: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

8 DRDC-RDDC-2017-R140

computed as being sufficient to contain a node’s location and other important status information (including any required security overhead);

• Commander’s Order: We consider commander’s order messages generated and broadcast once every 2 minutes. Each message is 1024 bytes. In addition, in this report, the commander’s orders are delivered without guaranteed broadcast, i.e., UDP, in order to evaluate the delivery ratio. In a real battlefield situation, these messages should be delivered with guaranteed broadcast technology;

• Unicast Traffic: We created 10 end-to-end unicast traffic flows among different nodes that persisted for the duration of the simulation. The source and destination of each unicast flow were selected such that we ensured these two nodes did not belong to the same “group.” Unicast traffic flows co-existed with other traffic types (including blue force SA and Commander’s orders). Each unicast traffic flow was set to transmit 1024 bytes per second.

3.3 MATLAB Simulation Scenario

Our MATLAB simulation is used to evaluate how the MPR-reduction capability of the SER-SA protocol compares with OLSR under the different network sizes and radio transmission ranges described in Section 3.1 and 3.2. The MATLAB simulation is run 500 times for each case, i.e., each case with a fixed network size and radio transmission range to obtain an average MPR value as the simulation result. For the same case, each run uses the same network size and radio transmission range but different network topology that is created by randomly placing nodes in the 2 km by 2 km simulation area. This simulation is more efficient to perform with MATLAB and was found to be very time-consuming with the EXata platform.

The performance metrics collected in the MATLAB simulation include total counts for MPR, local MPR, and MPR-reduction percentage, which are defined below.

• Total MPR: The number of all MPR nodes in a network (where MPRs are created by a routing protocol). In general, a larger number of total MPRs in a network will result in more nodes involved in forwarding broadcast traffic, and consequently more node interference, more bandwidth usage, and more nodes consuming their energy;

• Local MPR: Local MPR is related to each local node. It is the number of 1-hop neighbour nodes selected by a local node from its 1-hop neighbour list to serve as that node’s MPRs. Depending upon the MPR selection methodology, typically a 1-hop node selected as a local MPR by one node may not be used as an MPR by another node unless that node also independently (and locally) selects the MPR as its own. In general, the more local MPRs selected, the more potential (and possibly diversified) paths the local node has to reach its destinations, resulting in more reliable traffic delivery;

• MPR-Reduction Percentage: The ratio of removed MPRs (from using SER-SA) to total MPRs generated by the standard greedy heuristic method.

3.4 EXata Simulation Scenario

Our EXata simulation is used to evaluate the traffic delivery ratio and the associated wireless channel utilization of a network using the SER-SA protocol, and to compare these parameters

Page 17: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

DRDC-RDDC-2017-R140 9

with those of a network using OLSR and/or AODV protocols. In addition to the network size and range parameters described in Section 3.1 and 3.2, the EXata simulation setup is described below.

• Simulation Platform: EXata-5.2 integrated with the SER-SA protocol. Note the network healing function of SER-SA described in [5] is not implemented in the simulation and is left as future work. Without the network healing function, the performance of SER-SA under mobility and in lower neighbour-density networks8 is expected to be degraded—as we will see, this performance degradation becomes evident in Section 4.2.3;

• Bandwidth: 1 Mbps data rate with 1 Km radio transmission range;

• Broadcast Feature: The broadcast forwarding function is enabled following the mechanism described in Table 1, where MPR nodes relay broadcast traffic in the OLSR and SER-SA routing implementations, whereas all nodes relay broadcast traffic in the AODV routing implementation:

Modified MCBR: The original EXata multicast constant bit rate (MCBR) function is modified to be able to use broadcast address delivering specific broadcast traffic;

SA and Commander’s Order Traffic: All SA and commander’s order traffic is set as a specific broadcast traffic that can be delivered through the modified MCBR;

500 milliseconds jitter are added to the modified MCBR for SER-SA, OLSR, and AODV performance evaluation.

• Simulation Setting: 10 simulations are run for each case; each simulation uses a different random seed for groups’ and nodes’ placement and runs 30 minutes.

The performance metrics gathered by the EXata simulation include traffic delivery ratio and channel utilization, which are defined below.

• Traffic Delivery Ratio (TDR): The ratio of the total received traffic by each node to the total traffic sent to the node. The greater value of delivery ratio, the better performance of the routing protocol (in delivering the data intended to be delivered). Here Traffic means application data packets, it does not count for routing messages such as TC, HELLO in OLSR and RREQ, RREP in AODV;

• Localized Channel Utilization (LCU): The percentage of time that a node or any of its surrounding nodes within its reception range occupy the shared wireless channel. For instance, consider three nodes (na, nb, and nc) that are all mutual neighbours where each node can hear the other two nodes. If node na transmits unicast data to node nb for 30 seconds out of every minute, then all nodes—including node nc—will compute a channel utilization of 50%. Even though node nc neither transmits nor receives data directly, it can sense that the channel is occupied half the time and thus records a channel utilization of 50%. We denote the localized channel utilization for node ni as LCU(ni).

8 We define the term “neighbour-density” to refer to the average number of 1-hop neighbours per node in the network. A high neighbour-density network would have tightly clustered nodes with many 1-hop neighbours each, while a low neighbour-density network would consist of sparsely-packed nodes having few (if any) 1-hop neighbours. The relative positions and transmission ranges of the nodes directly impact the neighbour-density of the network. All other parameters being equal, a longer transmission range results in an increase in neighbour-density.

Page 18: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

10 DRDC-RDDC-2017-R140

4 Comparison of SER-SA, OLSR, and AODV

In this section, we discuss MATLAB and EXata simulation results based on the scenarios described in Section 3. The MPR performances including total MPR reduction and local MPR increase are simulated and evaluated with MABLAB. The traffic delivery and channel utilization performances are simulated and evaluated with EXata.

4.1 MPR Performance

MPR performance is simulated with MATLAB. The simulation scenarios and evaluation metrics are described in Sections 3.1, 3.2, and 3.3.

4.1.1 Total MPR Performance

Figure 2 depicts the total MPR reduction percentage of SER-SA over the greedy heuristic algorithm (OLSR) under different network sizes and radio transmission ranges.

Figure 2: SER-SA MPR reduction percentage over the greedy heuristic algorithm.

Figure 2 shows that SER-SA can achieve MPR reductions of over 50% for 40-node TEN networks using radio transmission ranges between 650 meters to 1050 meters. We note also from Figure 2 that the MPR reduction percentage is low in the case of large transmission ranges, i.e., the reduction percentage is less than 10% when the transmission range exceeds 1300 meters. This can be explained by realizing that when the transmission range is large, then it is possible to represent nearly the entire network (which is limited to a 2 km by 2 km area) as a 2-hop network. Since the MPR selection performed by the greedy algorithm performs its selection based on 2-hop information, this is essentially equivalent to global network information in a network that is limited to 2 hops (so the greedy algorithm has global information just like SER-SA). The MPR reduction percentage is also smaller in the case of short radio ranges. When the radio range becomes too small, a network with randomly placed nodes will contain clusters of nodes that are

Page 19: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

DRDC-RDDC-2017-R140 11

disconnected from one another. We have observed that these disconnected clusters themselves often contain no more than 2 hops. Thus, within a disconnected cluster, SER-SA makes little improvement since the nodes in a disconnected cluster require no more than 2 hops of information to ultimately have a “global” view of their cluster.

Further to the idea of having disconnected clusters when the radio range is small, we have shown in [15] that a TEN with 40 nodes is likely to contain disconnected clusters (or disconnected nodes) when the node radio transmission range is less than 800 meters. In Figure 3, we zoom into the plot of Figure 2 over radio transmission range of interest for the ISSP project, i.e., 800 meters to 1200 meters. The ISSP project requires 1 km point-to-point range for its dismounted soldier radios as presented in [10]; we add a buffer of 200 meters on either side of this target range since the radio transmission range may change in the real tactical battlefield due to the local physical environment, weather, terrain, etc.

Figure 3: SER-SA MPR reduction percentage over the greedy heuristic algorithm

under the proposed ISSP radio transmission ranges.

Figure 3 shows that SER-SA can reduce by over 30% the total MPRs in the network under most ISSP scenarios, e.g., network size from 20 to 50, radio transmission range from 800 m to 1100 m. The greater the network density has under the same radio transmission range, the greater MPR reduction SER-SA can reach.

4.1.2 Local MPR Performance

Figure 4 depicts the average numbers of local MPRs generated by SER-SA and by the greedy heuristic algorithm in a TEN network with 40 nodes, which is considered as a typical light infantry platoon size as described in Section 3.1.

Page 20: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

12 DRDC-RDDC-2017-R140

Figure 4: The average number of local MPRs generated by SER-SA and the

greedy heuristic algorithm in a TEN network with 40 nodes.

For local MPRs, first we explain the somewhat surprising result of how a reduction in total MPRs could result in an increase in the average number of local MPRs. Essentially, when the greedy algorithm is used for MPR selection, each node locally selects its MPRs and attempts to minimize this number without regard to how other nodes are performing this optimization. The following toy example can help illustrate this effect. Suppose that a node, A, has three 1-hop neighbours (call them B, C, and D), and that A has selected only one of these neighbours as a local MPR (node B). Suppose further that the other two nodes (C and D) are themselves selected as local MPRs by some of their other 1-hop neighbours. In this scenario, the total MPRs in the network is 3; however the local MPRs (from A’s perspective, and from the perspective of the node that chose C and the node that chose D) is only 1 local MPR. If SER-SA identified instead that the network could still be fully connected and covered using only nodes B and C as global (not local) MPRs, then the network now has only 2 total MPRs, but A suddenly has 2 local MPRs (since in SER-SA any MPR is treated as an MPR by all its 1-hop neighbours, whether or not they locally selected it).

For the local MPR values, we observe that SER-SA does not result in a significant increase at lower radio transmission ranges, e.g., less than 700 m, but results in significant increases for moderate ranges and even for very large radio transmission ranges, e.g., even larger than 1600 m. The reason is that at lower radio transmission ranges (consequently resulting a lower neighbour-density network), almost all 1-hop neighbours of a node are chosen as MPRs by the greedy heuristic algorithm (since nodes have very few neighbours to choose from). There is little room to increase the local MPR value with SER-SA in a lower neighbour-density network. However, for the high radio transmission ranges, many edge nodes require MPRs near the centre of the 2 km by 2 km network area to connect to the 2-hop nodes located at the opposite edge of the network. All of these MPRs located in the middle area of the almost fully connected network are available to be local MPRs to any nodes located near the centre of the network (thus increasing the average local MPR count).

Figure 5 depicts the average numbers of total MPRs reduced and local MPRs increased by SER-SA and by the greedy heuristic algorithm under different network sizes and radio transmission ranges.

Page 21: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

DRDC-RDDC-2017-R140 13

Figure 5: The average number of total MPR reduction and local MPR increase

with SER-SA compared to the greedy heuristic algorithm.

4.2 Traffic Delivery and Channel Utilization Performance

We performed simulations in EXata to evaluate the performance of OLSR, AODV, and SER-SA in terms of traffic delivery ratio and under different tactical edge scenarios. The simulation scenarios and evaluation metrics are described in Sections 3.1, 3.2, and 3.4. Table 2 shows the parameters and their related values that were used for our simulation.

Table 2: The parameters and their related values used in our EXata simulation.

Simulation Parameters Value

Jitter for Broadcast Traffic 0, 10, 500 milliseconds (ms)

Node Mobility Speed 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 m/s

Node Transmission Range 300, 400, 500, 600, 700, 800, 900, 1000, 1100, 1200, 1300, 1400, 1500 meters

Network Size 40 nodes

Wireless Channel 802.11b with 1 Mbps bandwidth data rate

Traffic9

Blue Force Tracking SA 64 bytes payload every 5 seconds for each node

Commander’s Orders (CO) 1024 bytes payload every 2 minutes

Unicast 10 parallel end-to-end UDP unicast traffic flows among different nodes; Each unicast flow transmits 1024 bytes payload per second

9 Here only three application data (i.e., BFT SA, CO, Unicast) are considered as traffic. All other routing and beacon related messages are not counted.

Page 22: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

14 DRDC-RDDC-2017-R140

Note that due to the long-running simulation time of EXata sessions, we didn’t consider all combinations of parameters described in Section 4.1 but instead focused on a representative subset.

4.2.1 Broadcast Performance Investigation Under Different Jitter

Our first investigation focused on evaluating the broadcast delivery ratio of OLSR, AODV, and SER-SA as a function of two important factors: jitter value and the number of source nodes (out of a total 40 nodes in a network) that broadcast MCBR traffic. Since a node shares the same wireless channel with its 1-hop neighbours in TEN, adding an appropriate jitter on broadcast traffic is expected to reduce collisions and consequently improve delivery ratio. In addition, the greater the number of source nodes sending broadcast traffic, the more collisions could happen (reducing delivery ratio).

Figure 6 depicts the broadcast delivery ratio of AODV, OLSR, and SER-SA under different jitter values and for different numbers of broadcasting source nodes. AODV_NoJ means that there is no jitter added into the broadcast traffic for the AODV routing protocol, AODV_J10 ms means a random jitter of up to 10 milliseconds (ms) is added to the broadcast traffic, and AODV_J500 ms means a random jitter of up to 500 ms jitter is added to the broadcast traffic. The legends for OLSR and SER-SA in Figure 6 have the same meaning as AODV.

Note the simulation of Figure 6 is based on the following configuration:

• 1 km radio transmission range as described in Section 3.4;

• Each broadcasting node generates and broadcasts a 64 bytes message every 5 seconds;

• No mobility.

Figure 6: The broadcast delivery ratios under different jitter,

number of broadcast nodes, and routing protocols.

To understand the results in Figure 6, we note at least three processes at work in a TEN that directly affect the broadcast delivery ratio:

Page 23: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

DRDC-RDDC-2017-R140 15

1. We expect that a greater number of nodes involved in re-broadcasting traffic will lead to an increase in the likelihood that a broadcast message is delivered, even if some of the nodes do not succeed in forwarding on the traffic.

2. In opposition to the effect described in (1), we expect that more nodes involved in re-broadcasting traffic will also lead to more collisions, which will result in a lower likelihood of a broadcast message being delivered.

3. As the number of source nodes in the network increases, we expect more collisions leading to a lower broadcast delivery ratio for all three routing protocols.

For the three processes described above, the effect of process (3) is apparent in Figure 6 as all curves trend downward as a function of the number of source nodes. The implications of processes (1) and (2) are more complex and are described in detail below.

With respect to process (1), with AODV—which uses pure flooding—all nodes participate in re-broadcasting. In contrast, for OLSR and SER-SA, only the MPR nodes participate in re-broadcasting. Furthermore, as shown in Figure 4, nodes in a network using SER-SA will have on average more local MPRs than nodes in a network using OLSR. In all of these cases (AODV, OLSR, and SER-SA), the number of nodes performing re-broadcasting does not change for the duration of the simulation in this static network. Thus, with respect to process (1), we expect AODV to achieve the best broadcast delivery ratio, followed by SER-SA and finally OLSR.

With respect to process (2), the greater number of re-broadcasting nodes in AODV should lead to a worse delivery performance than either OLSR or SER-SA due to the increased number of collisions. This effect works in direct opposition to the beneficial effect of an increased number of re-broadcasting nodes described for process (1). Jitter works primarily to mitigate the effect of process (2)—that is, the effect of collisions.

In Figure 6, with no jitter, we observe AODV has the highest overall delivery ratio performance (in all cases) followed by SER-SA and then OLSR when the number of broadcasting nodes is over 15. There is little to no difference among the protocols when the number of broadcasting nodes is less than 15. With jitter we observe the opposite effect—where OLSR achieves the best delivery performance, followed by SER-SA and then AODV. It seems reasonable that the effects of process (1) should not change based on the presence or absence of jitter. Therefore one possible explanation for the observed reversal of broadcast delivery performance among the three protocols in the presence of jitter could be that the negative effects of process (2) are mitigated to a much greater extent for a routing protocol that uses fewer re-broadcasting nodes.

In general, in order to reduce collision of broadcast traffic in a TEN, it is recommended that a random jitter be applied. For the remainder of this report, we add a random jitter of up to 500 ms to broadcast traffic in our simulations. We caveat these results (and our explanations) with the fact that our network is quite dense due to the limited area of operations and the large range.

Page 24: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

16 DRDC-RDDC-2017-R140

4.2.2 Routing Performance under Mobility

In this section, we examine the performance of SER-SA, OLSR, and AODV under mobility, 1 Mbps bandwidth channel, and 1 km radio transmission range in the cases of SA combined with Commander’s Order (CO) and unicast traffic.

Figure 7 depicts the average traffic delivery ratios for a network transmitting mixed SA, CO, and unicast traffic under three protocols. Average traffic delivery ratio is computed by considering the traffic delivery ratio (defined in Section 3.4) of all nodes (ni, i=1,…,N) in the network over 10 runs and computing TDRav = (1/10) ∑j [(1/N)∑iTDR(nji)] where j (j=1, …10) is the jth run.

Figure 7: The average delivery ratios of SA, CO, and unicast

traffic disseminated by SER-SA, OLSR, and AODV respectively.

Figure 7 shows that SER-SA has similar average delivery ratio for unicast traffic compared with OLSR and AODV. For unicast, all three protocols approach a high delivery ratio (of over 95%). For SA and CO traffic, SER-SA and OLSR both outperform AODV. We observe that for all three routing protocols, the CO traffic has a lower delivery ratio compared to SA—a contributing factor to this observation is the fact that the infrequent CO messages are considerably longer than the SA messages and thus are more susceptible to packet loss due to collision. Unicast traffic has a higher delivery ratio than both SA and CO messaging; this can be understood by the fact that a unicast message does not cause a “flood” of traffic in the network and hence is not as likely to be the victim of packet loss due to collision—such collisions would occur only if the unicast message were scheduled at the same time as an SA or CO broadcast.

Interestingly, for this specific simulated scenario, the mobility speed does not appear to have a significant impact on the delivery ratio for any of the three traffic types under any of the three studied protocols. Upon reflection, this is not surprising since the communication model used in the simulation was configured such that each node could communicate with an average range of 1 km. With 40 such configured nodes, operating in a 2 km x 2 km area, nodes are almost certain to have numerous connected neighbours at any instant; in fact, we would expect most unicast communication between nodes to be accomplished by point-to-point communications without relying on any routing at all. Consequently, mobility in this scenario is not expected to have a

Page 25: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

DRDC-RDDC-2017-R140 17

significant effect on the routes (and hence on throughput) since most “routes” remain direct for the duration of the simulation.

To validate our belief that mobility would have a stronger effect on the traffic delivery ratio under shorter radio range conditions, we repeated our simulations with new configurations to produce communication links whose average transmission ranges are 500 m. Figure 8 shows the results of a simulation focusing on a network using OLSR under this new scenario.

Figure 8: The average delivery ratios of the SA, CO, and unicast traffic disseminated

by OLSR under 500 meter radio transmission range and different mobility speeds.

We note that mobility speed has a significant impact on unicast delivery ratio. Interestingly, we note that for zero mobility, the delivery ratio is actually worse than at under low mobility (1 m/s). This surprising result can be explained by considering the relatively slow speed of the nodes compared to the frequency of routing updates in OLSR. Nodes that are disconnected upon initial random placement would have no way of ever sending or receiving data in a static (zero mobility) scenario; however, with slow mobility, those nodes would be likely to establish reasonably stable connections to deliver payloads.

From speeds of 1 m/s to 10 m/s, we notice the unicast delivery ratio uniformly decreasing as a function of speed. This can be explained by the fact that at higher speeds, the OLSR protocol does not respond quickly enough such that the routing tables computed by each node do not accurately reflect the true network connections existing in the network (as the nearest neighbours are continually changing).

Page 26: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

18 DRDC-RDDC-2017-R140

Finally, Figure 8 reveals that even at the lower range of 500 m, mobility speeds from 1 m/s to 10 m/s have less of an impact on the broadcast delivery ratio (for both SA and CO messages). This is understood by realizing that broadcast traffic is not dependent upon routes in the network, meaning that an “out of date” route will not have an impact on broadcast delivery performance. The gain in delivery performance from zero mobility to a mobility speed of 1 m/s is observed for broadcast traffic, as it was for unicast traffic. This gain is likely similar to the gain experienced by unicast under slow mobility, which was a result of an increase in likelihood of disconnected nodes moving slowly into connected (and stable) positions. We note that the gain in delivery ratio is significantly larger for CO than for SA traffic. This can be understood by noting that CO traffic originates from only one node in the group (the commander), whereas SA traffic originates from every node. Thus, if the commander node is “isolated,” this results in no CO traffic delivery at all—this is not an issue for multiple node traffic sources such as SA, where there will always be “some” traffic to deliver to “some” non-isolated nodes. Slow mobility would help an isolated commander node to establish connectivity and hence has a greater impact on CO traffic. In Section 4.2.3, this effect is noted across all routing protocols (OLSR, AODV, and SER-SA).

Figure 9 depicts the average channel utilization of SER-SA, OLSR, and AODV for dissemination of mixed SA, CO, and unicast traffic. Average channel utilization is computed by considering the localized channel utilization (defined in Section 3.4) of all nodes (ni, i=1,…,N) in the network under 10 runs and computing CUav = (1/10)∑j[(1/N)∑iLCU(nji)] where j (j=1, …10) is the jth run. SER-SA has the lowest average channel utilization (19%), compared with OLSR at 22% and AODV at 30%. We note that mobility speed appears to have little effect on the average channel utilization for all three protocols. However, a closer look at the distribution of the raw data—as opposed to the first-order statistics—reveals that mobility does indeed have an impact on channel utilization exhibited by the three routing protocols, as shown in Figure 10.

Figure 9: The average channel utilizations of SER-SA, OLSR, and AODV

on disseminating mixed SA, CO, and unicast traffic.

Figure 10 depicts the cumulative distribution function (cdf) of the localized channel utilization (LCU(ni)) for SER-SA, OLSR, and AODV for node speeds of 0 m/s, 5 m/s, and 10 m/s. We observe that for mobile networks (for speeds of both 5 m/s and 10 m/s), the cdf of the localized channel utilization indicates a lower variance, meaning that most nodes have comparable localized channel utilization values; hence, in this situation channel utilization is more

Page 27: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

DRDC-RDDC-2017-R140 19

predictable. However, for static networks (0 m/s) we observe a more gradual increase in the cdf, indicating a larger variance in channel utilization among the nodes. Thus, although Figure 9 indicated no significant impact of mobility on average channel utilization, Figure 10 suggests that mobility can make the channel utilization more predictable.

Figure 10: The cumulative channel utilizations of SER-SA, OLSR, and AODV

on disseminating mixed SA, CO, and unicast traffic.

4.2.3 Routing Performance under Different Radio Transmission Ranges

In Section 4.2.2, we observed that in the particular scenario examined, mobility had little effect on delivery ratio and channel utilization. This was explained to be a result of the fact that the large radio range (and small operating area) ensures that nodes are connected to a large number of 1-hop neighbours. In this section, we examine channel utilization and delivery ratio in the presence of mobility when the radio range is reduced such that nodes may become disconnected. We simulated the same traffic mix as described in Section 4.2.2 under a 1 Mbps bandwidth channel and considered two different mobility conditions—without mobility and medium mobility speed (5 m/s)—under a variety of radio ranges.

4.2.3.1 Baseline Performance—No Mobility

As a baseline, Figure 11 depicts the average delivery ratios for a network transmitting (concurrently) mixed SA, CO, and unicast traffic under different radio transmission ranges and routing protocols without mobility. In general, we would expect that as transmission range increases, the improved connectivity in the network would result in higher delivery ratio. As we see from the figure, however, this is not the case for all routing protocols and all traffic types—one explanation is that the increased transmission range leads to increased packet loss due to “collisions,” especially for broadcast traffic.

For SA broadcast traffic, we see from Figure 11 that AODV produces the best delivery ratio for the SA traffic under low radio ranges, i.e., in a low-neighbour-density scenario, whereas its delivery performance degrades as the radio range increases. This is explained by the fact that collisions have a larger impact in AODV in a high-neighbour-density scenario than they do in

Page 28: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

20 DRDC-RDDC-2017-R140

either OLSR or SER-SA. Both SER-SA and OLSR gradually increase in delivery performance for SA broadcast traffic as the radio range increases, outperforming AODV in the long-range scenario; SER-SA and OLSR do not suffer from collisions in the high-neighbour-density scenario, because unlike AODV they make use of MPRs to reduce the number of nodes forwarding broadcast traffic.

Figure 11: The average delivery ratios of the mixed traffic delivered by SER-SA, OLSR,

and AODV without mobility and under different radio transmission ranges.

For unicast traffic, we observe that all three routing protocols approach a similar delivery performance as radio range increases. This is a reasonable result, since the longer the radio range becomes, the more likely a unicast transmission can be accomplished with a single hop. In low-range scenarios, we see AODV offering a modest improvement over OLSR and SER-SA.

For CO traffic, we observe that both OLSR and AODV exhibit a large gap on their delivery ratios between CO and SA. The reason, as explained in Section 4.2.2, is that the increased CO packet length exposes it to a higher chance of collision; additionally, since the CO messages are infrequent, if a CO message collides with an SA message, it leads to a more significant drop in CO delivery ratio than to SA delivery ratio. We also note that the gap between SA and CO delivery ratios is reduced for OLSR under larger radio transmission ranges; this is reasonable because the number of MPR nodes in the network is reduced as the radio transmission range increases (therefore leading to fewer collisions). SER-SA has a more stable delivery performance across SA and CO traffic (smaller gap between curves) compared with OLSR under all considered radio transmission ranges—again, this is understood to be caused by the smaller number of MPRs in SER-SA, resulting in fewer collisions.

Figure 12 depicts the average channel utilization of SER-SA, OLSR, and AODV for disseminating the mixed traffic under different radio transmission ranges without mobility. Figure 12 shows that SER-SA has significantly lower channel utilization compared with OLSR and AODV under the different radio transmission ranges.

Page 29: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

DRDC-RDDC-2017-R140 21

Figure 12: The average channel utilizations of SER-SA, OLSR, and AODV on disseminating

the mixed traffic without mobility and under different average radio transmission ranges.

To explain the results and the shape of the curves in Figure 12, we note at least two processes at work in TENs:

1. For broadcast traffic, relays are employed to re-broadcast messages throughout the network. All other things being equal in a network, a higher number of relay nodes naturally gives rise to more re-broadcast messages—this leads to a higher channel utilization.

2. For unicast traffic, all other things being equal in a network, a higher number of hops en route between two nodes gives rise to more repeated (forwarded) traffic—this leads to a higher channel utilization.

For AODV, as the radio range increases we expect fewer disconnected nodes in the network. As more nodes become connected, they all serve as relays for broadcast traffic; by process (1), this naturally leads to an increase in channel utilization. We observe such an increase in Figure 12 for AODV up to 800 meters. As far as process (1) is concerned, we expect that once all nodes are connected the curve for AODV should remain flat since all nodes are relays—i.e., the number of relays can no longer increase. However, we observe a gradual decrease in the AODV curve after 800 meters. We suspect that this is a result of process (2), whereby the increasing radio range produces more nodes that are directly connected to one another, meaning that there are fewer hops required for unicast, leading to a lower channel utilization. Due to the traffic mix in our simulation, we suspect that the effects of broadcast are dominant, which is why we observe only a slow decrease in the AODV curve due to process (2).

For OLSR at low radio ranges, we observe a similar behaviour as AODV. We suspect a similar process is at work in these lower radio ranges, such that as more nodes become connected to the network we have more relays—here in the form of MPRs—leading to an increase in the channel utilization by process (1). A sharper decrease in channel utilization (than AODV) is observed for OLSR after 800 meters. We speculate that once there are no more disconnected nodes in the network (roughly around 800 meter range), then as the range further increases, more nodes

Page 30: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

22 DRDC-RDDC-2017-R140

become direct one-hop neighbours of one another. Nodes only require MPRs to reach two-hop neighbours (or further). If a node has more one-hop neighbours and fewer two-hop neighbours, it will require fewer MPRs. By process (1), this reduction in relays will lead to a reduction in channel utilization. Additionally, process (2) further contributes to a downward trend in the OLSR curve beyond 800 meters, since fewer hops are required for unicast communications as more nodes are direct one-hop neighbours.

To understand the SER-SA curve, we recall Figure 2, which shows the expected reduction in the number of MPRs in a network using SER-SA compared to a network using OLSR. Across all radio ranges SER-SA is expected to use fewer MPRs—thus, by process (1), we expect SER-SA to have a lower average channel utilization than OLSR. Furthermore, from Figure 2, the magnitude of the MPR reduction is most significant between 700 to 1000 meters—we speculate that this may explain the increasing difference in average channel utilization between SER-SA and OLSR over that range. As the radio range increases beyond 1100 meters, Figure 2 indicates less of a reduction in MPRs for SER-SA; this phenomenon may explain why the SER-SA and OLSR curves approach one another.

Figure 13 depicts the cdf of the localized channel utilization for SER-SA, OLSR, and AODV disseminating mixed traffic under different radio transmission ranges without mobility. We observe that when the transmission range is higher, the localized channel utilization has lower variance. This means that longer transmission ranges in our scenario result in the channel being utilized more uniformly by the nodes.

Figure 13: The cumulative channel utilizations of SER-SA, OLSR, and AODV on disseminating

the mixed traffic without mobility and under different average radio transmission ranges.

Taken together, Figures 11, 12, and 13 suggest that SER-SA can achieve a similar delivery performance to AODV and OLSR in terms of broadcast efficiency at medium to high network densities, but does so at a lower wireless channel utilization. At lower network densities, AODV appears to perform better in terms of broadcast delivery ratio, but comes at the cost of a modest increase in channel utilization.

Page 31: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

DRDC-RDDC-2017-R140 23

4.2.3.2 Performance—Group Mobility at 5 m/s

To investigate the performance of the three routing protocols under different radio transmission ranges and typical infantry mobility speed (5 m/s), Figures 14, 15, and 16 summarize SA, CO, and unicast traffic delivery ratios respectively, and compare them with the results without mobility.

Figure 14: The average delivery ratios of the SA traffic disseminated by SER-SA, OLSR,

and AODV under different radio transmission ranges and mobility speeds.

Figure 14 shows the average delivery ratios of the three routing protocols for SA traffic under different radio transmission ranges and mobility speeds (0 m/s and 5 m/s). SER-SA has similar delivery ratio when compared with OLSR under medium and long radio transmission ranges (over 800 meters). Both SER-SA and OLSR outperform AODV under medium and long radio transmission ranges. In low-range scenarios, AODV offers an improvement over OLSR and SER-SA. This is similar to the results observed and explained in Section 4.2.3.1. For AODV and OLSR, the mobility speed has little impact on average SA delivery ratio; for SER-SA under lower radio transmission ranges, there is a significant degradation under mobility. This delivery performance degradation is not surprising since the network is not fully connected and the SER-SA implementation in our simulation does not yet include a MPR re-computation and re-dissemination function as per [5].

Page 32: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

24 DRDC-RDDC-2017-R140

Figure 15: The average delivery ratios of the CO traffic disseminated by SER-SA, OLSR,

and AODV under different radio transmission ranges and mobility speeds.

Figure 15 shows the average delivery ratios of the three routing protocols for delivering CO traffic under different radio transmission ranges and mobility speeds. We note that for OLSR there is an improvement in delivery ratio for CO traffic when mobility is present. This improvement is present for AODV only under low radio transmission ranges. For SER-SA, there is no improvement in delivery performance for CO traffic under mobility. However, the degradation in SER-SA delivery performance that was observed for SA traffic under mobility (in Figure 14) is not apparent for CO traffic in Figure 15. These results are consistent with our expectations, based on the results and discussion provided in Section 4.2.2. Specifically, as was observed in Figure 8 for OLSR, we expected mobility to provide an increase in CO delivery ratio at lower radio ranges (recall that Figure 8 showed this for 500 m); we expected this effect to disappear at higher radio ranges as was observed in Figure 7 (which showed no significant effect of mobility for radio ranges of 1000 m).

Figure 16: The average delivery ratios of the unicast traffic disseminated by SER-SA, OLSR,

and AODV under different radio transmission ranges and mobility speeds.

Page 33: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

DRDC-RDDC-2017-R140 25

Figure 16 shows the average delivery ratios of the three routing protocols for delivering unicast traffic under different radio transmission ranges and mobility speeds. For all three protocols, at lower radio transmission ranges, we observe a delivery performance degradation with mobility. This is consistent with the observations in Section 4.2.2.

Once again, SER-SA is observed to have a greater degradation in delivery ratio at low radio transmission ranges with mobility compared to the degradations exhibited by OLSR and AODV. This is not surprising since the network is not fully connected and nodes change their 1-hop neighbours more frequently under these scenarios, and the SER-SA implementation in our simulation does not yet include a routing re-computation and re-dissemination function.

Figure 17 summarizes the average channel utilization of SER-SA, OLSR, and AODV on transmitting the mixed traffic under different radio transmission ranges and mobility speeds (0 m/s and 5 m/s). Figure 17 shows that SER-SA has lower channel utilization for low to moderate radio ranges compared with OLSR and AODV. This is consistent with observations and explanations appearing in Section 4.2.3.1.

Figure 17: The average channel utilizations of SER-SA, OLSR, and AODV on disseminating

the mixed traffic under different radio transmission ranges and mobility speeds.

Figure 18 depicts the cdf of the channel utilization of SER-SA, OLSR, and AODV for disseminating the mixed traffic under different radio transmission ranges and mobility speeds. Further to our observations from Figure 9 (which indicated an increase in speed resulted in a lower variance), we observe here that for lower to medium transmission ranges (cases (a) and (b)), adding mobility results in a lower variance of channel utilization. Additionally, in Figure 13 we observed that higher radio transmission ranges produced a lower variance for channel utilization. This is apparent in our results here as well, where from (a) to (c) the channel utilizations become decreasingly variant, meaning that channel utilization becomes more predictable. We note in Figure 18(c), where the range is 1500 m, adding mobility does not result in a significant change to the channel utilization.

Page 34: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

26 DRDC-RDDC-2017-R140

(a) Under 500 m transmission range. (b) Under 1000 m transmission range.

(c) Under 1500 m transmission range.

Figure 18: The cumulative channel utilizations of SER-SA, OLSR, and AODV on disseminating the mixed traffic under different radio transmission ranges and mobility speeds.

In general, both the higher mobility speed and radio transmission range appear to result in a lower variance in channel utilization in TENs, making the channel utilization more predictable. However, the higher mobility speed may increase average channel utilization and degrade the traffic delivery ratio; additionally, higher radio transmission ranges may require higher power and impose other physical limitations. Further research is required in order to moderate between speed and range to achieve the low and predictable channel utilization for specific use cases. In addition, the non-uniform mobility speed and transmission range of nodes under real world conditions can present additional complexity.

Page 35: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

DRDC-RDDC-2017-R140 27

5 Conclusions

Tactical edge networks (TENs) consist of mobile wireless nodes communicating through multi-hop links, and thus require routing protocols to ensure the continued connectivity of the network. This report presented the results of numerous simulations comparing a newly-proposed protocol—SER-SA—against the well-known OLSR and AODV routing protocols. Our simulations considered the effects of varying neighbour-density and mobility speed, and led to the following high-level insights regarding routing protocols in TENs:

• None of the routing protocols we considered had an overall “best” performance in all considered use cases. In selecting a routing protocol, network engineers must consider the expected tactical use case and its envisioned limits (e.g., network size, number of neighbours, speed, mobility pattern, traffic load, radio transmission range). Additionally, network engineers must also consider which performance metric(s) of a routing protocol is/are most important. In short, there is no “one size fits all” solution.

• SER-SA reduces the number of MPRs in a network compared to OLSR, given the same transmission range. This improvement is less pronounced in low- and high-neighbour-density networks. Because SER-SA MPRs serve all their 1-hop neighbours (as opposed to OLSR, where MPRs serve only a subset of 1-hop neighbours), the average number of MPRs seen by each node in a network using SER-SA is higher than in OLSR, although the total number of MPRs in the network using SER-SA is lower.

• Adding a random timing jitter to the transmission time of broadcast messages in TENs results in an improved delivery ratio for all three routing protocols considered—OLSR, AODV, and SER-SA. Jitter is effective in improving the delivery ratio by reducing the probability of message collision among neighbours. The improvement is more pronounced as the number of source nodes transmitting broadcast data increases.

• Traffic delivery ratio for unicast traffic is reduced as node mobility speed increases in TENs; this is attributed to the discrepancies between locally-known routing information and the ground truth of a changing network topology. Traffic delivery ratio for broadcast traffic is not negatively impacted under mobility; this is because the broadcast algorithms used by the considered routing protocols do not rely on local routing information, but simply deliver broadcast data to whichever neighbours are nearby.

• In low neighbour-density networks, AODV has a better delivery ratio than SER-SA and OLSR at the cost of higher channel utilization. At higher neighbour-densities, both SER-SA and OLSR outperform AODV in terms of delivery ratio. SER-SA and OLSR have comparable delivery ratio performance at higher neighbour-densities, with SER-SA exhibiting a lower channel utilization performance.

• Higher mobility speed and increased radio transmission range both resulted in more predictable (i.e., lower variance) channel utilization in the scenarios under consideration.

This paper has highlighted the complexity involved in identifying an overall high-performing routing protocol in a mobile ad hoc network. There is no obvious “best choice.” Further research is required to better characterize the interplay and trade-off between transmission range, node speed, network density, and routing protocol update frequency. For instance, in a high mobility

Page 36: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

28 DRDC-RDDC-2017-R140

environment, we expect it would be beneficial to use a routing protocol with higher frequency updates or use a radio with a longer range. We believe the trade-offs among these parameters are vital to matching a routing protocol to a TEN scenario. At present, this is a complex open question.

Page 37: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

DRDC-RDDC-2017-R140 29

References

[1] T. Hammel, Farther Out Networking, August 7, 2013. Retrieved from http://www.darpa.mil/WorkArea/DownloadAsset.aspx?id=2147487187, Access date: Dec 16, 2013.

[2] D.P. Dubhashi, O. Hggstrm, L. Orecchia, A. Panconesi, C. Petrioli, and A. Vitaletti, Localized Techniques for Broadcasting in Wireless Sensor Networks, Algorithmica 49(4): 412–446, 2007.

[3] A. Camill, M. Nati, C. Petrioli, M. Rossi, and M. Zorzi, IRIS: Integrated data gathering and interest dissemination system for wireless sensor networks, Ad Hoc Networks 11(2): 654–671, 2013.

[4] J. Crowcroft, M. Segal, and L. Levin, Improved Structures for Data Collection in Wireless Sensor Networks, In Proc. of the 33rd Annual IEEE International Conference on Computer Communications (INFOCOM 2014), Toronto, Canada, April 27–May 2, 2014.

[5] R. Song, J.D. Brown, H. Tang, and M. Salmanian, Secure and Efficient Routing by Leveraging Situational Awareness Messages in Tactical Edge Networks, In Proc. of the International Conference on Military Communications and Information Systems (ICMCIS 2015), Cracow, Poland, May 18–19, 2015.

[6] P. Jacquet, P. Minet, A. Laouiti, L. Viennot, T. Clausen, and C. Adjih, Multicast Optimized Link State Routing, Internet Draft, Retrieved from http://tools.ietf.org/html/draft-jacquet-olsr-molsr-00 in January 7, 2016.

[7] E.M. Royer, S. Barbara, and C.E. Perkins, Multicast Ad hoc On-Demand Distance Vector (AODV) Routing, Internet Draft, Retrieved from https://tools.ietf.org/id/draft-ietf-manet-maodv-00.txt in January 7, 2016.

[8] T. Clausen and P. Jacquet, Optimized Link State Routing Protocol (OLSR), IETF RFC 3626, October 2003.

[9] C. Perkins, E. Belding-Royer, and S. Das, Ad hoc On-Demand Distance Vector (AODV) Routing, IETF RFC 3561, July 2003.

[10] ISSP, The Integrated Soldier System Project. Retrieved from http://www.canadiandefencereview.com/news.php/news/877 in August 15, 2014.

[11] Z. Li, N. Yu, and Z. Deng, NFA: A new algorithm to select MPRs in OLSR, In Proc. of the 4th International Conference on Wireless Communications, Networking and Mobile Computing (WiCOM 2008), Dalian, China, October 12–14, 2008.

Page 38: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

30 DRDC-RDDC-2017-R140

[12] Y. Bai, Y. Liu, and D. Yuan, An Optimized Method for Minimum MPRs Selection Based on Node Density, In Proc. of the 6th International Conference on Wireless Communications, Networking and Mobile Computing (WiCOM 2010), Chengdu, China, September 23–25, 2010.

[13] A. Qayyum, L. Viennot, and A. Laouiti, Multipoint Relaying for Flooding Broadcast Messages in Mobile Wireless Networks, In Proc. of the 35th Annual Hawaii International Conference on System Sciences (HICSS 2002), Hawaii, USA, January 7–10, 2002.

[14] E.W. Dijkstra, A note on two problems in connexion with graphs, Numerische Mathematik 1: 269271. doi:10.1007/BF01386390. 1959.

[15] R. Song, J.D. Brown, P.C. Mason, M. Salmanian, and H. Tang, HMS: Holistic MPR Selection and Network Connectivity for Tactical Edge Networks, In Proc. of the 2015 IEEE Military Communications Conference (MILCOM 2015), Tampa, FL, USA, Oct 26–28, 2015.

[16] J.D. Brown, M. Salmanian, D. Simmelink, H. Tang, and R. Song, Tactical edge cyber command and control (TEC3) concept: A vision for network situational awareness and network command and control at the tactical edge, DRDC-RDDC-2014-R155, Defence Research and Development Canada, Dec 2014.

[17] A.P. Jardosh, K.N. Ramachandran, K.C. Almeroth, and E.M. Belding-Royer, Understanding Congestion in IEEE 802.11b Wireless Networks, In Proc. of the 2005 ACM Internet Measurement Conference (IMC 2005), Berkeley, CA, USA, Oct 19–21, 2005.

[18] EXata Simulation Software. http://web.scalable-networks.com/exata, Access date: May 4, 2016.

Page 39: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

DRDC-RDDC-2017-R140 31

List of Symbols/Abbreviations/Acronyms/Initialisms

AODV Ad Hoc On-Demand Distance Vector Routing

cdf cumulative distribution function

CO Commander’s Order

COP Common Operating Picture

ISSP Integrated Soldier System Project

LCU Localized Channel Utilization

MANET Mobile Ad Hoc Network

MPR Multipoint Relay

OLSR Optimized Link State Routing

SA Situational Awareness

SER-SA Secure and Efficient Routing by Leveraging Situational Awareness Messages

TC Topology Control

TDR Traffic Delivery Ratio

TEN Tactical Edge Mobile Ad Hoc Network

WSN Wireless Sensor Network

Page 40: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

32 DRDC-RDDC-2017-R140

This page intentionally left blank.

Page 41: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

CAN UNCLASSIFIED

CAN UNCLASSIFIED

DOCUMENT CONTROL DATA (Security markings for the title, abstract and indexing annotation must be entered when the document is Classified or Designated)

1. ORIGINATOR (The name and address of the organization preparing the document. Organizations for whom the document was prepared, e.g., Centre sponsoring a contractor's report, or tasking agency, are entered in Section 8.) DRDC – Ottawa Research Centre Defence Research and Development Canada 3701 Carling Avenue Ottawa, Ontario K1A 0Z4 Canada

2a. SECURITY MARKING (Overall security marking of the document including special supplemental markings if applicable.)

CAN UNCLASSIFIED

2b. CONTROLLED GOODS

(NON-CONTROLLED GOODS) DMC A

3. TITLE (The complete document title as indicated on the title page. Its classification should be indicated by the appropriate abbreviation (S, C or U) in parentheses after the title.) On the performance of the SER-SA protocol in tactical edge networks : A comparison among SER-SA, OLSR, and AODV

4. AUTHORS (last name, followed by initials – ranks, titles, etc., not to be used) Song, R.; Li, M.; Brown, J.D.; Salmanian, M.

5. DATE OF PUBLICATION (Month and year of publication of document.) October 2017

6a. NO. OF PAGES (Total containing information, including Annexes, Appendices, etc.)

38

6b. NO. OF REFS (Total cited in document.)

18 7. DESCRIPTIVE NOTES (The category of the document, e.g., technical report, technical note or memorandum. If appropriate, enter the type of report,

e.g., interim, progress, summary, annual or final. Give the inclusive dates when a specific reporting period is covered.) Scientific Report

8. SPONSORING ACTIVITY (The name of the department project office or laboratory sponsoring the research and development – include address.) DRDC – Ottawa Research Centre Defence Research and Development Canada 3701 Carling Avenue Ottawa, Ontario K1A 0Z4 Canada

9a. PROJECT OR GRANT NO. (If appropriate, the applicable research and development project or grant number under which the document was written. Please specify whether project or grant.)

9b. CONTRACT NO. (If appropriate, the applicable number under which the document was written.)

10a. ORIGINATOR’S DOCUMENT NUMBER (The official document number by which the document is identified by the originating activity. This number must be unique to this document.) DRDC-RDDC-2017-R140

10b. OTHER DOCUMENT NO(s). (Any other numbers which may be assigned this document either by the originator or by the sponsor.)

11. DOCUMENT AVAILABILITY (Any limitations on further dissemination of the document, other than those imposed by security classification.)

Unlimited

12. DOCUMENT ANNOUNCEMENT (Any limitation to the bibliographic announcement of this document. This will normally correspond to the Document Availability (11). However, where further distribution (beyond the audience specified in (11) is possible, a wider announcement audience may be selected.)) Unlimited

Page 42: On the performance of the SER-SA protocol in tactical edge … · 2018. 1. 22. · In this report, we present a recently proposed routing strategy (SER-SA) by leveraging SA messages

CAN UNCLASSIFIED

CAN UNCLASSIFIED

13. ABSTRACT (A brief and factual summary of the document. It may also appear elsewhere in the body of the document itself. It is highly desirable that the abstract of classified documents be unclassified. Each paragraph of the abstract shall begin with an indication of the security classification of the information in the paragraph (unless the document itself is unclassified) represented as (S), (C), (R), or (U). It is not necessary to include here abstracts in both official languages unless the text is bilingual.)

This report evaluates the performance of a recently proposed protocol for simultaneously delivering broadcast situational awareness traffic and performing proactive routing in a Tactical Edge Mobile Ad Hoc Network (TEN). The proposed routing strategy—called SER-SA, for secure and efficient routing by leveraging situational awareness messages—takes advantage of the fact that nodes in future tactical networks are expected to periodically broadcast Situational Awareness (SA) messages to other nodes in the network, indicating their position and status.

With simulations examining various scenarios such as different network size, radio transmission range, and mobility patterns, we compare the performance of SER-SA against other popular Mobile Ad Hoc Network (MANET) routing protocols—specifically, Optimized Link State Routing (OLSR) and Ad Hoc On Demand Distance Vector Routing (AODV). When considering channel utilization and traffic delivery ratio as the performance metrics for our study, we observe that there is no “silver bullet,” i.e., there is no single best routing protocol for every considered TEN scenario. This report highlights the cases under which certain protocols offer benefits over others, and when a particular routing protocol is advantageous.

---------------------------------------------------------------------------------------------------------------

Le présent rapport évalue les performances d’un protocole récemment proposé qui permet la distribution simultanée du trafic de connaissance de la situation à destinations multiples et la réalisation du routage proactif dans un réseau mobile ad hoc tactique en périphérie (TEN). La stratégie de routage proposée, appelée SER-SA (acheminement protégé et efficace en exploitant des messages de connaissance de la situation), tire profit du fait que les nœuds dans des réseaux tactiques futurs devraient diffuser périodiquement des messages de connaissances de la situation (CS) aux autres nœuds dans le réseau pour indiquer leur position et leur état.

À l’aide de simulations visant à examiner divers scénarios, comme la taille de différents réseaux, la portée de transmission radioélectrique et les tendances concernant la mobilité, nous comparons les performances de SER-SA aux autres protocoles populaires de routage de réseaux mobiles ad hoc (MANET), en particulier le protocole Optimized Link State Routing (OLSR [routage destiné aux réseaux maillés, sans fil ou mobiles]) et le protocole de routage Ad-hoc On-demand Distance Vector (AODV [vecteur de distance ad hoc sur demande). Lorsque nous examinons le taux d’utilisation des canaux et de distribution du trafic comme paramètres de rendement dans notre étude, nous constatons qu’il n’y a pas de « solution miracle », c.-à-d. qu’il n’y a pas de protocole de routage universel pour tous les scénarios de TEN examinés. Le présent rapport met en évidence les cas où certains protocoles présentent des avantages par rapport aux autres, et ceux où un protocole de routage en particulier est avantageux.

14. KEYWORDS, DESCRIPTORS or IDENTIFIERS (Technically meaningful terms or short phrases that characterize a document and could be helpful in cataloguing the document. They should be selected so that no security classification is required. Identifiers, such as equipment model designation, trade name, military project code name, geographic location may also be included. If possible keywords should be selected from a published thesaurus, e.g., Thesaurus of Engineering and Scientific Terms (TEST) and that thesaurus identified. If it is not possible to select indexing terms which are Unclassified, the classification of each should be indicated as with the title.) Tactical Edge Network; Routing Protocol; Traffic Delivery Ratio; Channel Utilization; Radio Transmission Range; Mobility Speed