Analysis of CSMA broadcast performance in tactical ad hoc networks
Publish date: 2016-02-24
Report number: FOI-R--4219--SE
Pages: 37
Written in: English
Keywords:
- CSMA
- OLSR
- ad hoc networks
- multicast
- broadcast
Abstract
This report investigates the possible use of carrier sense multiple access (CSMA) for transmitting broadcast traffic in mobile military ad hoc networks. CSMA is a widely used medium access control (MAC) protocol, most prominently as being part of the distributed coordination function (DCF) in the IEEE 802.11 standard for wireless local are networks (WLAN). Unlike most non-military wireless networks, tactical ad hoc networks focused on here, are often characterized by a majority of the traffic sent being multicast or broadcast. The report describes how the parameter setting and function of CSMA need to be adjusted when used for broadcast traffic rather than unicast traffic and ultimately how this affects the performance of the protocol. An overview of related work at adapting CSMA to better handle broadcast-type traffic is given and the applicability of the proposed methods is discussed. To evaluate the performance of CSMA in tactical ad hoc networks simulations are used. An important factor that determines the broadcast traffic performance in ad hoc networks, aside from the MAC protocol, is the routing protocol and its coupling with the MAC protocol being used. In the simulations, multipoint relay (MPR) flooding is used, which is a commonly used protocol for multicast and broadcast traffic in ad hoc networks and part of the Optimized Link State Routing (OLSR) standard. The results show that when MPR flooding is used with CSMA, robustness problems arise due to the contention-based access that CSMA provides for the control packets needed for MPR-node selection. The result is that under high traffic loads or in sparse networks, MPR-node selection fails due to a high number of collisions, which in turns causes more duplicate packets to be sent and thereby increasing the collisions further. This behaviour results in a very limited network capacity for broadcast-type traffic when CSMA is used, due to high packet losses from collisions. Finally, CSMA is compared to a benchmark system using a fixed time division multiple access (TDMA) protocol instead of CSMA. The comparison shows that the performance of CSMA is inferior to the benchmark system in all cases except under low traffic loads when the network is very well-connected.