Performance analysis of CSMA-based half duplex and full duplex MAC protocols for vehicular networks

Abstract

Beaconing mechanism used in vehicular networks is an essential mechanism to improve the cooperative awareness between vehicles by periodically broadcast-ing vehicle kinematic information and other traÿc entities. Cooperative aware-ness between vehicles a˙ects the performance of safety applications in vehicular networks. Therefore, the Quality of Service (QoS) provided by the beaconing mechanism determines whether the strict timeliness and reliability requirements of safety applications are met. The most important state parameter of the bea-coning mechanism is the beacon generation rate. Even if increasing it increases the cooperative awareness between vehicles, it also increases the channel conges-tion. This trade-o˙ poses a challenge in the MAC protocol design of the beaconing mechanism. Moreover, the broadcast nature of beacon packets prevents the usage of acknowledgement mechanism in the beaconing mechanism, further complicat-ing the problem. In this thesis, we examine the performance of two CSMA-based MAC protocols used for the beaconing mechanism. The first one makes use of Half-Duplex radios (HD-MAC), and it does not support collision detection, whereas the second one makes use of Full-Duplex radios (FD-MAC), and it sup-ports collision detection and retransmission. These protocols are analyzed using Markov Chain based queueing models, taking queue state and beacon generation rate into account. FD-MAC model considers missed and false collision detection occuring due to non-ideal behavior of the self-interference cancellation. Using the Markov Chain models, QoS measures such as successful transmission probabil-ity, throughput, service delay, packet loss probability and age of information are derived. The accuracy of the Markov Chain based models are validated through simulations. It is shown that the FD-MAC protocol is superior to HD-MAC protocol in terms of throughput, packet loss probability, and age of information.