Analytical modeling and simulation validation of IEEE 802.11 DCF in vehicular-to-infrastructure networks

Abstract

Vehicle-to-infrastructure (V2I) communication over IEEE 802.11 access points (APs) deployed along the roadside must operate under the contention of many vehicles sharing one channel. We investigate the transmission and network performance of vehicles passing through an AP over 802.11 WLANs. We first derive an analytical traffic model that yields the number of vehicles within the transmission range of an AP. Coupling this traffic description with a Markov-chain representation of the backoff process and an M/G/1/K queue for arriving packets, we obtain a model that evaluates DCF performance under an adaptively chosen retransmission limit, and we derive the probability of the mean arrival rate to the AP. A distinctive aspect of this work is that it integrates the vehicular traffic model and the backoff procedure with the M/G/1/K queueing model to examine the impact of traffic load and system parameters on the network. We validate the analytical framework against an independent ns-3 packet-level simulation: the throughput, collision, and transmission-probability predictions agree closely with simulation. Using the validated framework to examine the optimal retransmission number, we find that under realistic finite-buffer conditions the standard 802.11 retransmission limit is already close to optimal for delay and information freshness, so the large gains predicted by an infinite-buffer analysis do not survive packet-level validation. We report this finding explicitly and use it to delimit the operating regime in which retransmission-limit tuning is beneficial.