The Modeling and Cross-Layer Optimization of 802.11p VANET Unicast

Vehicular ad-hoc networks (VANETs) play an important role in enabling ubiquitous communications and connectivity among vehicles in intelligent transportation systems. Various messages can be transmitted in a VANET to improve road safety and furnish multiple types of application services. Therefore, the evaluation of VANET performance and its optimization should be considered. Previous conventional considerations regarding VANET modeling merely incorporated a general homogeneous road traffic scenario. Furthermore, prior research works primarily focused on the broadcasting performance in VANETs, since the safety beacon packets are transmitted in periodic broadcast. However, the exchange of some important data between vehicles is better accomplished by using unicast instead of broadcast with the retransmission mechanism. On the other hand, in the context of VANET optimization, most conventional schemes required continuous monitoring of the network by measuring the number of neighboring nodes to configure the transmission power or adjusting the transmission rate accordingly. Such constant tracking leads to large transmission overheads and measurement delay. In this paper, we propose a set of 802.11p unicast modeling and optimization methodologies to determine the optimal network parameters without continuously monitoring the vehicles in vicinity. This is accomplished by integrating a stochastic urban traffic model in the analysis then performing a cross-layer optimization for each network node to reduce packet collisions. The optimal transmission range and contention window size at different locations are derived based on the spatio-temporal velocity profile and are made known to entering vehicles. These aid the vehicles to configure their transmission power and rate accordingly upon entering a road segment. We evaluate the proposed system in terms of the network delay and throughput performance. Considerable simulation runs to verify the feasibility of the proposed model and optimization methods. Our results indicate that delay (throughput) is improved by about 53% (120%) on average for homogeneous traffic and 45% (104%) on average for heterogeneous traffic.