Predictive Contention Window-Based Broadcast Collision Mitigation Strategy for VANET

In vehicular ad-hoc network (VANET), safety applications require the dissemination of safety information and vehicular states to all nearby vehicles through broadcasting. However, due to the tremendous information, broadcast channel will be confronted with great pressure. Therefore, vehicles should adjust their contention window to avoid intense competition on channel. In this paper, we propose a new contention window adjustment mechanism based on classification strategy and prediction scheme. The classification strategy availably classifies vehicular states into attribute sets by taking account of multiple factors. With this processing, we can obtain contention window corresponding to the attribute sets by BPR (Bayesian Personalized Ranking) algorithm. Subsequently, we adopt the prediction scheme to obtain vehicular states at next moment which map directly to the attribute sets. Finally, the prediction result is employed to explore the contention window at next moment according to retrieving attribute sets. The simulation and analysis show that our scheme provides outstanding performance compared with other classical single factor schemes in term of reducing collision probability and transmission delay.