Travel Prediction-based Data Forwarding for Sparse Vehicular Networks

Abstract

Vehicular Ad Hoc Networks (VANETs) represent promising technologies of cyber-physical systems for improving driving safety and communication mobility. Due to the highly dynamic driving patterns of vehicles, effective packet forwarding, especially for time sensitive data, has been a challenging research problem. Previous works forward data packets mostly utilizing statistical information about road network traffic, which becomes much less accurate when vehicles travel in sparse network as highly dynamic traffic introduces large variance for these statistics.With the popularity of on-board GPS navigation systems, individual vehicle trajectories become available and can be utilized for removing the uncertainty in road traffic statistics and improve the performance of the data forwarding in VANETs. In this paper, we propose Travel Prediction based Data-forwarding (TPD), in which vehicles share their trajectory information to achieve the low delay and high reliability of data delivery in multi-hop carry-and-forward environments. The driven idea is to construct a vehicle encounter graph based on pair-wise encounter probabilities, derived from shared trajectory information. With the encounter graph available, TPD optimizes delivery delay under a specific delivery ratio threshold, and the data forwarding rule is that a vehicle carrying packets always selects the next packet-carrier that can provide the best forwarding performance within the communication range. Through extensive simulations we demonstrate that TPD significantly outperforms existing schemes of TBD and VADD with more than 5% more packets delivery while reducing more than 40% delivery delay.