Decentralized traffic rerouting using minimalist communications

Vehicular ad-hoc networks have the potential to greatly decrease travel times while increasing traffic safety. Potential applications range from simple turn assist to complex traffic management via integration with city infrastructure like traffic lights. In this paper, inspired by observed behaviors of certain insects, we propose a decentralized congestion avoidance scheme. The approach measures the level of congestion in a road transportation network by the amount of wireless network traffic generated by vehicle-to-vehicle communications. When experiencing high congestion, a vehicle re-routes non-deterministically using a modified K shortest path algorithm whose paths are weighted using a Logit model. To illustrate the workings of the proposed solution, we use a microscopic vehicular traffic simulator coupled with a communication network simulator. The resulting decentralized congestion avoidance scheme requires only a minute amount of bandwidth and it can be implemented on top of existing vehicle-to-vehicle communication applications by both autonomous and nonautonomous vehicles alike. Significant improvements were obtained in traffic throughput with the proposed approach.