A predictive collision detection protocol using vehicular networks

Vehicular traffic accidents are a crucial problem in big urban centers. Nearly one million people die in road crashes each year. In this paper, we propose a predictive vehicular collision detection protocol for VANETs. The proposed protocol takes advantage of the vehicle-to-vehicle communication in VANETs to predict potential collisions with vehicles in the proximity. Simulation results show that our proposed protocol achieves 92% accuracy in detecting collisions using real case Ottawa urban scenarios and several accidents scenarios. Also, applying a prediction model to estimate future trajectories of nearby vehicles, has significantly reduced the overhead of transmitted packets.