Traffic Collision Avoidance with Vehicular Edge Computing

Traffic accidents are on the rise due to a multitude of reasons including the increasing number of vehicles on the roads. Therefore, novel mechanisms to predict traffic accidents and issue timely warnings are of great interest to academia and industry. So far, the computational and communication infrastructures were not able to provide fast enough services to implement real-time traffic collision avoidance and warning systems in a collaborative manner. Most of the conventional systems are inbuilt into vehicles and operate with the limited information acquired from their own sensors. However, the latest edge computing technologies when paired with ultra-reliable low-latency communications in 5G/6G networks can provide suitable computational power and latency levels for effective collision avoidance mechanisms in a collaborative setting. In that regard, this paper presents a vehicular edge computing architecture that collects the positional and dynamic information of vehicles of a given area and predicts possible collisions well in advance. The system is capable of sending warning signals to vehicles in danger using the low latency 5G/6G links and avoiding possible collisions. The simulation results prove that the 5G/6G links are capable of supporting less than 10 ms latency for vehicular edge computing even with a high number of vehicles in the area. Further, the collision prediction algorithm is tested with real-life data sets and it demonstrates high accuracy in predicting imminent traffic collisions.