Context Aware Accidents Prediction and Prevention System for VANET

Worldwide, traffic accidents cause over a million fatalities every year. Thus, improving road safety and saving people's lives is an international priority. One major challenge faced by researchers is to design an ideal system that is able to predict road accidents and implement efficient prevention actions. Context-aware systems are those systems that are able to sense, reason and react upon the current contextual information. Utilising those systems in intelligent transportation systems (ITS) might improve road safety and enhance traffic efficiency. This paper introduces a context-aware accidents prediction and prevention system taking into account the most contributory factors that cause road accidents including factors related to the driver, the environment, the vehicle and other vehicles on the road. A context-aware architecture based on VANET's On Board Unit (OBU) is presented. The architecture is divided into three phases: physical phase, thinking phase and application phase, which represent the three main subsystems of context-aware systems: the sensing, the reasoning and the acting subsystem respectively. In the thinking phase, a Dynamic Bayesian Networks (DBN) model has been proposed to predict the accident likelihood and the severity level. The evaluation of the proposed system showed good results in predicting accidents and their levels of severity.