Application Of Multi Sensor Cooperative Detection Method In Automotive Anti-collision Alarm

With the development of the country’s urbanization process, the number of vehicles is increasing rapidly, the complex traffic network makes frequent traffic accident collision. Some researchers in the field of transportation hope to solve the traffic problems with the help of the information technology, and they put forward the concept of intelligent transportation system(ITS). As an important part of ITS, anti collision system could significantly reduce the probability of collision and severity of rear-end collision.In this thesis, we propose an anti collision scheme which based on wireless sensor network and signal detection technology. The scheme is composed of two parts, one is the rear-end collision model, the other is signal detection model. In the scheme, when the vehicle suddenly slows down, it will send collision alarm signal to the vehicle behind it. The signal detection model is responsible for detecting collision alarm signal, when the signal detected, the model will remind the driver that the vehicle in front of it has slow down. The rear-end collision model is responsible for measuring actual distance between two vehicles, if the actual distance is less than the safe distance then to remind the driver to immediately slow down.In order to detect the rear-end collision warning signal efficiently, we propose three signal detection model by using the temporal correlation and spatial correlation of sensor node sampling information, they are temporal correlation signal detection model, spatial correlation signal detection model and spatio-temporal correlation signal detection model. The simulation results show that the detection performance of spatio-temporal correlation signal detection model is the best. The frequency domain GLRT(Generalized Likelihood Ratio Test) detector under appropriate SNR conditions can ensure a high probability to detect collision alarm signal and then ensure the performance of the rear-end collision detection scheme. The rear-end collision detection scheme proposed in this thesis overcomes the problem of delay in the traditional vehicle network, and avoid the problem of the blind zone of the anti collision system based on radar ranging. Finally, we use data on the rear-end collision simulation program, prove the rear-end collision detection scheme proposed in this paper is feasible.