A SVM based Routing Scheme in VANETs

In recent years, VANET as the key communication networking technology has been attracted by academia and industries with remarkable development. However, there is still many shortcomings especially in communication efficiency where routing algorithm is one of them. In the design phase of routing algorithms, only very few features of vehicle nodes are considered manually to improve the classic algorithms e.g. GPSR. This paper studies the features of vehicle nodes and drivers. In addition, we introduce machine learning algorithm in particular Support Vector Machine (SVM) to process the vehicle data and generate routing metric to enhance the effect of these features. This paper studies the methods of analyzing and processing probe car data and also discusses the possibility of applying machine learning algorithms in the generation of VANET routing algorithm. The simulation results show better reliability and communication efficiency is achieved.