Location Prediction of Vehicles in VANETs Using A Kalman Filter

Location information is very important for many applications of vehicular networks such as routing, network management, data dissemination protocols, road congestion, etc. If some reliable prediction is done on vehicle's next move, then resources can be allocated optimally as the vehicle moves around. This would increase the performance of VANETs. A Kalman filter is employed for predicting the vehicle's future location in this paper. We conducted experiments using both real vehicle mobility traces and model-driven traces. We quantitatively compare the prediction performance of a Kalman filter and neural network-based methods. In all traces, the proposed model exhibits superior prediction accuracy than the other prediction schemes.