Congestion Avoidance Algorithm Using Extended Kalman Filter

Abstract

The prediction of traffic congestion is quite an important issue in vehicle navigation to smoothly control traffic flow, and improve the quality of driver's convenience. However, it is not easy to make accurate predictions since traffic change is highly nonlinear and complex dynamic process. First, we present a new traffic prediction algorithm on the basis of the combined knowledge of both the historical and the real-time traffic information. Based on this traffic prediction result, this paper presents a novel routing technique capable of providing intelligent route services completely adequate to dynamic route guidance systems. In our experiments, we have performed the proposed algorithms on two road networks; one of the complex urban areas and the city. Overall, the results of traffic prediction indicate that our prediction algorithms provide more accurate (nearly 90%) traffic information compared with previous traffic prediction solutions. In addition, our implementation of route determination provides the adaptive routes for traffic conditions, as well as scalable routing services for users' preferences.