An Approach to Estimate Traffic Speed Based on Cellular Network Signaling Data on Highways

Traffic speed is one of the most essential parameters representing traffic conditions in intelligent traffic system (ITS). In recent years, there have been several approaches estimating traffic speed based on cellular network signaling data. However, the accuracy of these approaches is unsatisfactory because they have a poor performance in filtering out noisy data and minimizing deviations of traffic speed values´ trend in adjacent time intervals. In this paper, a new approach is proposed to solve the two problems above. The approach filters out noisy data according to educated judgment, and adopts a modified Kalman filter algorithm to minimize the deviations. The performance study on real data sets of Beijing shows that the accuracy of the proposed approach is higher when compared with existing two notable estimation approaches. Further the approach will contribute to developing intelligent navigation systems and pursuing artificial intelligence applications.