Recognition of Traffic Congestion Based on Mobile Phone Sensor Data

Accurate recognition of traffic congestion is the basis of intelligent transportation system. This study develops a new method to evaluate the running environment surround the target vehicle based on two parameters:the real-time acceleration and angular acceleration collected from the smart phone. After analyzing the statistical characteristics of the parameters, the data is found to have a close relationship with the corresponding traffic flow states. Support vector machine and parametric optimization method are utilized to test these data. The experiments show a highest accuracy of 92%, indicating that acceleration and angular acceleration could be considered together as the characterization parameters of traffic flow states. To reduce the computation cost while maintaining the accuracy of the traffic state identification, strong explanatory variables of the support vector machine algorithm are recognized by the relative optimal regression model and the least angle regression method.