Estimating Real-time Traffic Variables On Freeway Based On Mobile Phone Locating And Clustering

The estimating real-time traffic variables on freeway is the base of traffic flow guidance, the prediction of route travel time, the detection of traffic incident, the analysis of road safety, and the analysis of traffic capacity and performance. It plays important roles in the modern traffic supervision and management system, so it is significant to study the methods of estimating real-time traffic variables on freeway. At present, it is rather difficult to estimate real-time traffic variables using the traditional methods, which get detecting information from detector that is sited in the road section. Then traffic variables are calculated from that information. Because the cost price of installation and maintenance of detecting equipment is too high cost to be used those detecting methods for detecting traffic information that is the whole day, low cost and high efficiency in the wider range. In this paper, a new method of estimating real-time traffic variables based on mobile phone locating is presented. The state of vehicles on freeway is estimated by clustering and analyzing location, velocity and acceleration of mobile phones. Then traffic variables are calculated from the state of separate vehicles. It can estimate traffic variables in the wider range, low cost and high efficiency. This paper introduces several existing methods of estimating traffic variables and discusses their advantage and disadvantage. And this paper brings forward a new scheme of estimating traffic variables: identifying mobile on the vehicle based on the map matching, identifying of independent vehicle and estimating the number of independent vehicle, estimating traffic variables after summarizing the character of running of vehicle, the character of mobile locating data and rule of running of cell phone. Then the theory and implement of clustering are expatiated and the new method of clustering applied to estimate traffic variables is narrated in detail. Finally, the paper compares the estimating performance of mobile phone locating and clustering with that of R&F theory via a great deal of simulation experiment. The experiment results prove that the mobile phone locating and clustering can estimate the traffic variables in real time well and the adaptability and accuracy of mobile phone locating and clustering are better than that of R&F theory.