Urban Road Travel Speed Based On Phone Location Data Extraction

Extracting traffic information from mobile phone location data is a hot research in recent years, compared with the fixed detectors and GPS floating cars, the mobile phone location data which based on GSM cellular network, it doesn't need to cost much money for installation and maintenance, doesn't need to upgrade the terminals. Its investment is small, but the collected data is massive, it can collect the location data all day long, and it can collect the traffic information all over the road, not just limited to the intended roads. The cellular mobile phone location is a wireless navigation and positioning technology which does not dependent on the GPS, it has broad research and application prospects. However, it is a challenge for map matching and path searching, not only for the reason that there are noise mixed in the data, such as the ping-pong switching data and the stationary data, but also the data's sampling interval is large and random, the data's accuracy is low in most of time. Therefore, how to extract the useful traffic information from the massive mobile phone location data, such as the travel speed, is the current problem need to be resolved.This paper reviews the research significance and research results using the mobile phone location data to extract the traffic information at home and abroad, introduces the related theories, such as GSM wireless communication network, wireless position technology, intelligent transportation system, OD trip characteristics of the residents, spatial data mining, clustering analysis, path search algorithm and so on. We use a city's several users'mobile phone location data as the sample, combined with the GSM base station data and the traffic road network dataset. First of all, we preprocess the location data, create the Thiessen polygons to express each base station's signal coverage, and then use the cluster method of spatial neighborhood relationship to eliminate the no-movement data reasonably. The result shows that the method is effective. And then, use several mobile phone's history location data which have regular commuting trajectory for our study, this paper proposed a method using the sample statistics of large probability, extract the location cells which have large sampling rate, assign the property of the road links according to the superposition of the weighted factors, combined with the improved high probability optimal path algorithm(IHPOPA), reconstruct the trajectory and calculate the average travel speed, compared with the travel speed which calculated by the GPS floating cars of the same road link at the same time. The experiment result shows that:the relative error is below 15%.Finally, the author summarizes and prospects the full paper's studies and research results, puts forward the problems which needed to be solved in the future.