Urban Road Usage Analysis Based On Base Station Data

As China’s economy grows smoothly, traffic congestion has become an issue haunting every city. Merely expanding traffic lines is not sufficient and not sustainable. Only grasping the whole road usage, could people raising measures correspondingly. However, because of limitation in coverage, modern traffic flow detection techniques (radar-based detection, magnet-based detection and video-based detection) and GPS based traffic flow detection fail to reveal comprehensive road usage. On the other hand, with unmatched penetration, mobile phones have become sources for massive traffic information gathering. Therefore, with thorough study of existing methods, base station data is used to calculate and analyze Nanjing urban road usage. Main works completed are as follows:1) Base stations are hierarchical clustered to relieve problems existing with current base station coverage estimating method. Besides, OD-matrix is elegantly calculated with SQL statements.2) Road network is constructed from OSM.3) In light of the shortage of modeling car-routing process with Dijkstra, a routing algorithm based on transition probability is proposed.4) To incorporate routing algorithm proposed, a new matching method from area to road network is proposed, based on which road usage as well as its generating sources are calculated.In lack of read road usage data, the result calculated is validated against common sense. Main discoveries are as follows:1) OD-areas can be divided into two clusters, and the ones with more people occurrence at night mainly belong to residential quarters.2) The fluctuations of OD and road usage volume calculated are in accordance with people’s daily commuting rule. Specifically, on weekdays, volumes’ morning peak occurs between 7:30 and 9:00, and evening peak occurs between 17:00 and 19:00, while there is a small volume peak at 12:00 which corresponding to short travelling distance.3) Road usage volume is in positive correlation with its amount of generating sources. And road usage contribution made by sources follows Pareto principle.4) Calculated high-level usage roads have high recall rate for congested road. These roads include Nanjing Yangtze Bridge, Zhongyang Road, Xuanwu Lake tunnel, Yingtian Street viaduct and so on.These comparison and discoveries validate the methods designed as efficient. Finally, with road usage sources computed, potential application in traffic field is proposed, which includes potential congestion recognition, congestion cause analysis and aided decision support for urban traffic planning.