Traffic information is gained from a lot of sources. The most common type of traffic sensors is stationary devices (cameras, loops). Commercial or public sensor systems are usually restricted to the highway network and a number of main urban roads. Stationary sensors provide only local information. Another approach is floating car data (FCD). In an active FCD system the position of each vehicle is observed by a GPS device or another positioning system. Passive FCD systems work with stationary devices where vehicles can be identified with the help of onboard transponders communicating with stationary devices on subsequent positions on a road. The coverage of FCD systems is limited by the expenses for the data transfers between the vehicles and a central server. The general question for all methods is how to improve knowledge the about traffic situation in order to get traffic information as close as possible to an ideal system: area wide, all time available and able to get prediction about traffic situation on short time scales, and, finally, cost-effective. An idea to get travel information with high coverage but without using extra infrastructure is using mobile phones as traffic sensors. They communicate with the next available base station. The stations form a network of cells. The time when a phone is crossing a cell boundary can be recorded and conclusions on the motion can be drawn from this. In this paper the performance of mobile phone GSM data for traffic information is discussed. It is validated how effectively the traffic speed and flow can be measured. For this purpose the data are compared with data from traffic loops and with classical FCD from a taxi fleet. For the covering abstract see ITRD E134653.
Abstract