iTrip: Traffic signal prediction using smartphone based community sensing

Providing drivers with traffic signal scheduling information in advance can enable many novel applications, such as optimal speed advisory and shortest trip planning. Existing solutions employ either infrastructure (e.g. wireless transmitter) or vision (e.g. cameras) based approaches. However, these solutions may be limited by high infrastructure cost or low air visibility. In this paper, we propose iTrip, a novel community sensing service that only utilizes smartphone accelerometer to detect and predict accurate traffic signal schedules. In iTrip, on-vehicle smartphones detect and report vehicle´s events, such as start and stop moving, to the server. Using the collected data contributed by a group of vehicles, iTrip can predict the traffic signal in near future by estimating the traffic signal schedule. We conduct extensive simulation under different traffic scenarios. Results show our proposed method is able to efficiently estimate the schedule with accuracy less than 1 second in a few signal cycles.