Detection of traffic jams using T-flock patterns

The widespread use of GPS devices on cars enables the collection of time-dependent positions of vehicles and, hence, of their movements on the road network. It is possible to analyze such huge collection of data to look for critical situation on the traffic flow. The offline analysis of traffic congestions represents a challenging task for urban mobility managers. This kind of analysis can be used by the traffic planner to predict future areas of traffic congestions, or to improve the accessibility to specific attraction points in a city. Many traffic systems adopt ad-hoc sensors like cameras, induction loops, magnetic sensors to monitor the status of the traffic flows: these systems are very expensive for installation and maintenance, and they are restricted to the local monitoring of the road arcs where they are installed. On the contrary, the use of GPS data to check the traffic conditions requires low installation costs (a part for the installation on the vehicle) and it enables to virtually monitoring the entire road network. In this demo we present an innovative tool that exploits the data collected from GPS- enabled cars to detect the occurrences of traffic jams on the road network. The detection of potential traffic jams is based on the discovery of slowly moving flock patterns, i.e. a set of objects slowly moving together for a minimum amount of time. The tool has been integrated in the M-Atlas system exploiting the implementation of the T-Flock algorithm provided by the system.