A comparative assessment of multi-sensor data fusion techniques for freeway traffic speed estimation using microsimulation modeling

Real-time traffic speed estimation is a fundamental task for urban traffic management centers and is often a critical element of Intelligent Transportation Systems (ITS). For this purpose, various sensors are used to collect traffic information. For many applications, the information provided by individual sensors is incomplete, inaccurate and/or unreliable. Therefore, a fusion based estimate provides a more effective approach towards traffic speed estimation. In this paper, seven multi-sensor data fusion-based estimation techniques are investigated. All methods are implemented and compared in terms of their ability to fuse data from loop detectors and probe vehicles to accurately estimate freeway traffic speed. For the purposes of a rigorous comparison, data are generated from a microsimulation model of a major freeway in the Greater Toronto Area (GTA). The microsimulation model includes loop detectors and a newly implemented traffic monitoring system that detects Bluetooth-enabled devices traveling past roadside Bluetooth receivers, allowing for an automated method of probe vehicle data collection. To establish the true traffic speed that each fusion method attempts to estimate, all vehicles in the microsimulation model are equipped with GPS devices. Results show that most data fusion techniques improve accuracy over single sensor approaches. Furthermore, the analysis shows that the improvement by data fusion depends on the technique, the number of probe vehicles, and the traffic conditions.

► Several multi-sensor data fusion-based estimation techniques are investigated.
► Compared in terms of their ability to accurately estimate freeway traffic speed.
► Most investigated techniques improve accuracy over single source approaches.
► Improvement depends on technique, number of probe vehicles, and traffic state.