Average speed estimation for road networks based on GPS raw trajectories

For applications involving displacements around cities, planners cannot count on moving at the legal speed limits. Indeed, the amount of circulating vehicles decreases the average speed and consequently increases the estimated time for daily trips. On the other hand, the number of available trajectories generated by GPS devices is growing. This paper presents a methodology to compute statistics about a road network based on GPS-tracked points, generated by vehicles moving around a city. The proposed methodology allows selecting the most representative data to describe how speeds are distributed along the days of week as well as along the time of the day. The results obtained may be used as an alternative to the shortest-path routing criterion for route planning.