A Fluid Dynamic Approach for Traffic Forecast from Mobile Sensor Data

In this paper we propose an algorithm for vehicular traffic forecast. The mathematical model is developed to exploit a set of experimental data, which comes from a large number of mobile sensors located on cars. Data refers to traffic flow in a urban highway in Rome, Italy. The model is based on a fluid dynamic approach and is able to make previsions up to 80 minutes with good accuracy in free, congested and unsteady conditions. Different kinds of initializations and algorithms are discussed, then forecast is compared with the exact solution. Traffic data were provided by Octo Telematics(c) SpA.