Evaluating the Performance of Link Travel Time Estimation Based on Floating Car Data

Travel time estimation is one of the core problems in Intelligent Transportation System (ITS) and Floating Car Data (FCD) is an important data source used by many cities to estimate it. This paper utilizes the Next Generation Simulation (NGSIM) project dataset to simulate the FCD scenario, to test the accuracy of FCD based link travel time estimation. In the first part, the calculated time from simulated FCD, which is referred as “estimated time” and the average travel time of selected floating cars, which is referred as “sample time”, is compared. In the second part, the “sample time”, and average travel time of all cars on the target link, which is referred as “real time” is distinguished, and a binomial distribution model is proposed to explain the situation. Finally, the robust of time estimation to exceptional cars is discussed.