Optimal Averaging Time for Predicting Traffic Velocity Using Floating Car Data Technique for Advanced Traveler Information System

Many metropolitan cities are facing the problem of traffic congestion in large scale and high frequency. The congestion can be lessened by employing Intelligent Transportation Systems (ITS) including Advanced Traveler Information Systems (ATIS). ITS systems have been demonstrated and implemented in a few advanced countries. For example, the technology of Electronic Toll Collection (ETC) in Japan has completely eliminated the traffic congestion ahead of toll gates and has reduced CO2 emission by 130000 tons-per-year. Recently, an ATIS system, the so called Floating Car Data (FCD) technique, has received much attention due to its cost effectiveness and wide coverage in comparison to traditional systems in providing real-time traffic information. The success of the technique is made possible with the existing and vast coverage of the wireless network and information technology. In regard to FCD technique, although many publications have discussed various issues, none has elaborated the traffic data discrepancy between that provided by the FCD technique and the actual traffic data. This paper discusses the issue and demonstrates that there is an optimum averaging time interval in the FCD technique such that the data recorded by a probe vehicle can reasonably predict the traffic data. The analysis is based on experimental data recorded by Sugiyama et al. (2008) where 22 vehicles were deployed to establish a platoon of vehicles moving in a circular road having 230 m perimeter length. Various averaging time intervals are studied, and the one that provides the best estimate of the traffic flow is selected as the optimum averaging time interval.