Traffic signal phases' estimation by floating car data

The paper focuses on the estimation of traffic signal phases by floating car data. Given a signalized intersection, by approximating and analyzing the velocity-time curves of all vehicles passing by this signal, not only cycle length but also green time of each movement is calculated respectively. First, because of lack of passing by vehicles, green time of corresponding movement cannot be estimated with high precision, especially for left-turn movement. On the other hand, green time of each movement is estimated separately so that the integral signal logic is not satisfied for any intersection. In this paper, offset between Green Beginning Time (GBT) of through and Green Beginning Time of left-turn (OTLT) is estimated first, which is useful to determine the actual phase plan later. Then, a novel method is presented to adjust all these estimations so that the integral signal logic is satisfied. Experimental results prove that the estimation of traffic signal phases is efficient and reliable.