Delay analysis of signal control policies for an isolated intersection

In this paper, we analyze the delay performance of three traffic signal control algorithms that can attain maximum system throughput, for a single four-way intersection. The first is a static (state-independent) fixed-time scheduling (FTS) policy which is followed in current practice. The second is a dynamic maximum weight (backlog) scheduling (MWS) policy that determines the traffic signal phase (the set of collision-free lanes chosen to be scheduled) based on the backlogs (number of queued vehicles) in the different lanes of the intersection, but the schedule length (green time) is kept fixed. The third is adaptive length MWS (aMWS), a variant of MWS in which the schedule length (green time of the phase) is also adjusted based on the backlogs, in addition to the phases. We show that while the dynamic signal control policies (MWS and aMWS) outperform the static (FTS) policy in general, the performance of a fixed-length schedule like MWS can be seriously affected when loss of scheduling time due to “lag effects” is considered. However, with adaptive adjustment of the schedule lengths (green times), a dynamic signal control policy (like aMWS) can outperform the FTS policy at all traffic loads.