Traffic signal optimization for oversaturated urban roadway networks

The goal of this research is to develop a signal optimization algorithm that can equalize queue growth rates across links in oversaturated roadway networks, thereby postponing queue spillbacks that form at the localized sections of roadway networks.

Article  |  Fall 2015

Traffic in urban areas has rapidly increased over the past decades while the space for constructing roadway infrastructure has decreased. As a result, urban roadway networks have become increasingly congested. Generally, with urban congestion, queues form at a few oversaturated intersections and spill over into the adjacent links. This queue spill-over triggers urban gridlock that eventually restricts all traffic movements in the entire network.

In this condition, traffic signal settings have a significant impact on the activation and evolution of spill-overs because they determine queue growth rates in the links. Thus, it is critical to properly operate traffic signals at each intersection to postpone spill-over or even to prevent gridlock.

Since the inception of traffic signals, extensive research efforts have been devoted to operate traffic signals more efficiently. However, most algorithms targeted undersaturated intersections, in which links have only endogenous queues; neither queue spill-over nor gridlock is anticipated in such links. In oversaturated conditions, on the other hand, queues that form in other intersections may affect traffic conditions in distant links via queue spill-overs. Therefore, no closed-form expression can be formulated for network delay and throughput as functions of signal parameters because the state of one intersection can be affected by the traffic signal settings at other intersections.

The present study develops a novel method for traffic signal optimization under oversaturated demand. This method defines the degree of queue growth equalization (QGE) in a given network as a new measure of effectiveness and then minimizes the degree of QGE to balance queue growth rates (i.e. the difference between discharging capacity and traffic demand) over the network. If traffic signals and cycle times are optimized using the degree of QGE, wider, smoother transitions of queue growth in a network can be achieved even under intense, localized oversaturated conditions, as shown in Fig. 1. This can help vehicles gradually and uniformly accumulate at the network level and thereby postpone spill-overs in the given network. As a result, the total outflow from the network can remain high for a longer duration, which implies that we can approach the network-wide system optimum. Simulation-based experiments demonstrate that the traffic signal settings optimized by QGE could successfully distribute queues across the underused links and that the proposed method is superior to conventional methods in terms of traffic performance measures and computational cost.

An extended method is currently being developed in order to reflect real-world conditions such as complicated network, varying cycle time, and pedestrian crossing (Fig. 2). This further study will open the door to new ways of mitigating traffic congestion in complicated, large-scale urban networks.