Font Size: a A A

Optimization of Traffic Signal Timing by Local Computation

Posted on:2012-09-05Degree:Ph.DType:Dissertation
University:George Mason UniversityCandidate:Nagle, DanielFull Text:PDF
GTID:1452390011457574Subject:Engineering
Abstract/Summary:
Traffic congestion is a large problem, with costs including public safety, economic, and psychological. One way to address traffic congestion is by improving the timing of the traffic signals. Most currently installed traffic signals can detect the passage of vehicles, but they make minimal use of this information. Many schemes have been proposed to make better use of this locally-gathered information. But concerns have been raised that local optimization may lead to traffic flows that are globally suboptimal, perhaps even worse than current conditions. One difficulty is that traffic simulators either lack sufficient detail to provide insight into synchronization of traffic signals, or they lack sufficient computational performance to cover a large area. Here, a simulator with sufficient detail to give insight into traffic signal optimization schemes, and sufficient performance to simulate a large grid of traffic signals, is discussed. The use of this simulator shows that traffic signals can, in fact, gather detailed information about the traffic passing through each traffic signal. The traffic signal may then use that information to improve the flow of traffic through itself, and do so while also improving the global flow of traffic. One key idea to emerge is that of tiling traffic signals within an idealized regular grid of intersections executing an idealized synchronous switching plan. Only traffic signals within a preset subset, determined by their position within the tiling pattern, are allowed to decide how to change. Over a range of conditions, this use of technique can give substantial improvements in traffic flow. No new traffic sensors or traffic signal communications is required.
Keywords/Search Tags:Traffic, Optimization
Related items