Speed Harmonization-Based Perimeter Control With Network Clustering for Large-Scale Urban Systems

Abstract

Traffic congestion results in increased travel times and frequent delays. This paper introduces a novel speed harmonization perimeter controller (SHPC) that integrates variable speed limit control with sliding mode theory and is evaluated using the INTEGRATION microscopic traffic simulator. The proposed controller adopts a speed-based perimeter control strategy by regulating the speed at the gated links of the protected network, rather than modifying traffic signal timings. The developed controller was first applied to a medium-sized grid network inspired by downtown Washington, DC, and compared with two benchmark controllers: a fixed-time plan (FP) controller and a decentralized adaptive traffic signal controller that optimizes phase splits and cycle lengths (PSC). At the entire-network level, SHPC reduced queue length and total delay by 24.7% and 31.2%, respectively, relative to FP, and by 11.7% and 19.8%, respectively, relative to PSC. SHPC also reduced travel time, fuel consumption, and CO2 emissions. At the protected-network level, SHPC further reduced queue length and total delay by 26.1% and 31.0%, respectively, relative to FP, and by 9.9% and 9.8%, respectively, relative to PSC, while also improving the remaining measures of effectiveness (MOEs). The controller was then evaluated on a large-scale downtown Los Angeles network. Due to the size and heterogeneity of this network, a geographical self-organizing map (GeoSOM) was developed as a preprocessing step to identify a homogeneous congested region suitable for perimeter control. At the entire-network level, SHPC reduced queue length and total delay by 15.6% and 14.7%, respectively, relative to FP, and by 5.4% and 4.7%, respectively, relative to PSC. Similar reductions were achieved in the remaining MOEs. At the protected-network level, SHPC reduced queue length and total delay by 9.4% and 4.1%, respectively, relative to FP, and by 4.5% and 5.1%, respectively, relative to PSC, with corresponding improvements in the remaining MOEs. These results demonstrate that the proposed SHPC framework can be effectively scaled to large-scale urban networks while improving mobility, mitigating traffic congestion, and providing environmental benefits.