Exploring geometric, kinematic and behavioral scalability of microscopic traffic simulation systems

Abstract

Even with today's remarkable advancement in computing power, microscopic simulation modeling remains a computationally intensive process that imposes limitations on its potential use for modeling large-scale transportation networks. Research and practice have repeatedly demonstrated that microscopic simulation runs can be excessively time-consuming, depending on the network size, the number of simulated entities (vehicles), and the computational resources available. While microscopic features of a simulated system collectively define the overall system characteristics, it is argued that the microscopic simulation process itself is not necessarily free of redundancy, which if reduced, could substantially improve the computational efficiency of simulation systems without compromising the overall integrity of the simulation process. This research study explores the concept of scalability for microscopic traffic simulation systems in order to improve their computational efficiency and cost-effectiveness. More specifically, we present an optimized downsampling procedure for transforming the full-scale simulation system (prototype) into a geometrically, kinematically, and behaviorally equivalent reduced-scale system (microcosm). The ultimate goal is to execute the microscopic simulation process in the microcosm environment, observe all necessary macroscopic characteristics and performance measures, and upsample the results back to the prototype environment. Experimental analysis was conducted on a homogeneous freeway corridor to examine the effect of different operating conditions on the optimal solutions for the downsampling procedure. The study also investigates the tradeoff between performance and scalability of microscopic simulation systems.