Parallel microscopic simulation of metropolitan-scale traffic

Abstract

We present a scalable and high-performance microscopic simulator capable of simulating metropolitan-scale traffic. We analyze the requirements for a realistic level of detail and present techniques for optimizing the performance of the simulator and maintain coherence when scaling into parallel computing architectures. To demonstrate its capabilities, we simulate the San Francisco Bay Area road network. Finally, we provide a performance comparison with other free/open-source simulators and show that our simulator not only outperforms existing solutions but also scales well when running in multi-core environments.