Multi-modal traffic simulation platform on parallel and distributed systems

Abstract

In this paper we describe a highly scalable multi-modal traffic simulation platform on parallel and distributed environments ranging from multi-core or many core machines to multi-node clusters or supercomputer environments. We evaluated our platform with multi-modal transportation networks from the capital city of Ireland, Dublin, and verified its high scalability and real-time performance on both a 1-node with 12 core machine for a multi-threading environment and a cluster of 12 nodes (with a total of 144 cores) for a distributed system. With this platform, municipal planners or traffic engineers at transportation companies can quickly conduct many series of simulations with various what-if scenarios in a parallel and distributed manner, allowing them to assess and select the best responses to sudden incidents, or plan for major events or disaster responses.