Conference Papers

A Framework for simulating mobility services in large scale agent-based transportation system

AuthorChristian Rakow
SourceSummer Simulation Conference 2019 - WIP (to appear), Berlin 
LinksDownload   |   BibTeX 

Mobility services nowadays offer an enormous variety for modes of transportation. The development and planning of such systems requires careful consideration, in which regard traffic and demand simulations can play an important role. In this paper we introduce JIACVI-ITS, a framework for simulating large scale agentbased transportation scenarios. It is based on the actor model in which agents communicate entirely via message passing. Simulation events are synchronized conservatively but, by exploiting model uncertainty, are processed in parallel between small time steps, which enables the utilization of multiple CPU cores. The framework also offers a programming language and platform agnostic model definition, which facilitates the integration into external systems. We have evaluated our simulation approach with regard to speedup with an increasing number of CPU cores and show significant performance gains.