Developing a Distributed Agent-Based and DES Simulation Using poRTIco and Repast

• Authors: Athar Nouman;Anastasia Anagnostou;Simon J. E. Taylor
• Venue: DS-RT '13 Proceedings of the 2013 IEEE/ACM 17th International Symposium on Distributed Simulation and Real Time Applications
• Year: 2013

Simulation modeling is a significant tool for systems analysis that is being used in industry and services for several years. As the world becomes more connected, organizations complexity increases and traditional simulation techniques often are inadequate to represent modern problems. Moreover, simulation building requires high-skilled modelers and experimentation with large-scale simulations is highly computation intensive exercise. Distributed simulation, by enabling experimentation across multiple processors over a network or the cloud, has the potential to provide solution for the required computation capacity. Additionally, reusability of models is supported by enabling individual models to interoperate. In this paper we demonstrate the development of a distributed hybrid agent-based and discrete event simulation using high level architecture (HLA) technology. We implemented our project in open source software which has the advantage of cost-free access and source code accessibility. Using emergency medical services as exemplar case study, we developed an agent-based model of an ambulance service and several discrete event simulations of accident and emergency departments using Repast toolkit. The interface runtime infrastructure was developed in portico software. We tested the performance of the 1.3 and the Evolved version of the IEEE 1516 HLA standard. The yielded results show that the overall performance of both HLA versions is satisfying. However, interestingly, HLA 1.3 performs better on a single processor while HLA Evolved demonstrates better performance when executing in a network.