Previously we posted about our work on advancing MASON, part of which we briefly discussed making it distributed in order to run large scale models including geographical explicit ones along for optimization and validation purposes. To this end we recently had a paper accepted and presented at the 22nd International Symposium on Distributed Simulation and Real Time Applications (DS-RT 2018), entitled "Scalability in the MASON Multi-agent Simulation System".
In this paper we describe a distributed version of the MASON, and use three existing MASON models: HeatBugs, Flockers, and CampusWorld, to demonstrate how Distributed MASON achieves highly scalable performance, in terms of linear performance increases as the size of the simulations grow using Amazon Web Services. Below you can read the abstract of the paper, see some figures relating to how we go about data management and some of the results. Finally, at the bottom of the post you can see the full reference and access the paper itself.
Abstract:
This paper describes Distributed MASON, a distributed version of the MASON agent-based simulation tool. Distributed MASON is architected to take advantage of well known principles from Parallel and Discrete Event Simulation, such as the use of Logical Processes (LP) as a method for obtaining scalable and high performing simulation systems. We first explain data management and sharing between LPs and describe our approach to load balancing. We then present both a local greedy approach and a global hierarchical approach. Finally, we present the results of our implementation of Distributed MASON on an instance in the Amazon Cloud, using several standard multi-agent models. The results indicate that our design is highly scalable and achieves our expected levels of speed-up.
Full Reference:
Wang, H., Wei, E., Simon, R., Luke, S., Crooks, A.T., Freelan, D. and Spagnuolo, C. (2018), Scalability in the MASON multi-agent simulation system, in Besada, E., Polo, Ó.R., De Grande, R. and Risco J.L (eds.). Proceedings of the 22nd International Symposium on Distributed Simulation and Real Time Applications, Madrid, Spain, pp. 135-144. (pdf)
This research is supported by the National Science Foundation (Grant 1727303).