Learning-agent-based simulation for queue network systems

Abstract

Established simulation methods generally require from the modeller a broad and detailed knowledge of the system under study. This paper proposes the application of Reinforcement Learning in an Agent-Based Simulation model to enable agents to define the necessary interaction rules. The model is applied to queue network systems, which are a proxy for broader applications, in order to be validated. Simulation tests compare results obtained from learning agents and results obtained from known good rules. The comparison shows that the learning model is able to learn efficient policies on the go, providing an interesting framework for simulation.

Keywords:

• Simulation
• agent
• machine learning
• queue network

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

The authors would like to thank the Carlos Chagas Filho Foundation for Research Support in Rio de Janeiro, FAPERJ, for supporting the research by means of Grant E-26/202.789/2015. This work was partially supported by the National Council for Scientific and Technological Development – CNPq, under Grants 307126/2017-0 and 311075/2018-5.