Decentralized control of cooperative and autonomous agents for solving the distributed resource allocation problem

Abstract

This paper focuses on the decision-making mechanism of coordination protocol for a multi-agent system in collaborative, distributed environment requiring resource allocation solutions. The model of distributed collaboration network (DCN) for distributed resource allocation is developed. The use of monitored viability of each agent as a feedback for the agent's reflexivity and goal adjustment mechanisms during the coordination process is introduced. Question: Will (can) a multi-agent system, as a whole, achieve efficient resource allocations, when each agent makes independent decisions, adjusted by feedback of its own viability measure to maximize its own goal based on limited information about the entire system? To investigate this question, experiments were conducted by using the parallel simulator TIE/Protocol to model the DCN and multi-agent system behavior.