A ROBUST ORGANIZATIONAL POLICY FOR TASK AND RESOURCE ALLOCATION: A NOVEL FRAMEWORK

Abstract

In a Multi-Agent System existing formalisms for implementing organizational policies assign specific roles to each agent. Examples are hierarchical organization, contract net protocol, social reasoning mechanism, and the use of matchmaker agents. These policies allow the problem solving roles of the agents to change dynamically but do not adapt to variations in computational load on the multi-agent system. They are designed to operate for a predefined maximum problem-solving load and fail to respond, when the number of task requests, exceed this limit. This paper proposes an adaptive organizational policy, called E-RTA (Efficient Resource and Task Allocation), for Multi-Agent-Systems that operate under time constraints and varying computational loads. We consider MAS (Multi-Agent Systems) as consisting for several problem-solving organizations where each organization is comprises multiple agents that may be grouped into teams for specific problem solving.