Multi-agent Model for Real Time Resource Allocation

Abstract

Abstract. A large number of important applications are reducible to combinatorial models. Almost all of them are at least exponential complexíty and cannot be solved in a traditional way. In the paper, we consider an agent -based approach to solve a class ofcomplex combinatorial problems in the area ofplanning and scheduling ofbounded resource al/ocation under real time and temporal constraints. The model ofthe problem is formulated in terms ofcontract al/ocation over a set of contractors and specified as an auction-based competítion of intelligent agents-contractors under agentmanager supervision. The paper contributions are repeatable auction-based scheme of random search of admissible decisions: knowledge-based specification ofreal-time and temporal constraints that is used lo order of contrae! al/ocation from step lo step ofauction procedure; dynamic programming approach for forming strategy ofbargaining by agent-contractor.