A multi-agent memetic algorithm approach for distributed object allocation

In this paper, we present a multi-agent system (MAS) approach to allocate objects over different servers in distributed computing environment (DCE). We show that distributed object allocation problem in DCE with an objective to allocate objects over different servers so that the number of inter-server communications can be minimized is similar to the graph bisection problem. We test the performance of the MAS approach against other heuristic approaches and semidefinite programming (SDP) based lower bounds, and show that the MAS approach fares well when compared to other competing heuristic approaches.

► Proposed distributed allocation problem (DOA) as a graph bisection problem.
► Proposed a multi-agent particle swarm optimization algorithm.
► Solved the DOA problem using the multi-agent procedure.