A game-theoretic model for dynamic load balancing in distributed systems

Abstract

Process of improving the performance of a parallel and distributed system through a redistribution of load among the processors is called load balancing. If it is being done at run time then it treats as dynamic load balancing. In this paper we present the framework for obtaining a user optimal load balancing scheme in distributed system. We formulate the Dynamic load balancing problem in distributed systems as a noncooperative game among users. The system model considers various parameters and accordingly the load balancing algorithm is defined. For the proposed noncooperative load balancing game, we consider the structure of the Nash equilibrium. Based on this structure we derive a new distributed load balancing algorithm. Our focus is to define the load balancing problem and the scheme to overcome it, by using new area called game theory.