An effective game theoretic static load balancing applied to distributed computing

作者：Siar Hajar^*; Kiani Kourosh; Chronopoulos Anthony T

来源：Cluster Computing, 2015, 18(4): 1609-1623.

DOI：10.1007/s10586-015-0486-0

• Distributed computing system
• Static load balancing
• Non-cooperative games
• Nash equilibrium
• Evolutionary algorithms

In this paper an algorithm has been proposed to balance the loads in a distributed computing system based on game theory which models the load balancing problem as a non-cooperative game among the users. The proposed load balancing game, which is infinite and with perfect information, aims to establish fairness both in system and user level. The optimal or near-optimal solution of the game is approximated by a genetic algorithm and an introduced hybrid population-based simulated annealing algorithm, using the concept of Nash equilibrium. Since all users responses are shown to converge to their near-optimal solution, distribution of users' jobs is "fair". Simulations demonstrate near-optimality of the proposed algorithms in terms of makespan and fairness for the proposed load balancing scheme.

• 出版日期2015-12