A Genetic algorithm with a Local Improvement Mechanism for Dynamic Load Balancing in Distributed Systems

摘要

Load balancing algorithms can improve distributed system's performance by redistributing the load maong its processors. In a receiver-initiated load balancing algorithms, underloaded processor continues to send unnecessary request messages for load acquistion until overloaded processor is searched while the system load is light. Therefore, it yields ineficient inter-processor communications and much overhead until the additional load received. This paper represents a new dynamic load balancing scheme based on genetic algorithm with a local improvement mechanism. In this scheme, a subset of processors to which the requests are sent off is determined by GA to decrease unnecessary requests. The genetic operators applied to a population of binary strings each of which stands for a list of processors to which the transfer request messages are sent off. Our experiments show the effectiveness of the proposed approach.