Skip to Main Content
In this paper we present two methodologies, one is to use MPI collective communication functions as performance measures to measure communication time between peers. The other is to use a Distributed Genetic algorithm with MPI functions running on each peer node for solving a variety of optimization problems. Genetic Algorithms are found useful in variety of problems, such as in searching and optimization. Distributed Genetic Algorithms are inherently embarrassingly parallel which leads to efficient implementation on the nodes. In this work DGA is used first to distribute resources on nodes to maximize availability within budget and second to find in-best network routes within links cost and end-to-end delay. The iterations for DGA to converge are measured. It is seen overall performance of DGA is not affected as nodes join or leave the network.