Using spare network computing power for genetic algorithm problems

Traditional network design incorporates a failure-recovery model in order to allow calculation of problems independent of knowledge of the network tool layer. This paper explores the possibilities of improving the calculation throughput by constructing a tool for the specific solution of problems which have an inherent ability to deal with partial calculation failure. Using a modified Genetic Algorithm as the client tool, the amount of information the network layer needs to have is brought to an extremely minimal level; this allows for a large scalability factor of the tool due to the reduction of network management tables.