Optimizing Variance of Link Betweenness Centrality distributing among links in Elastic Optical Networks

Link betweenness centrality of a link indicates how probabilistically a path request will use the link. The link with higher betweenness centrality are expected to be more loaded and hence more congested. We have used genetic algorithm to find link weight distribution in network, so that betweenness centrality has minimum variance among all links. We expect that this should reduce blocking probability and also give better resource utilization. In order to verify the hypothesis, simulations were done with determined link costs. The blocking probability is lower as compared to the case of determined link costs in the network and resource utilization is found to be higher then benchmark algorithms.