Routing and spectrum assignment algorithm with prediction for elastic optical networks under self-similar traffic

Abstract

Poisson traffic model was usually used in existing investigations for evaluating the performance of routing and spectrum assignment (RSA) algorithms in elastic optical networks (EONs). However, many studies showed that the self-similar traffic could reflect the Internet traffic property more accurately than Poisson traffic. In addition, the performance of RSA could be improved by using the predicted holding time awareness in EONS. In this paper, we propose a dynamic RSA algorithm that achieves better performance by using Back Propagation neural network (BPNN) to predict the information of each future connection with holding time awareness under self-similar traffic. Experimental results show that the proposed RSA algorithm could obtain lower blocking probability compared to existing RSA algorithm under self-similar traffic.