Traffic lifetime-aware routing considering dynamic grooming in WDM networks

Abstract

This thesis presents a lifetime-aware scheme for routing considering dynamic grooming for WDM mesh networks, which is adaptive to the lifetime of traffic flows and resource utilization on the virtual topology. Although routing and grooming sound to be independent, they are actually closely related particularly at the virtual topology. In this scheme, routing decision is combined with dynamic grooming policy adaptive to resource utilization at virtual topology with the objective to minimize the blocking probability of connection requests. Paths are selected among a candidate set of paths chosen by the grooming policy and utilization of some resources is referred as constraints. Moreover, since lightpath is logical link in virtual topology, in the case of multiple lightpaths existing between a node pair, grooming policy can help routing algorithm to choose one logical link from multiple logical links connecting the same node pair.As suggested in the scheme, grooming policy and routing decision can take lifetime of traffic session into account. A lot of research work shows that although the "long-lived" flows just occupy a very small ratio of all the traffic flows in the Internet, they account for a high proportion of packets(bytes) on a link. Viewing this, we conduct grooming and routing considering this traffic characteristic by classifying flows as short-lived and long-lived. Thus, different services are provided to these flows to produce better resource utilization.Simulation results show that the proposed scheme is adaptive to dynamic network states and can get better usage of resources. As a result, lower blocking probability and higher wavelength utilization are achieved as compared to some benchmark schemes.