Novel survivable logical topology routing in IP-over-WDM networks by logical protecting spanning tree set

Abstract

The survivable logical topology mapping problem in an IP-over-WDM optical network is to map each link (u, v) in the logical topology (at the IP layer) into a lightpath between the nodes u and v in the physical topology (at the optical layer) such that failure of a physical link does not cause the logical topology to become disconnected. It is assumed that both the physical and logical topologies are at least 2-edge connected. Generating a survivable routing is an NP-complete problem. For this problem two lines of investigations have been pursued in the literature: the mathematical programming based approach initiated by Modiano et al., and the structural approach initiated by Kurant and Thiran and pursued further by Thulasiraman et al. The mathematical programming approach is not scalable for large networks, though it gives considerable insight into certain important aspects of the problem. The structural approach requires contraction and expansion of logical graphs and computing link-disjoint lightpaths between pairs of nodes in the physical topology. In this paper, we propose a novel approach based on the concept of protecting spanning tree set of the logical topology. The basic idea is to identify a set of spanning trees of the logical topology and a routing of the logical links such that at least one of these trees remains connected after a physical link failure. Given a set of trees of the logical topology we first present three optimization problems with varying degrees of difficulty relating to this approach and discuss their Integer Linear Programming formulations. We then consider the general case when both the tree set and a survivable routing are to be determined. For this general case we present a heuristic approach. We incorporate in this heuristic a method to augment the logical topology with additional links to guarantee a survivable routing. This approach has several interesting features. It only requires a shortest path algorithm and an algorithm to generate appropriate spanning trees. It also provides a framework for generating a survivable routing for the SRLG failure case. Contractions of graphs and disjoint path generation are not required that greatly reduces the computation time. We provide results of extensive simulations conducted to evaluate our new approach.