DWDM network protection/restoration with ring/cycle topologies.

Abstract

In this thesis, we first discuss the differences between plain optical switches and wavelength routers, which are the building blocks for static WDM networks and wavelength routing networks respectively. We study the different features of these building blocks, and clarify the different meanings of 'ring/cycle topologies' in those two types of optical WDM networks. Then we propose a heuristic algorithm to find a set of link-covering rings to be used in WDM network protection/restoration. Our algorithm yields a scalable set of rings that covers most links in a sparse mesh network. To study whether it is beneficial to use P-cycles instead of pure rings in the WDM ring cover approach, we design a generic linear programming model for comparison purposes. By doing simulation studies with the model, we show that the method of P-cycle cover can only achieve very little performance improvement over pure ring cover. Thus from an engineering point of view, P-cycle cover is not practical. Finally we propose a distributed protocol to find rerouting paths by P-cycle decomposition, in wavelength routing networks. The restoration performance of the distributed protocol can be very close to the theoretical upper bound under our traffic distribution assumptions. That justifies P-cycle is a highly efficient topology for network protection/restoration.