Considering Linear and Nonlinear Impairments in Planning WDM Networks

In a wavelength division multiplexed (WDM) network, each fiber link carries high-rate data on several different wavelengths, thus creating multiple channels within a single fiber. The most common architecture utilized for establishing communication in WDM optical networks is wavelength routing, where optical pulse trains are transmitted through lightpaths, that is, all-optical WDM channels that may span multiple consecutive fibers [1–3]. Since the lightpaths are the basic switched entities of a wavelength-routed WDM network, their effective establishment and usage is crucial. Thus, it is important to find efficient algorithms to select the routes for the requested connections and to assign wavelengths on each of the links along these routes, so as to serve the traffic and optimize a certain performance metric. This is known as the routing and wavelength assignment (RWA) problem. The constraints are that paths that share common links are not assigned the same wavelength (distinct wavelength assignment). Also a lightpath, in the absence of wavelength converters, must be assigned a common wavelength on all the links it traverses (wavelength continuity constraint). The RWA problem is usually considered under two alternative traffic models. When the set of connection requests is known in advance (e.g., given in the form of a traffic matrix), the problem is referred to as offline or static RWA and is encountered in the planning phase of a WDM network, while when the connection requests arrive at random times, over an infinite time horizon, and are served on a one-by-one basis, the problem is referred to as online or dynamic RWA and is encountered in the operation phase. In this chapter we will focus on offline RWA, which is known to be a NP-hard problem [4]. Offline RWA is generally considered more difficult than online RWA, since it aims at jointly optimizing the lightpaths