The research focuses in this thesis are on optical switching systems based on WDM, including their performance, scheduling mechanisms and routing and wavelength assignment (RWA).; First of all, we introduced an analytical model to obtain the performance bounds for a WDM switch based on a passive star coupler. We modeled the system under saturation traffic loading while ignoring the specific channel access scheme as a multi-class closed queuing network ( BCMP network), for which a product-form solution for the steady-state probabilities exists. Therefore the exact solutions can be derived. Based on the model, we investigated the effect on the system performance measures of a variety of different system parameters.; Next, we designed a scheduling mechanism for WDM packet switch systems to support Quality of Service (QoS) for multimedia applications. We introduced a novel hierarchical scheduling framework to use in a class of photonic packet switching systems based on WDM, in which we separated the flow scheduling from the transmission scheduling. We showed such separation was essential for achieving scalability such that large input/output ports could be accommodated, and also for offering flexibility in that optimal scheduling algorithms could be derived in different level. The key feature of the proposed scheduling mechanism is that it takes into account potentially different quality of service requirements from different traffic flows. In addition, we studied its performance under different system configurations.; Finally, we examined the routing and wavelength assignment (RWA) problem in wavelength-routed wide-area networks. Based on an extensive study of the existing algorithms adopted in the virtual topology design, we proposed several critical improvements. The algorithms we have proposed can be used to set up virtual topologies, the realizing virtual topologies have lower blocking probability, thus achieving better performance. Specifically, we first introduced a genetic algorithm to solve RWA problem, which is proven to be a NP-hard problem. The major advantage of the genetic algorithm is its general efficiency, therefore, it is effective in dealing with routing and wavelength assignment problem under a wide range of traffic conditions. To reduce the running times of genetic algorithm, a greedy random adaptive search procedure (GRASP) is adopted to solve RWA problem. In addition, we proposed several heuristic-based algorithms, including the Lightpath Setup, Alternative Path and Continuous Wavelength Assignment algorithms. They take into consideration the alternative paths, multihop nature of the physical topology and wavelength continuity when doing the wavelength assignment. Compared to some well-known algorithms, our proposed algorithms can obtain better performance with reduced degree of complexity.
展开▼
