Routing And Wavelength Assignment Algorithms In Optical Networks

With the development of broadband video,multimedia and internet business, optical network is becoming increasingly important. For the purpose of effectively utilizing network resources, when business requests arrive in optical networks, effective routing and wavelength assignment (RWA) algorithm must be implemented to build path(s) with available wavelength. In addition, to achieve equilibrium between network block rate and cost, designing efficient wavelength converter placement algorithm is requisite. Aiming at these problems, this paper made the following efforts:In the first place, systematic classifications on routing subproblem and wavelength assignment subproblem are made respectively. Then for routing and wavelength assignment joint problem on layered graph, some influencing factors are skillfully normalized to be a meaning comprehensive and calculation simple figure, which is designed to represent the abstract length of each link in the layered graph. On the base of this normalization, a genetic algorithm employing variable-length directly coding is put forward to solve this joint problem. List based coding method, principle and concrete steps of crossover and mutation are elaborately expounded in the proposed genetic algorithm. Owing to the intuition of the particular coding, crossover and mutation method, current solution in evolutionary process and the optimal solution can be detected directly without decoding. Being convenient and practical, this algorithm has widespread meaning.To provide theoretical support for the wavelength converters placement algorithm, a model of inexact network state information is established and probability estimation of demanding convert wavelengths on node is obtained in the model. Finally, several heuristic elements influencing performance of placement algorithms are achieved through reasoning analysis, and inspired by these elements, a clustering paths sorting nodes algorithm is proposed to solve wavelength converters placement problem. In this algorithm, integer category scale series are obtained via fine tuning arithmetic progressions, and influence of other paths in the network on the current path is also taken into account. To conclude, this algorithm is rigorous designed and has strong overall interactivity.