No abstract

The study of protection against failures in WDM optical networks plays strategic importance due to huge bandwidth of optical fiber. The p-Cycle is a novel protection approach based on pre-configured cycles to provide a fast recovery for single link failure. The optimal selection of cycles is the central problem to get a high protection performance of p-Cycle approach. In that sense, the importance of the length of a cycle for fast restoration is studied in this paper, because it affects directly the quality of restoration, therefore, the quality-of-service can be affected. This paper proposes a new heuristic approach and a new evaluation metric to solve the p-Cycle protection problem which simultaneously optimizes cost, protection, length and fairness of solutions in a mono-objective approach. Experimental results show a good performance of the heuristic proposed, when is compared with a recent Genetic Algorithms approach and the state-of-the-art ILP method.

The huge bandwidth of optical fibres is exploited through wavelength division multiplexing technology, which introduces new complexities in the routing problem. In this context, the wavelength converter allocation problem has become a key factor to minimize blocking. The wavelength converter allocation problem has been treated as a mono-objective problem minimizing the number of wavelength converters or minimizing blocking; however, both criteria are in conflict with each other. Therefore, the wavelength converter allocation problem is studied here in a pure multi-objective optimization context for more appropriate decision making. This work proposes a multi-objective optimization approach based on an evolutionary algorithm which simultaneously minimizes blocking and the number of wavelength converters. Extensive simulations on three real optical networks show promising results in the sense that our algorithm generates the trade-off curve between blocking and the number of converters needed, and outperforms a recently proposed approach.

The techniques for protection against failures in optical networks have been discussed in many scientific papers. In the case of Wavelength Division Multiplexing (WDM) optical networks, the p-cycles strategy of protection is a promising approach that combines the speed of ring protection and efficiency of resource utilization in mesh protection. In real networks, several optical links may fail at the same time when nodes or ducts failures happen. These sets of links with simultaneous fails can be determined a priori, and they are called Shared Risk Link Groups (SRLG). This paper proposes an approach based on Genetic Algorithm to find the best set of p-cycles that protects the network traffic against SRLG. Independent SRLG restoration is also achieved for better and faster network recovery in case of failures. This makes recovery faster and more efficiently. In addition we present a new algorithm for selecting candidates cycles, specific to this problem. The experimental results show viability of the proposed approach by achieving good performance in comparison to an exact algorithm of the state of the art. Enrique Dávalos, Máster en Ingeniería en Sistemas por la UNA. Ingeniero Electromecánico por la UNA. Programador de Computadores por la Universidad Nacional del Este-Paraguay. Docente Investigador en la FP-UNA.Diego P. Pinto-Roa, Ingeniero Electrónico por la Universidad Católica de Asunción. Alumno del Doctorado en Ciencias de la Computación y Docente Investigador por la FP-UNA.

No abstract