Cooperative versus selfish routing in WDM networks a study in multi-objective context

The performence of centralized and distributed routing in wavelength converter allocation problem are studied in this work. The distributed routing is based on selfish routing in which each connection tries to improve its blocking probability. In counterpart, in centralized management, the routing of connexions are calculated by cooperative approach to improve the overall blocking probability of system. In the cooperative context, it is proposed a pure evolutionary algorithm which calculates simultaneously the converters allocation and traffic load flows. For selfish routing, an evolutionary algorithm calculates the converters allocation while the traffic load flow assignment that maximizes the benefit of each connection is accomplished by simulations. Both approaches are compared using Pareto Anarchy Price measure which is a proposal of this work. Experimental results indicate that, when the traffic load increased the Pareto Anarchy Price improves, and paradoxically, the quality of solutions gets worse.