Avaliação de redes ópticas do mundo real através da análise de curvas de bloqueio

Abstract

The continuous change in the scenario of traffic demands in telecommunications networks has increased the number of studies that seek to improve network performance. The flexibility that networks have to apply changes facilitates the implementation of new techniques, generating results that can be confusing and difficult to classify. The performance of a network can be evaluated in several ways, one of them is through the request blocking rate. A request is blocked when the network is overloaded and there is no spectrum available for new demands, and the smaller the amount of blocking, the better the network performance. In the blocking rate results represented by blocking curves, there may be intersection points, making the curves, when compared to each other, have at one point the lowest blocking rate and not at another time, which makes it difficult to identify the better results. The modification scenarios that can be applied in telecommunications networks range from the physical layer to the ways in which information travels through the network, and all these scenarios can interfere with the network’s blocking rate. From the difficulty of identifying the best result and the number of scenarios that impact the blocking rate of a network, in this work, two methods are implemented for the analysis of blocking curves and classification of results, which enable the identification of the best results. One of the methods is based on the area under the blocking curve and the other method makes a comparison between the curves blocking values. In order to test the proposed methods, let’s analyze the blocking results of 10 sets of topologies. Each set contains different topologies that have the same information traffic capacity and, all were generated from changing topologies in the real world. With the classification of results using the methods, it was possible to perform topological analyzes that can be the starting point for decision making in increasing the efficiency of telecommunications networks. For all sets, both methods were able to identify the best and worst results and, consequently, identify patterns that are repeated in the best and worst topologies of each set. The methods have different forms of analysis, and both proved to be efficient to be used in decision making in telecommunications networks based on blocking results.