Spectrum fragmentation metrics and their use in optical channel allocation algorithms

A wide overview of metrics to evaluate the fragmentation is presented together with their application for Routing and Spectrum Assignment (RSA) in Elastic Optical Networks (EONs). A basic heuristic spectrum allocation algorithm conceived for keeping the fragmentation index low during the connection allocation process has been defined. Starting from the application of such algorithm on a single link, the analysis has been extended to encompass sample networks and a real size case study. It has been found that both in a context of dynamic traffic and in a semi-static incremental network loading, good heuristics based on simple and low computational demanding allocation policy (i. e. First Fit), perform very close in terms of average traffic loss to methods which use more complex algorithms based on fragmentation metrics. However some of the applied metrics result still beneficial as they obtain an equalization of the connection blocking rates between the various types of multi granular traffic, in particular by reducing considerably the probability of blocking of high spectrum demanding superchannels.