MULTI-PATH PROVISIONING STRATEGIES FOR PROTECTION IN ELASTIC OPTICAL NETWORKS UNDER DYNAMIC TRAFFIC.
Optical networks, Multipath protection, split spectrum, squeezing, optimization, evolutionary algorithms.
Optical transport networks are widely used to support the traffic of large volumes of data demanded by various services used by society nowadays. Several surveys show a high growth in such demands over the past few years. In this context, elastic optical networks (EON) represent a promising concept to achieve flexibility of spectral use and increase the capacity of optical transport networks to meet growing traffic demands. The occurrence of failures due to human causes or network equipment is a possible event that usually generates large losses of data and revenue for operators. Thus, the fault survivability area has become an essential requirement for EONs. Among the survival mechanisms found in the Literature, multipath protection uses the split spectral technique to divide the requested transmission rate into partitions that are transmitted by disjoint optical lightpaths used by the protection scheme. The squeezing technique, which allows a reduction in the bandwidth required by a connection during a failure, is also commonly used by multipath protection schemes. These two associated techniques, to promote multipath protection, allow the application of optimization strategies and network performance improvement. This work proposes the investigation of different heuristics or meta-heuristics to optimize the performance of multipath protection schemes operating in EONs under dynamic traffic. A dedicated multipath protection scheme is proposed that allows the asymmetric partitioning of the total connection bitrate and uses an evolutionary algorithm to find the best partitioning for each source-destination pair, in order to provide a reduction in the blocking probability and in the fraction of bandwidth squeezed from the network. A protection scheme that performs the optimization of the prioritization ordering of application resources is proposed and investigated in this work. The results obtained in the simulations show that multipath provisioning together with bitrate squeezing restoration technique applied to protection in elastic optical networks provide optimization opportunities, which can be performed through heuristics or meta-heuristics (evolutionary algorithms) and results in improved network performance.