MULTIPATH PROTECTION WITH OPTIMIZATION OF TRANSMISSION RATE PARTITIONING AND RSA PRIORITY ORDERING IN ELASTIC OPTICAL NETWORKS.
Elastic optical networks, Multipath protection, asymmetric partitioning, RSA ordering, optimization, evolutionary algorithms.
Transport optical networks have been widely used to support the significant traffic volumes demanded by various services in the society. Researches have shown a high growth in such demands over the last years. In this context, elastic optical networks (EONs) provides flexibility on spectral utilization and an increase in optical networks’ capacity adequately to meet the growing traffic demands. The occurrence of failures, usually due to human fault or network equipment breakdown, is a possible event that typically results in significant data and revenue losses for network operators. Thus, the network survivability area is an essential requirement for EONs. Among the survivability mechanisms found in the literature, multipath protection uses the split-spectrum technique to divide the requested transmission rate into smaller partitions and transmit them simultaneously throughout disjoint optical paths. The squeezing technique, which allows a reduction in the required bandwidth for a connection during a failure event, has also been commonly investigated in works addressing multipath protection. The association of these two techniques enable the application of optimization strategies for the improvement of EON performance.
This thesis presents some proposed heuristics and metaheuristics to optimize the performance of elastic optical networks under multipath protection schemes and dynamic traffic. Firstly, some prioritization resource allocation ordering strategies in the Routing, Modulation Level, and Spectrum Assignment (RMLSA) process are proposed. Then, a metaheuristic that utilizes an evolutionary algorithm to find the best ordering strategy for each source-destination pair in order to reduce the network path-request blocking probability (PB) is introduced. Next, the use of asymmetric partitioning of the total transmission rate of connections in multipath protection schemes is proposed to enhance network resource utilization. For this, a multi-objective optimization is proposed to determine the best partitioning for each source-destination pair capable of mitigating the PB as well as the average squeezing factor of connections. Additionally, this research introduces a routing algorithm that generates groups of disjoint routes, strategically ordered, and uses them to perform fixed-alternate or dynamic routing in networks with multipath protection. A new metric is also presented, allowing the measurement of the average percentage reduction in the requested transmission rate in protected networks that admit traffic squeezing in the event of a failure. Simulation results showed that the use of a hybrid RMLSA priority ordering, applied individually to each source-destination pair in an optimized manner, can provide a significant reduction in the network path-request blocking probability. The results also demonstrated that the use of asymmetric partitioning of the total transmission rate, optimized for each source-destination pair, can promote a reduction in PB and further improve the average squeezed bandwidth fraction of connections under single link failure scenario.