Protection And Restoration Mechanisms Under The Multiple Failures In Flexible Bandwidth Optical Networks

The amount of bandwidth capacity greatly improves with the rapid development of Internet applications and traffic services. Currently, the transport networks have the large-scale, complex, and large-capacity characteristic. As a result, the multiple failures become an important issue in flexible bandwidth optical networks. In traditional WDM networks, the wavelength channels, the line rates, and modulation format are fixed, which causes the waste problem of bandwidth resources. It is also not good for protection and restoration techniques in the future optical networks, especially towards the low redundancy of the network resources and high restoration rate. In order to address the above problems, flexible bandwidth optical networks provide a finer granularity according to client traffic demands, which is also a significant feature of variable-bandwidth in this network architecture. Aim at the protection and restoration problem, the trade-off problems between the protection resources requirement and the limitation of the network capacity provisioning are resolved by allocating the network resources on recovery paths under the multiple failures.The protection and restoration problems under the multiple failures are solved with the support of973program (the national basic research and development plan)"The survivability for ultra-large capacity optical networks under multiple failures." Based on the characteristic of both multi-link failures and the recovery traffic, it is possible to solve the network survivability by employing the dynamic load balancing schemes. According to the failure probability theory, the protection and restoration mechanisms are developed by considering the joint failure probability of primary path and backup path, solving the trade-off problem between the joint failure probability and the network resources. Taking into account the survivable virtual optical network mapping problem, the optimized models of both the energy consumption and the network cost are proposed by using protection and restoration technologies. The mainly innovative contributions of the dissertation are listed as follows.Firstly, according to characteristic of both multi-link failures and the dynamic recovery traffic, a novel dynamic load balancing shared-path protection (DLBSPP) is proposed to efficiently restore failure traffic under multi-link failures, where the primary path and link-disjoint shared backup path are computed by using Dijkstra algorithm. The DLBSPP algorithm employs first fit and random fit schemes to search and assign the available spectrum resources. Traffic-aware restoration (TAR) mechanism is adopted in the DLBSPP algorithm to compute new recovery routes for carrying the traffic affected by the multi-link failures and then the multi-link failures are efficiently restored. Simulation results show that, compared with the conventional shared-path protection (SPP) algorithm, the DLBSPP algorithm achieves lower blocking probability, better spectrum utilization ratio, and higher failure restoration ratio. Especially, the DLBSPP algorithm is1.5times as failure restoration ratios as the SPP algorithm under the case of three-link failures.Secondly, in both dedicated protection and shared protection, the overall objective of the problems is to minimize spectrum consumption, where the joint failure probability between primary and backup paths must meet the maximum joint failure probability requirement. We mainly identify the relationship between total frequency slots consumed and average joint failure probability.1) For the dedicated protection problem in flexible bandwidth optical networks, we develop an ILP model of dedicated-path protection in order to minimize the total frequency slots consumed, and also propose a spectrum-aware dedicated protection algorithm with joint failure probability (SADP). For comparison, a conventional dedicated-path protection (CDPP) algorithm is introduced. Compared with the SADP algorithm, simulation results show that the ILP model of dedicated-path protection minimizes the total frequency slots consumed, reducing about12.3%spectrum consumption. Moreover, the SADP algorithm achieves better trade-off between total frequency slots consumed and average joint failure probability than ILP model and the CDPP algorithm.2) Considering maximum shared degree (MSD) of backup spectrum resources, a minimum free spectrum-block consumption algorithm (MFSB) is first proposed for providing shared-path protection with joint failure probability in flexible bandwidth optical networks. For a comparison, both the conventional shared spectrum-block (CSSB) consumption algorithm and the maximum shared spectrum-block (MSSB) consumption algorithm are introduced. Simulation results show that the MFSB algorithm provides a better trade-off between the minimum spectrum-block consumption and the average joint failure probability.Thirdly, focusing the trade-off problem between the resources requirement of network survivability and the limitation of the network capacity, different optimized models of spectrum resources, energy consumption, and network cost are developed by employing the different protection and restoration approaches.1) Minimized spectrum resources consumption is investigated by considering the rescaled failure probability in a flexible bandwidth optical network. The rescaled failure-probability-aware algorithm (RFPA), the traffic cognition algorithm with rescaled failure probability (RFPTC), and an ILP model are proposed to address the trade-off between spectrum consumption and average rescaled failure probability.2) A virtual optical network (VON) request is mapped to the physical network by employing the principles of both minimum energy submatrix mapping and largest bandwidth requirement versus smallest distance mapping. Considering VON mapping problem, a survivable energy-aware approach (SEA) is proposed to minimize energy consumption in flexible bandwidth virtual optical networks. Simulation results show that SEA can efficiently reduce energy consumption, decreasing about37.5%for the line rate400Gbps.3) An ILP model and the LBSD (the largest bandwidth requirement (LB) of virtual links versus the shortest distance (SD)) mapping approach are developed to minimize the network cost, and two baseline mapping approaches are introduced, named LCLC (the largest computing resources (LC) requirement versus the largest computing resources provisioning) and LCSD (the largest computing resources requirement versus shortest distance) mapping approaches, for comparison. Compared with LCLC and LCSD, simulation results show that LBSD achieves network cost near the ILP solutions and greatly reduces the cost.