Research On Bandwidth Allocation Optimization Mechanism In Elastic Optical Networks

The backbone network is bearing a heavier and heavier burden on bandwidth allocation. But the traditional wavelength division multiplexing optical networks have not met the demand of such a heavy bandwidth requriment. Therefore, the future research focus of the backbone network is the elastic optical networks who support large capacity, dynamic and flexible transmission. The wavelength assignment in the wavelength division multiplexing optical networks lead to poor flexibility and low bandwidth utilization of network resources. But the elastic optical networks with the introduction of OFDM technology garantee a flexible allocation of network resources. At the same time, the elastic optical networks structure itself offering finer spectrum granularity and good flexibility pose new challenges in the spectrum allocation optimization process. For example: the spectrum continuity and spectrum contiguity constraints, the recession of fairness, spectrum fragmentation issue and so on. This thesis mainly studies the optimization of spectrum allocation mechanism on unfairness, spectrum fragmentation problem in the elastic optical networks.The finer spectrum granularity in the elastic optical networks make wider spectrum services more likely to be restricted by the constraints of spectrum continuity and spectrum contiguity, leading to higher blocking probability. Obviously, it would deteriorate the fairness of multi-granularity services. Therefore, a fair dynamic routing and spectrum allocation mechanism is proposed. By virtualizing the network resources as ‘resource pool’, a two-dimensional rectangle packing model in combination of service holding-time is introduced to satisfy the two spectrum constraints and to optimize the allocation of the ‘resource pool’ for dynamic multi-granularity services. Besides multigranularity service fairness is achieved by tackling heterogeneous-size services together. Simulation results show that the benefit in terms of fairness can be reaped. Meanwhile, we also investigate the proposed mechanism’s impact on other performance metrics, namely, spectrum resource utilization and blocking probability. The results indicate that the proposed mechanism can achieve lower blocking probability and higher spectrum efficiency.The frequent setting up and tearing down of hybrid granularity services in elastic optical networks leads to the generation of unavailable resources-fragments. And the accumulation of such fragments will increase service congestion, reduce resource utilization and induce unfairness. To settle the problem of spectrum fragments in the elastic optical networks, a mechanism of spectrum allocation optimization based on spectrum defragmentation is proposed. The mechanism employs ant colony mechanism, in which the transition probability is defined by combining the spectrum compactness and transmission length factors together. At the same time the real-time path information is attained through the updating rules. Thus spectrum fragments are lessened and the blocking probability is reduced. On the basis of this, a fair spectrum allocation optimization mechanism based on spectrum defragmentation is proposed by adding the process of service ordering. Simulation results show that the proposed mechanisms can effectively guarantee resource utilization on the premise of blocking probability.