Font Size: a A A

Link State-aware Resource Allocation Technology In Elastic Optical Netwoek

Posted on:2021-05-12Degree:MasterType:Thesis
Country:ChinaCandidate:S LiFull Text:PDF
GTID:2428330632963029Subject:Information and Communication Engineering
Abstract/Summary:PDF Full Text Request
The emergence of elastic optical networks provides flexible resource allocation method for services,greatly increases the amount of traffic carried on a single link.This brings huge benefits to operators,but it also brings potential risks.The risk is:due to the increased traffic carried by a single link,when the external factors(such as natural disasters)or internal factors(such as the age of the link itself or environmental factors cause the attenuation coefficient to increase),the link status is abnormal(link failure Or link channel quality degradation),and the network capacity will be greatly affected thereby.As the resource allocation method under abnormal link state determines the size of the final network capacity,this topic being promoted.The link state in this topic refers to two abnormal link states:link failure and link channel quality degradation(referred to link degradation).When a multi-link failure occurs,a large number of services are interrupted,and network resources are tightened,fragmentation is more serious.This is not conducive to the recovery of interrupted services.In response to this problem,existing studies use the idea of load balancing to optimize resource allocation.The load distribution in a balanced network can effectively increase the recovery probability of future services,thereby improving the overall service recovery rate to a certain extent.However,load balancing alone does not effectively improve the fragmentation of the network,Therefore,there are certain restrictions on the ability to recover.Based on this,this paper combines the resource usage of the current faulty network and optimizes the service recovery sequence to efficiently configure the remaining network resources to further improve the recovery rate.When multi-link degradation occurs,the communication quality decreases,the spectrum consumption rate increases,and the network capacity decreases.In response to this problem,existing studies judge whether fiber needs replacement based on the corresponding evaluation model,or adjust the modulation format based on the corresponding technology to avoid the problem of service interruption caused by degradation.But even if the service is received normally after adjusting the modulation format,the increase in spectrum consumption caused by the change in the modulation format will also affect the network capacity.Based on this,this paper studies the impact of link degradation on network capacity from the perspective of resource allocation.In general,from the perspective of abnormal link state,this paper analyzes the link state awareness resource optimization problem in elastic optical networks,completes the establishment and simulation of network models under different link states,and the design and simulation of resource allocation algorithms,providing a theoretical reference for network optimization and planning in network with multiple link failure and large-scale link degraded.The research work and innovations of this article are as follows:1.In view of the shortage of network resources and the difficulty of recovering from large-scale service interruptions in the case of multi-link failures in elastic optical networks,research was conducted in two scenarios:network expansion and no network expansion.Three genetic-based algorithms were proposed(traditional Genetic Algorithms(GA),genetics based on minimum resource priority recovery,mixed genetic taboo)to optimize the sequence of recovery service by sensing network resource usage,to achieve efficient allocation of remaining network resources,thereby improving service recovery rate and reducing network expansion costs.The simulation results show that the three genetic-based service order-aware recovery algorithms proposed in this paper are significantly improved in terms of recovery rate and expansion cost compared with the traditional maximum resource-first recovery,minimum resource-first recovery,and random recovery algorithms.The ranking of the three genetic-based algorithms in terms of time efficiency and recovery performance is:time-efficiency ranking(the order of iterations from small to large when reaching the optimal solution)is genetic based on minimum resource priority recovery algorithm>hybrid genetic taboo algorithm>traditional genetic algorithm;The order of recovery performance(that is,the recovery rate is from large to small,and the expansion cost is from small to large)is a hybrid genetic taboo algorithm>genetics based on minimum resource priority recovery algorithm>traditional genetic algorithm,operators can choose according to specific circumstances in terms of time cost and recovery performance.At the same time,the results show that the traditional minimum resource-first recovery algorithm and maximum resource-first recovery algorithm cannot take into account the two scenarios of network expansion and non-network expansion(the performance degradation is severe in a certain scenario).The genetic algorithm proposed in this paper can take good care of this different target scenario,so it has a stronger generalization ability.2.Aiming at the problem of shortage of network resources caused by large-scale link degradation in elastic optical network,first,a network model under link degradation is established according to the phenomenon that the accumulated noise value of the service through links with different degradation levels is different,and then based on the proposed network model,a resource allocation algorithm based on Q-learning is being proposed,which learns and perceives the impact of different link degradation states in the network to plan routes for services and allocate resources.Simulation results show that the resource allocation algorithm proposed in this paper improves the blocking rate significantly compared with the traditional algorithm in the case of large-scale link degradation in the network.The results also show that when the number of degraded links in the network is small and the degree of degradation is small,the impact of the degradation on the communication capacity is small.On the contrary,when the number of degraded links is large and the degree of degradation is large,the impact on the network capacity cannot be ignored.This provides operators with a certain theoretical reference for network planning.
Keywords/Search Tags:elastic optical network, links failure, links degradation, resource allocation
PDF Full Text Request
Related items