Research On Resource Management Technologies In Elastic Optical Networks

With the rapid development of 5G networks,cloud computing,artificial intelligence and other emerging internet technologies in various fields,the applications with demands of high-speed and high-bandwidth in the network keep increasing.These have become the challenges for the transmission rate,communication capacity,and operating cost of the network.In the meantime,abundant service request types,such as delaysensitive IR requests,AR requests that need to reserve bandwidth in advance,and MR requests that need to accomplish bulkdata transfer timely,are in need of higher standards for heterogeneous services and transmission quality in future networks.In order to provide more flexible spectrum resource allocation and diversified services,while ensuring the quality of network transmission,the Elastic Optical Networks(EONs)has emerged as an important means to improve network performance.This thesis aims to solve the generation and optimization of spectrum fragmentation in dynamic elastic optical networks and the allocation of time-frequency resources in static elastic optical networks.With the goal of improving the utilization rate of spectrum resources and reducing the probability of traffic congestion,the resource management technologies in elastic optical networks are researched for different types of traffic requests.The dynamic traffic grooming algorithm based on double-fused with Link Optimizing,the static RMSTA algorithm based on proportional optimization and the dynamic RMSA algorithm based on fuzzy logic control in the elastic optical network are mainly studied.The main work and innovations of this thesis are as follows:(1)Dynamic Traffic Grooming based on Double-fused with Link Optimizing for elastic optical networksIn order to reduce the impact of the continuous spectrum resource fragmentation on the network performance due to the random arrival and deployment of service requests in the elastic optical networks,based on the research of traffic grooming,this thesis proposes a dynamic traffic Grooming based on Double-fused with Link Optimization(DTG-DFLO)algorithm in dynamic elastic optical networks.In the early stage of establishing the network topology model,an Optical Network Virtualization based on Double-fused(ONV-DF)model is proposed.By establishing the double-fused network topology,calculating and constructing the alternative path set of the double-layer fusion network.In path selection and resource allocation,the Degree of Link Selection(DLS)and the Degree of Fragmentation Increasing(DFI)are introduced,and a Dynamic Traffic Grooming with Link Optimization(DTG-LO)algorithm is proposed.The simulation results show that when the traffic load increases to 1000Erl in NSFNET,DTG-DFLO algorithm can reduce the blocking rate by about 12.3%,and improve the spectrum resource utilization by about 2.2%,compared with the T-RMSA algorithm.When the traffic load increases to 2000Erl in USNET,DTG-DFLO algorithm can reduce the blocking rate by approximately 14.1%,and improve the spectrum resource utilization by approximately 1.3%.(2)Static RMSTA algorithm based on proportional optimization in elastic optical networkIn order to improve the joint utilization of time and spectrum resources in the elastic optical network with known AR service requests,this thesis proposes a static Routing,Modulation format,Spectrum,and Time Assignment algorithm based on Proportional Optimization after researching the static RMSTA problem.First,an Integer Linear Programming(ILP)model with multi-objective proportional optimization for static RMSTA problem is constructed.The objective is to minimize the maximum utilized frequency and time slots index,and to balance resource utilization in frequency domain and time domain by adjusting proportional parameter.For large-scale network,three heuristic algorithms based onascending order of the elastic time window,descending order of the data transmission and ascending order of the alternative schemes are propoesd.The simulation results show that the proposed ILP model can effectively solve the RMSTA problem in the elastic optical network.In the large-scale network,by adjusting the proportional parameter of the objective function,the PO-AET algorithm achieved the best performance when minimizing the maximum utilized frequency slot index,whereas the PO-DDV algorithm worked best when minimizing the maximum utilized time slot index.As for the compromise between two indexes,both PO-AET and POAAS provided satisfying results.(3)Dynamic RMSA Algorithm Based on Fuzzy Control in elastic optical networksin view of the constantly changing network resources and randomly generated spectrum fragmentation problem,In order to improve the optimization capability of fragmentation,on the basis of researching the dynamic RMSA problem in elastic optical networks,in this thesis,a Dynamic Routing,Modulation format,and Spectrum Assignment based on Fuzzy Logic Control(RMSA-FLC)algorithm is proposed.Based on the research of fuzzy logic,a dual-input single-output fuzzy control system model based on Degree of Fragmentation Optimization(DFO)and DLS is constructed.The path-control weight(P W)is achieved as the output of the fuzzy control system,and the path and spectrum allocation scheme with the maximum PW is selected for the IR requests of real-time arrival.The simulation results show that when the traffic load increases to 1000Erl in NSFNET,compared with the T-RMSA algorithm,RMSA-FLC algorithm can reduce the blocking rate by about 10.5%,and improve the spectrum resource utilization by about 1.1%.When the traffic load is constant at 300Erl,the number of requests increases from 500 to 3000,the difference in blocking rate of the RMSA-FLC algorithm is less than 0.06%,and the difference in spectrum resource utilization is less than 1.43%.Compared with other algorithms,the stability of the network performance of the RMSA-FLC algorithm is better.