Research On Resource Allocation Algorithm Of Elastic Optical Network Based On Data Analysis

In recent years,with the explosive growth of network traffic and the rapid development of optical network technology,the elastic optical network which can allocate spectrum resources flexibly emerges at the right moment.In the EON,resource allocation is one of the important issues.However,the current resource allocation method has the problem of low resource utilization.In order to cope with surging traffic,we need to reduce resource waste by optimizing resource allocation.We need to further optimize resource allocation and reduce resource waste.In addition,optical network is one of the most important infrastructure of the Internet,in order to ensure the safe and stable operation of the network,it is necessary to carry out comprehensive evaluation.Network comprehensive evaluation can not only reflect the running state of the network,but also provide reference for network planning and construction.In this thesis,the resource allocation and network comprehensive evaluation of elastomer optical network are the core direction of research.Through network traffic prediction,business information can be obtained in advance,network resource allocation can be optimized and network resource utilization can be improved.Through the combination of subjective and objective network comprehensive evaluation,the resource allocation algorithm proposed in this thesis is applied to verify the effectiveness.The main work and achievements are as follows:(1)The design and verification of resource allocation algorithm based on prediction and link state are completed.In addition,a neural network traffic model based on attention mechanism is constructed to improve the stability and accuracy of network traffic prediction.Based on the premise of time-varying service,the algorithm can selectively reserve resources by predicting the traffic information and combining with the current link state.The feasibility of the proposed algorithm is verified by modeling and simulation.Compared with the common resource reservation algorithm,the proposed algorithm can reduce the network blocking rate and improve the number of service redistribution accordingly.(2)An evaluation method of network survivability based on node importance is proposed.The method combines the importance of nodes and the number of shortest paths to establish the evaluation index of network destructiveness,in order to reflect the merits and demerits of network destructiveness,and the importance of nodes is solved based on PageRank algorithm.The simulation results of different network topologies show that the proposed method can accurately reflect the anti-destruction performance of the network.(3)A comprehensive evaluation method of network performance is proposed by combining the two subjective and objective evaluation methods of analytic hierarchy process and coefficient of variation method.The performance of the whole network is divided into basic performance index and anti-damage performance index to establish the evaluation index system and evaluation level,and the effectiveness of the algorithm is verified by the simulation results of various indexes.It includes the resource allocation algorithm proposed in this thesis.The simulation results show that the evaluation results can accurately judge the state of the network and provide some reference opinions for the network planning.