Research On Complex Topology Construction Mechanism And Performance Evaluation In Multidimensional Optical Interconnection Network

In recent years,with the rise of cloud computing and big data services,internet concepts and technologies have undergone revolutionary changes,which put forward higher and higher requirements for the network communication of data centers.Optical interconnection technology,especially multi-dimensional and multi-granularity all-optical switching technology,is regarded as the most promising network technology for cloud service under this background,relying on its low power consumption and abundant bandwidth resources and other advantages.At the same time,in order to adapt to the traffic characteristics of data center with high burst and high difference,it is very important to design an efficient topology reconstruction algorithm.Therefore,based on OpenScale,a small-world optical Internet architecture for data centers,this paper proposes a topology construction method for changing and different distribution network traffic,so as to give full play to the flexibility and reconfiguration of optical switching and improve the performance of cloud services.The main contents of this paper are expressed as follows:Firstly,this paper introduces the design principle,networking mode and node structure of OpenScale architecture.On the basis of this architecture,the topology construction method without considering traffic distribution is discussed at first.A traffic-independent deterministic wavelength connection topology generation algorithm is proposed,the simulation results show that this algorithm can reduce the average path length of the topology;A traffic-related dynamic threshold-based topology reconstruction mechanism is proposed.The simulation results show that this mechanism can effectively improve the delay and throughput performance.Secondly,the topology construction method considering traffic distribution is discussed.A topology reconstruction mechanism based on hot spot area identification is proposed,by identifying hot spot areas,the network congestion can be judged more accurately,so as to reduce the impact of regional congestion on network performance;Traffic-prediction and traffic-identification topology reconstruction mechanism is proposed,in this paper,traffic prediction is abstractly understood as the prediction of time series,through traffic prediction,can adapt the topology reconstruction to the latest traffic demand,traffic identification is used to identify whether the traffic is global or regional traffic,topology is clipped according to traffic identification results,can save the OCS connection number,raise the utilization ratio of wavelength connection.