Dynamic Dispersion Compensation And Elastic Resource Management Mechanism In Data Center Optical Network

With the massive amount of information produced every day and the explosive growth of various data,the society has stepped into the information age and the era of big data has come.To meet the demand of emerging businesses brought by big data,cloud services,artificial intelligence,Internet of Things and 5G services.the next generation of data center optical network needs to develop in the direction of high speed,high scalability,high reliability,high energy efficiency,flexibility and low time delay.The application of SDN to the resource management of data center optical network.relying on SDN controller to realize the centralized control and comprehensive scheduling of the whole network resources,is an optimal solution to satisfy the emerging business demand of data center optical network.In this paper,dynamic dispersion compensation and elastic resource management mechanism in data center optical network are described in detail.A tunable dispersion compensator based on lithium niobate waveguide grating is designed to compensate the chromatic dispersion.Combining machine learning with existing spectrum algorithms is proposed to improve the efficiency of spectrum resource utilization.The main work and innovation of the paper are as follows:(1)The perception of chromatic dispersion,one of the damages in data center optical network.The chromatic dispersion values at different transmission distances of the same wavelength and at different wavelengths of the same transmission distance are perceived respectively.(2)A tunable dispersion compensator based on lithium niobate waveguide gratings is proposed to compensate chromaticity dispersion.The dispersion compensation system is built,and the dispersion compensator is used to compensate the chromatic dispersion in the optical network.The simulation results of BER and constellations show that the designed tunable dispersion compensator based on lithium niobate waveguide grating has a good effect on chrominance dispersion compensation.(3)In the part of elastic resource management mechanism,SDN controller is used to realize the centralized control of data center optical network resources.On this basis,this paper proposes an improved spectrum allocation algorithm combined with machine learning--MLF.MLF first uses k-means algorithm to classify business requests.Then when a certain business arrives,it matches and classifies the business with the previous classification,and then selects the corresponding spectrum allocation algorithm according to different categories.The strategy of this algorithm focuses on the comprehensive consideration of the convergence time and the spectrum blocking rate.That is to reduce the convergence time while ensuring the lowest possible blocking rate so as to maximize the efficiency of spectrum utilization.The algorithm is verified by matlab simulation,which can effectively reduce the convergence time and blocking rate of the algorithm,improve the bandwidth utilization efficiency and the system's capacity for elastic businesses.