Research On Resource Allocation Optimization Technology And Its Application In Data Center Optical Network

With the advent of the 5G era,the types of network services and transmitted data traffic are showing a rapid growth trend.The traditional wavelength division multiplexing network has been unable to meet the current needs of people due to its low utilization rate of spectrum resources.The flexible optical network technology has become one of the main technologies of current network transmission due to its more flexible bandwidth resource allocation.The optical network based on the data center is the focus of current attention.This thesis mainly starts from the purpose of improving the utilization of data center optical network resources and reducing the transmission blocking rate,and uses theoretical analysis and numerical simulation methods to conduct in-depth research on the optimization strategy of data center optical network spectrum resource allocation.In view of the low utilization rate of the internal transmission resources of the optical network link in the data center,the thesis proposes a Markov chain-based spectrum defragmentation evaluation model,and introduces an approximate Markov model to calculate the RBP before and after the defragmentation.The change of RBP before and after sorting is used as an evaluation index,which can measure the degree of optimization of spectrum resource utilization by spectrum fragmentation.The simulation experiment results show that the spectrum defragmentation technology can effectively improve the utilization of spectrum resources and reduce the transmission blocking rate.Compared with the traditional accurate Markov model,the approximate Markov model can greatly reduce the model calculation when calculating RBP.Complexity,increase flexibility.Furthermore,in order to solve the problem of insufficient resource utilization due to the transmission link topology,the thesis proposes a data center spectrum resource allocation optimization strategy based on machine learning assistance.Optimize the transmission link topology and link transmission strategy by combining machine learning technology and improved genetic algorithm,and introduce improved genetic algorithm to reconstruct the topology of the transmission link before transmission,and solve the problem of resource utilization decline caused by topology structure;At the same time,machine learning technology is used to classify the traffic transmitted in the data center during transmission,and different types of traffic are mixed and transmitted to prevent a large number of similar traffic from being transmitted on the same link,causing waste of resources.The simulation experiment results show that the introduction of machine learning technology and improved genetic algorithm in data center transmission can effectively improve the problem of low spectrum resource utilization due to topology structure,and reduce transmission delay and blocking rate.