Research On Dynamic Resource Allocation In OLT Pooling

With the development of network,the new services represented by cloud computing,Internet of things,Internet of vehicles,virtual reality and video conference are emerging,bringing new opportunities and challenges to the development of optical access network.The traditional access network architecture cannot fit with the dynamic services,which leads to the mismatch between the bandwidth demand and supply,resulting in the problem of bandwidth fragmentation.When the granularity of the bandwidth resource required by the service is large,the bandwidth utilization efficiency of the access network will be seriously affected.In view of the above bandwidth fragmentation problem,this paper studies the architecture of the next generation flexible optical access network,and based on the architecture,this paper also studies the dynamic resource allocation schemes in different situations.Specifically,the innovation of this paper are as follows:1.A bandwidth defragmentation mechanism based on genetic-tabu search algorithm is proposed.In order to solve the problem that part of the wavelengths cannot meet the bandwidth requirement of a new user/service,from the perspective of bandwidth fragmentation,this paper proposes an ONU dynamic migration with wavelength sharing mechanism under the above mentioned flexible optical access network architecture,to implement the integration of the bandwidth fragments in the network.By defragmenting,it helps to provide more bandwidth resource for a new user/service,so as to improve the service passing rate and the network bandwidth utilization.This mechanism uses the genetic-tabu search algorithm,which can find the optimal solution quickly in a huge solution space.Under the assumed conditions,the simulation results show that the algorithm above can reduce the degree of bandwidth fragmentation of the TWDM PON system effectively,within the acceptable migration loss range.When no extra wavelength is added,the maximum reduction is about 38%.When adding additional wavelength,the maximum reduction is about 78%with the help of wavelength sharing mechanism.2.An ONU optimized migration mechanism based on greedy-genetic algorithm is proposed.This mechanism is mainly aimed at the phenomenon of the insufficient bandwidth resource supply caused by the sudden large amount of bandwidth demand under a certain wavelength in OLT pooling.From the perspective of bandwidth fragmentation,an ONU dynamic migration mechanism is proposed under the above-mentioned architecture to try to provide a large amount of bandwidth resource required by the burst large-scale services,so as to reduce the delay problem caused by traffic congestion,and improve the total bandwidth utilization efficiency in OLT pooling.The typical scenario of the above phenomenon is railway scenario.In view of the special requirements of the algorithm time consumption in the railway scenario,this mechanism uses the greedy-genetic algorithm.Compared with the common genetic algorithm,the satisfaction rate of bandwidth of the above-mentioned algorithm is increased by about 3 1%under the assumed conditions.At the same time,the average convergence generations of the above-mentioned algorithm to reach the nearly optimal solution is only about 1/4,which greatly accelerates its convergence speed to give the ONU migration scheme in a relatively short time.Also,compared with the mechanism without defragmentation,this mechanism can improve the satisfaction rate of bandwidth by up to about 56%at most.Compared with the random migration mechanism,this mechanism can greatly increase the satisfaction rate by about 68%at most.And Compared with the no migration mechanism,this mechanism can greatly improve the satisfaction rate by about 88%at most.