Research On Virtual Network Mapping Algorithm Based On Multi-objective Optimization

The number of Internet users has increased dramatically in the short time,because of the internet greatly facilitates people's work and life.As a result,the traditional network has been greatly impacted and unable to meet all application requirements.In order to accommodate more users and not reduce the access speed of network resources,network virtualization(Network Virtualization,NV)technology came into being which quickly received extensive attention.Virtual network mapping problem(Virtual Network Mapping Problem,VNMP)is a very important part in the process of network virtualization.Specifically,it is based on the va rious needs of users,build s virtual network request(Virtual Network Request,VNR)and maps it to a limited shared physical network.Most virtual network mapping issues are based on a single objective.However,various problems need to be considered in reality.Virtual network mapping is no exception.Multi-objective optimization of virtual network mapping has become an important issue in recent years.When a goal is optimized,it does not guarantee that other goals can be optimized and sometimes it even reduces the performance of other goals.Such an optimization method cannot naturally guarantee a comprehensive optimization solution.So we need to consider a compromise method in which each objective function can be optimized simultaneously.The so lution set like this is called pareto optimal set(Pareto-optimal Set).In the multi-objective optimization problem,with the increase of the number of target,the difficulty of Pareto entropy approximation to the optimal solution is also increasing.In addition,it is worth noting that with the continuous expansion of the size of network users,the physical network is prone to failure due to excessive load,and the stability of the network is a factor that must be considered.This article focuses on the above issues.In order to obtain a better resource mapping solution,the following inquiry is made:1)In order to improve the optimization speed and accuracy rate,a comprehensive optimized virtual network mapping scheme is obtained,and the network acceptance rate is improved.A virtual network mapping algorithm(MOPSOVNE)based on multi-objective particle swarm optimization is proposed.Firstly,MOPSOVNE introduced crossover operator in the basic particle swarm algorithm and expanded the search space of population optimization.Secondly,non-dominant sort and crowded distance are introduced in the multi-objective optimization algorithm to speed up the convergence rate of the population.Finally,a multi-objective particle swarm optimization algorithm is used to solve the problem of virtual network mapping by minimizing the cost and the node load balancing.The experimental results show that the algorithm which would solve the problem of virtual network mapping has advantages in the network request acceptance rate,average cost,average node load balance,and infrastructure providers' benefits.2)In order to balance the resource load of the physical network,improve the stability of the network,and further increase the acceptance rate of virtual requests,a multi-objective particle swarm virtual network mapping algor ithm(GRC-mopso-VN E)based on global resource capabilities is proposed.The algorithm refers to the existing new metric GRC(measuring the potential mapping ability of nodes),which makes the load of physical network more balanced,and greatly improves the network acceptance rate,and achieves higher returns.