Multi-Objective Optimization Model And Algorithm For Virtual Network Function Service Chain Deployment

As a key new infrastructure to support the digital transformation of domestic economy and society,5G/6G network promotes the vigorous development of the fourth Industrial revolution,which is mainly characterized by digitalization,networking and intelligence.As one of the key technologies of 5G/6G communication,network function virtualization technology aims to realize network functions in a software-based way to achieve on-demand distribution of network resources and flexible loading of software,reduce the deployment cycle of network services,improve the flexibility of the network and thus greatly enhance the flexibility and intelligence of network management.How-ever,this technology also brings challenges to resource allocation(computing resources,storage resources,network bandwidth resources,etc.).The construction of efficient re-source allocation scheme(mapping of service function chain)has become the core problem of network function virtualization.This paper mainly studies under the background of network function virtualization,and its innovation is mainly reflected in the following two aspects:(1)To tackle the problem of service chain mapping of virtual network functions with fixed sequence of virtual network functions,a multi-objective constrained optimization model was established to minimize network delay,network energy consumption and maximize link load balancing and network node load balancing.In order to solve the model,an efficient coding method and a population initialization method based on uniform distribution were designed,and a selection strategy was proposed considering the uniformity,diversity and convergence of constraint violation degree and settlement.On this basis,an improved multi-objective brainstorming algorithm based on MOEA/D framework is proposed to obtain a virtual network service chain mapping scheme.The simulation results show that the proposed method can obtain the front surface with better uniformity,diversity and convergence,and the network delay and energy consumption of the network are reduced by 5% and 7.8% compared with the comparison algorithm.The performance of link load balancing and network node load balancing is improved by 10.2% and 8.5%compared with the comparison algorithm.(2)Aiming at the problem of service chain mapping of virtual network functions with flexible function order,a four-objective constrained optimization model was established,an efficient coding method was designed to determine the optimal implementation order and migration of virtual network functions,and a new individual generation algorithm and selection operator were proposed to solve the multi-objective optimization model.The simulation results show that the proposed method can obtain a virtual network function mapping scheme with better network delay,network energy consumption,link load balance and network node load balance,and provide more decision schemes for decision makers.The designed algorithm has some problems such as insufficient resources of functional nodes,unbalanced load and long queuing time in the network nodes,so the network performance can be improved by adjusting the sequence of virtual network functions and reasonable migration.