Research On Swarm Intelligence Based Virtual Network Embedding Algorithm

With the increasing variety and quantity of new Internet services,the current Internet architecture has become increasingly rigid,which is difficult to meet the diversified needs of users.Network virtualization is one of the important technologies to overcome the gradual fossilization of network.To solve this problem,we need to consider the constraints of many different resources,which makes virtual network embedding NP hard.Most of the existing research results uses node ranking and mathematical programming algorithm,and the embedding scheme is simple.The collaboration mechanism of swarm intelligence algorithms can produce a variety of mapping schemes,which can effectively solve this problem and provide a new idea for the study of virtual network embedding.In this thesis,different factors affecting the mapping are combined with swarm intelligence algorithm to study the virtual network mapping.The main work completed is as follows:(1)Proposed Teaching-Learning-Based Optimization for Virtual Network Embedding(TLBO-VNE).The algorithm simulates the real class teaching scene,taking the product of node local centrality,node residual resources and node adjacent bandwidth as the learning ability of each subject of the class,and the revenue-cost ratio as the grade of the students.In the teaching stage,the average grade of students in the class is improved,and in the mutual learning stage,the grade of students themselves is improved.Experimental results show that TLBO-VNE algorithm performs better in terms of request acceptance rate and revenue-cost ratio.(2)Proposed Consultant Guided Search algorithm for Virtual Network Embedding(CGS-VNE).In this algorithm,the indexes such as node cpu,node degree and adjacent bandwidth resources are taken as the strategy of the consultant respectively.During the process of building the embedding scheme,the client will consult multiple consultants,and the consultant will give suggestions to the client according to the strategy.CGS-VNE uses the consultant-guided search mechanism to ensure the diversity of embedding schemes,fully excavates the solution space,and obtains high-quality embedding schemes through iteration.Simulation is carried out for three scenarios of low resource demand,medium resource demand and high resource demand of virtual network.The simulation results show that the cgs-vne algorithm has high request acceptance rate and revenue-cost ratio.