Research On Resource Allocation Algorithm For Network Virtualization

In the scenario of network virtualization,the development of network virtualization technology,network function virtualization technology and software-defined network technology make the allocation of network resources more flexible.The demands of network operators or users for network and network functions are gradually changing to the demands for virtual network and virtual network functions.In this case,the research on resource allocation algorithm of virtual network embedding and virtual network functions forwarding graph embedding has gradually become an important research direction for people to study the problems in the network domain.The problem of virtual network embedding and virtual network function forwarding graph embedding is mainly based on the needs of users or network operators,that is,in the form of virtual network request and virtual network function forwarding graph.The corresponding physical network resources are allocated in the substrate network.In this process,it is necessary to allocate network resources reasonably according to certain allocation decisions,so as to make efficient use of physical network resources and improve the benefits of network resources.Therefore,this paper studies the problem of virtual network embedding and virtual network function forwarding graph embedding,and puts forward the corresponding algorithm to efficiently allocate network resources.The main research contents of this paper are as follows:(1)The research background,significance and challenges of resource allocation problem for network virtualization were described.The research status,related technologies and research methods of the research challenges were also introduced.(2)A kind of virtual network embedding algorithm based on reinforcement learning Q-learning was proposed.The virtual network embedded process could be modeled as a Markov decision process by defining the state sets of the reinforcement learning,action sets and a reward function.This algorithm could train the agent by constantly rewarding for high quality work,accumulating total rewards,maximizing the long-term revenues.At the same time,based on the event-driven simulation environment,the process of virtual network embedding was simulated,and the proposed algorithm was compared with an advanced heuristic algorithm,which proved the effectiveness of the proposed algorithm.(3)A new algorithm based on subgraph matching for virtual network function forwarding graph embedding was proposed.The optimal embedding decision was obtained by combining nodes and links at the same time through subgraph matching.Meanwhile,the process of virtual network function forwarding graph embedding was simulated in the simulation environment,and the proposed algorithm was compared with the same type of heuristic algorithm,and the experimental analysis proved the effectiveness of the proposed algorithm.(4)According to the technical methods of resource allocation algorithm,the research was summarized.At the same time,the future research of virtual network embedding and virtual network function forwarding graph embedding was prospected.