Energy-saving Virtual Network Embedding Algorithm Based On Topology Awareness

Network virtualization effectively solves the problem of network rigidity through the abstraction,isolation,and sharing of resources,and promotes the development of innovative business.With the emergence of communication-intensive businesses such as virtual reality and high-definition video and computation-intensive businesses such as big data and cloud computing,different embedding requirements are proposed for network links and nodes.If traditional businesses and emerging businesses are deployed without distinction,which reduces service quality and energy efficiency.Therefore,this paper studies the virtual network energy-saving embedding algorithm for different scenarios and the main work is as follows:(1)Aiming at the problems of more hops link embedding and scattered node embedding in the algorithm of deploying traditional business,the energy-saving virtual network embedding algorithm based on sliding region particle swarm is proposed,which is concerned with the distribution characteristics of network energy-saving regions and the coordinating role of topology information on the process of node and link embedding.When a virtual network request arrives,the backbone topology is extracted and the link embedding priority is set.The energy-saving area is found through swarm intelligence collaboration,where link and node embedding are performed alternately.Experimental results show that in traditional business scenarios,our algorithm improves the level of node aggregation and shortens the length of link embedding,thereby reducing the embedding cost and network energy consumption.(2)Aiming at the problem that the two-stage embedding algorithm ignores the structural characteristics between nodes and leads to the low success rate of deployment of communication-intensive business,the virtual network embedding algorithm via diffusion wavelet is proposed,in which the structure among nodes is analyzed by spectral graph theory.Through unsupervised learning for the node topology and comprehensive evaluating for topological proximity,connection density,and communication capacity,a node that contributes to the success rate of link embedding is selected preferentially.The experimental results show that in the communication-intensive business environment,the algorithm's embedding success rate and link load balance are improved significantly,which improves network throughput and reduces network energy consumption effectively.(3)Aiming at the problem of the low success rate of computing-intensive business deployment due to the limitation of the coordinated embedding algorithm on the nodes and the fixedness of the greedy strategy,the improved fish swarm based on diffusion wavelet algorithm for energy-saving virtual network embedding is proposed,in which the spectrum graph analysis is used to redefine the topological space and combination with the swarm intelligence collaboration theory.Due to using the loose constraint strategy to coordinate the node and link embedding process,the selection ability of the nodes with high computing power requirements is improved.Through the fish foraging,rear-end and clustering behaviors,the embedding strategy is found in the entire substrate network,thus exploring the influence of uncertain factors on node energy consumption effectively.Experimental results show that in computing-intensive scenarios,our algorithm improves the embedding success rate and node load balancing,thereby increasing the network throughput.In a non-saturated substrate network environment,our algorithm presents lower network energy consumption.