Tongxin xuebao (Jun 2016)
Time and energy aware virtual network embedding using Bayesian theory analysis
Abstract
Aiming at the energy consumption problem in virtual network embedding,a virtual-network-embedding algorithm was proposed by combining the time and energy aware.Taking the running time during the evaluation of physical nodes and physical paths into account,it considered multiple factors which included the processing time of physical devices,and used probability theory to help analyze the selected probability of each available physical node for a virtual node.During the selection of substrate nodes,the factors of remaining resources,the increment of CPU utilization,the switch state and the amount of extended time of physical nodes were considered.The theory of conditional probability was further used to analyze the importance of available nodes.The factors of the switch state,the amount of extended time and the length of physical paths were also considered.The proposed approach could effectively map the current virtual network request onto a smaller set of nodes and links which are switched on,and also the devices which have less amount of extended time.Experimental results show that the proposed approach has better performance,and can effectively decrease energy consumption comparing with the methods without taking the time factor into consideration.
