International Conference on Computer and Knowledge Engineering

Home / 11th International Conference on Computer and Knowledge Engineering

‎Virtual Network Embedding based on ‎Univariate ‎Distribution ‎Estimation‎‎‎

Authors :

Arezoo Jahani¹

1- Assistant Professor

Keywords :

‎‎Virtual network ‎embedding,‎ Substrate ‎provisioning,‎ ‎incremental ‎learning,‎ ‎‎univariate ‎EDA,‎ ‎Distribution ‎estimation.‎

Abstract :

‎Network virtualization has been extensively used in the next generation of internet architecture‎. ‎Substrate provisioning is a challenging problem in ‎Virtual Network Embedding (VNE) which provides the infrastructure of various virtual elements on a similar substrate/physical infrastructure‎. Previous works usually ‎complete two phases of node mapping and link mapping to embed ‎Virtual Networks (VN) ‎and ‎try to check all possible scenarios ‎in virtual network embedding‎‎. ‎This paper uses the ‎estimation ‎of ‎distribution‎ algorithm ‎(EDA) ‎to embed virtual networks on the shared ‎Substrate Network (SN)‎. ‎‎Discrete EDAs can be divided into three groups: ‎univariate, ‎bivariate, and ‎multivariate‎. Since in the simplest case there is no relation between the selected substrate nodes, we have chosen the ‎univariate ‎EDA‎. In the coding method, we examined the mapping of nodes and links together, so we can easily use a univariate EDA. By removing the link mapping from coding, the multivariate EDA should be used. ‎‎‎EDA ‎has ‎incremental ‎learning ‎ability ‎which ‎makes ‎it‎ able to learn the distribution of the best embedding solutions ‎with high revenue and less cost ‎to generate the best ‎answers‎ in the next generations‎. ‎The proposed method completes two phases of node mapping and link mapping concurrently‎. ‎The evaluation results show the effectiveness of the proposed method in convergence and high acceptance ratio and high revenue in the VNE process ‎in the comparison with Presto and ACO-‎VNE.‎‎‎