An enhanced node mapping algorithm for VNE problem based on Path Relevance

摘要

Network virtualization has been proposed to be a promising way to solve the Internet ossification problem. One of the main challenges for network virtualization is the virtual network embedding (VNE) problem, which focuses on the efficiency of the substrate resources allocation and involves the virtual node and link mapping problems. Node mapping process has serious influence on the whole mapping performance. Existing node mapping algorithms cannot efficiently support complex virtual network mapping processes, especially virtual networks with mesh topologies. The reason relies on mutual interference between shortest paths for various virtual links that are determined by node mapping process. In this paper, we introduce Path Relevance (PR) which is used to decrease the mutual interference among shortest paths, and an ant colony optimization (ACO) algorithm is proposed for the node mapping process. Our node mapping algorithm is combined with various link mapping algorithms in simulation. Simulation results show that our enhanced node mapping algorithm can help the successive link mapping process to find more suitable paths for virtual links and improve the acceptance ratio of complex virtual networks. Increasing revenue is also gained in various combinations than existing G-SP and D-ViNE algorithms.