A Parallel Link Mapping for Virtual Network Embedding with Joint Load-Balancing and Energy-Saving

Virtual Network Embedding (VNE) is an emerging area of Telecom Networks, in an era where the physical capacity of the substrate network is pushed to the limits in order to get the maximum achievable performance. Link mapping optimization is one of the two sub-problems which the VNE is subdivided into. The objective of the paper is to demonstrate the efficiency of a parallel approach with respect to the sequential one, in particular when there is a high level of heterogeneity among the Virtual Networks to be embedded, such as network slicing scenario. In this regard, this paper investigates the significance of a parallel approach in effectively implementing VNE. Specifically, we focus the attention to the link mapping, considering the node mapping is already known a priori. In contrast to the trend in the literature to use Genetic Algorithms (GAs) for a parallel computation, this paper proposes a novel and efficient parallel link mapping algorithm for enabling VNE in realistic system scenarios. Two important objectives are considered, load-balancing and energy-saving, and the presented results demonstrate the superiority of the proposed parallel approach over the sequential one in terms of these objectives, at the expenses of the computing time. Furthermore, the scalability of the proposed algorithm is evaluated over a range of substrate network sizes.