A New Robust Heuristic for Assigning Substrate Network Resources to Virtual Networks

Actually used in the Cloud-Computing concept, network virtualization is a promising way to share distant resources, offer multiple services and architectures to multiple heterogeneous Virtual Networks (VNs) on a common Substrate Network (SN). The main challenge in the VN embedding problem is the efficient mapping of virtual nodes and links to the substrate network resources, avoiding network congestion and responding as well to the desired Service Level Agreement (SLA). Faced with this NP-Hard problem, research focused on designing algorithms, processes, and heuristics which had separate node's allocation and links allocation, where the load balancing on the substrate components is not reached. This paper provides a heuristic solution based on A-Star algorithm. The heuristic relies on constraints of both nodes and links, and aims to maintain a load balancing upon the SN. The solution is gradually built following this heuristic function values with the possibility of back-tracking to explore the maximum of solution components before selecting the suitable allocation. To validate the proposed solution, we carry out several experiments and compare our results with some classical result's works.