Traffic-Aware Virtual Network Embedding With Joint Load Balancing and Datarate Assignment for SDN-Based Networks

Network virtualization is one of the key technologies to enable network slicing scenarios for the efficient resource management of beyond 5G and 6G networks. Mapping algorithms, such as Virtual Network Embedding (VNE), have been proposed to manage such a challenging scenario due to heterogeneous traffic requirements. Despite the numerous VNE contributions, the literature is currently lacking VNE implementation platforms, which can exploit real-time traffic statistics and jointly optimize the routing and resource assignment. In general, VNE solutions foresee complex optimizations with worst-case resource dimensioning, without considering the real-time traffic statistics. In this paper, we propose a statistics-collection aware (SCA) link mapping algorithm for VNE, named SCA-VNE. SCA-VNE is formulated as a Mixed Binary Linear Programming (MBLP) problem which jointly minimizes the load balancing and the data rate assignment to each Virtual Network Request (VNR). VNRs are differentiated based on priority level, i.e., tolerated queuing delay and user satisfaction probability. Due to the exponential complexity of SCA-VNE, we propose a relaxed version, named SCA-R, to significantly reduce the computation time. We show via testbed experimental results that, compared to four baseline schemes, the proposed algorithm increases the acceptance ratio up to 11% and drastically minimizes the average queuing delay, in highly-loaded scenarios.