Hybrid Bio-Inspired Heuristics for the Robust Network Slice Design Problem

Abstract

We consider the task of provisioning a generic network slice request on a given physical substrate network infrastructure-a problem that arises in the context of next generation networks-under traffic uncertainty, with the objective of minimising the capital and operational expenditures incurred to accommodate the network slice. As the resulting formulation can be hard to tackle using commercial MIP solvers even for problem instances of moderate size, we devise a hybrid biased-random key genetic algorithm to solve the robust network slice design problem. Finally, we present a performance evaluation of the proposed solution methodologies using realistic datasets from SNDlib [1].