A Fast, Scalable Meta-Heuristic for Network Slicing Under Traffic Uncertainty

Abstract

Perceived to be one of the cornerstones of the emerging next generation (5G) networks, Network slicing enables the accommodation of multiple logical networks with diverse performance requirements on a common substrate platform. Of particular interest among different facets of network slicing is the problem of designing an individual network slice tailored specifically to match the requirements of the big-bandwidth next generation network services. In this work, we present an exact formulation for the network slice design problem under traffic uncertainty. As the considered mathematical formulation is known to pose a high degree of computational difficulty to state-of-the-art commercial mixed integer programming solvers owing to the inclusion of robust constraints, we propose a meta-heuristic based on ant colony optimisation algorithms for the robust network slice design problem. Experimental evaluation conducted on realistic network topologies from SNDlib reveals that the proposed meta-heuristic can indeed be an efficient alternative to the commercial mixed integer programming solvers.