Optimal resource sharing in multi-tenant 5G networks

Abstract

The potential price for enabling network slicing in multi-tenant virtualized mobile networks is the underutilization of the scarce wireless and/or network resources. One way to increase overall network utilization would be to allow inter-tenant sharing, i.e., sharing of resources between different network slices. To this end, this work tries to shed further light into this issue by discussing different possible degrees of network sharing together with the associated linear integer mathematical programs that allows to investigate upper bounds on the achievable performance improvement. Furthermore, in order to realize real time and adaptive sharing of resources a scale-free heuristic is also presented that is amenable for real-time implementation. Based on the depth of the aforementioned multi-tenant sharing, we propose a sharing scheme with two options named Tight Coupling (TX) and Loose Coupling (LX). Under the 3GPP baselines and certain assumptions, a set of numerical investigations have been carried out demonstrating significant gain in aggregated network throughput and per user throughput compared to traditional fully isolated network slicing method.