Multi-Scale Hierarchical Resource Management for Wireless Network Virtualization

Abstract

To deal with the conflict between limited radio resources and dramatic growth of wireless traffic, wireless network virtualization (WNV) is proposed as an efficient network sharing solution in 5G era. In this paper, we propose a multi-time-scale hierarchical model for WNV, which consists of primary users (PUs) with low latency requirement, secondary users (SUs) with high data rate requirement and tertiary users (TUs) without strict QoS requirement, such as machine-to-machine communications. The resource allocation scheme is decomposed into inter-slice subchannels pre-assignment in large time period and intra-slice subchannels and power scheduling in small time slot. In large time period, the subchannels pre-assignment problem is formulated as an integer optimization problem considering the requirement of PUs, SUs, and TUs. In small time slot, the subchannels and power scheduling of PUs is formulated as a mixed optimization problem with integer variables, for which we adopt Lyapunov drift-plus-penalty function with tradeoff factor ${V}$ to transform the problem into a Lyapunov optimization problem and propose a two-step algorithm consisting of a heuristic subchannel assignment procedure and a fast barrier-based power allocation procedure. SUs access spectrum in a static manner and TUs access spectrum in a cognitive way. Simulation results show that our proposed algorithm achieves great performance in terms of users’ performance with almost linear computation complexity.