Computing-Assisted Task Offloading and Resource Allocation for Wireless VR Systems

Abstract

Wireless virtual reality (VR) has become an entry point for 5G applications. The widespread application of wireless VR in people's production, life, and so on, requires joint utilization of mobile network resources. However, the high dynamics of multi-dimensional resources bring severe challenges to improve the quality of experience (QoE) of wireless VR users. This challenge can be largely addressed by a recent network architecture known as mobile edge computing (MEC) network, which enables caching and computing at the edge of mobile networks. In this paper, we propose a wireless VR-oriented task offloading and resource management scheme which jointly considers caching, computing and spectrum allocation to minimize the content delivery delay under the promise of QoE requirements. To solve the problem, we decompose the original problem into two complex integer nonlinear programming (INLP) sub-problems. To solve the INLP problem, a proactive caching algorithm is proposed to determine the VR video's caching status. Then, a graph coloring method based on a greedy strategy is proposed to solve the remaining problem, which includes computation offloading and RB allocation decision. In the simulations, the delivery delay and the offloading capacity are evaluated. Simulation results demonstrate the effectiveness of the proposed scheme.