Low-Latency Mobile Virtual Reality Content Delivery for Unmanned Aerial Vehicle-Enabled Wireless Networks With Energy Constraints

In this paper, we leverage the benefits of unmanned aerial vehicle (UAV)-enabled wireless networks in providing high link qualities and on-demand services to achieve low-latency mobile virtual reality content delivery. Taking the limited energy storage at the UAVs into consideration, we formulate an average latency minimization problem, subject to the constraints on the UAV's kinematic, computation, and transmission capabilities. Utilizing the Lyapunov optimization approach, we first transform this NP-hard problem into a weighted-latency-plus-energy minimization problem for each time slot. Then, we further decompose the transformed problem into three subproblems, namely, the UAV's trajectory design subproblem, the UAV's processing frequency allocation subproblem, and the UAV's transmission power control subproblem. An iterative algorithm is proposed to solve these three subproblems. To reduce the complexity, we propose a low-complexity algorithm that designs the UAV's trajectory with the target of flight energy minimization. Through numerical results, we show that, compared to other benchmarks, our proposed algorithms can achieve a good trade-off between the average latency and the energy consumption by judiciously tuning the Lyapunov control parameter.

I. Introduction

Mobile virtual reality (VR) has become a popular service, since it can provide an immersive experience. However, in order to satisfy the quality-of-service (QoS) of diverse VR services, such as VR live streaming and VR game, low latency and high transmission rate are required for mobile VR content delivery [1]. In addition, a large amount of traffic data and the VR users' mobility make it increasingly difficult to achieve low-latency mobile VR delivery. Against this background, significant research attention has been paid on improving the latency performance of mobile VR content delivery [2], [3], [4], [5]. In [2], the optimal spectrum allocation for both uplink and downlink transmissions was studied to reduce the end-to-end latency, in which the VR users are classified into different social communities and served based on such classifications. In [3], a view caching placement problem was studied to minimize the download latency and an online view synthesis-based caching algorithm that reduces the distance between two cached adjacent views was proposed. In [4], the closed-form expression of end-to-end latency in mobile VR video streaming systems was analyzed, and its dependence on the caching, computation, and transmission resources was discussed. The authors of [5] investigated a joint caching placement and offloading optimization problem to minimize the system latency and the energy consumption for the VR video delivery, in which a trade-off between the energy efficiency and the latency reduction was revealed.