Multi-UAV Cooperative Task Offloading and Resource Allocation in 5G Advanced and Beyond

In 5G advanced and beyond, latency-critical and computation-intensive applications require more communication and computing resources. However, remote areas without available terrestrial edge/cloud infrastructure fail to satisfy these applications’ demands. This motivates the emergence of the UAV-enabled aerial computing paradigm. Single UAV-enabled aerial computing (SUEAC) is limited by small coverage area and insufficient resources, which cannot meet the application requirements. Multiple UAV-enabled aerial computing (MUEAC) has broken through the limitation of SUEAC and has attracted wide attention. Cooperation among multiple UAVs in MUEAC can fully utilize UAV resources and achieve load balancing. Furthermore, for divisible tasks with data-dependent characteristics, using partial offloading makes task scheduling more flexible compared to binary offloading, thus reducing task processing delay. Therefore, we propose a software defined networking enhanced cooperative MUEAC system. To minimize the processing delay of divisible tasks, we study the problem of joint task scheduling and computing resource allocation under task data dependency and UAV energy consumption constraints. To solve the non-convex problem, a multi-UAV cooperative communication and computing optimization (MCCCO) scheme is proposed. Experimental results corroborate that MCCCO can achieve better performance in task processing delay reduction and load balancing on UAV energy consumption than the traditional schemes.

I. Introduction

In 5G advanced and beyond, more and more Internet of Things (IoT) applications emerge in our lives, such as automatic driving, face detection, etc [1]. Due to their computation-intensive and latency-sensitive nature, these applications often demand more communication and computing resources. Mobile edge computing (MEC) can be used to resolve conflicts between these applications and resource-constrained user devices, and can greatly reduce task processing delay and energy consumption [2]. However, because edge servers are typically embedded in fixed base stations (BSs) or access points (APs) close to user devices, they cannot handle computing tasks in temporary or emergency scenarios (e.g., remote areas without available terrestrial edge/cloud infrastructure, emergency rescue sites, etc.) [3], [4], [5]. Recently, owing to the strong adaptive ability, fast movement rate, and low deployment cost of unmanned aerial vehicles (UAVs), a new diagram of UAV-enabled aerial computing (UEAC) has attracted increasing attention [6]. By embedding computing servers in UAVs or utilizing UAVs as mobile relays, UEAC can provide flexible communication, computing, and cache services for ground devices [7]. Compared with conventional MEC, UEAC has many advantages. First, UAV platforms can be quickly deployed to specific locations to meet user demands for computing resources due to UAVs’ flexibility, mobility, and ease of deployment when edge computing servers are overloaded or unavailable [8]. Second, taking advantage of the line-of-sight (LoS) property of air-ground links, UEAC can deliver higher data rate and significantly minimize energy consumption and task processing delay, so as to guarantee user quality of service (QoS) [9].