Fairness-Aware Offloading and Trajectory Optimization for Multi-UAV Enabled Multi-Access Edge Computing

Abstract

Multiple unmanned aerial vehicles (UAVs) can compensate for the performance deficiencies of a single UAV in multi-access edge computing (MEC) systems, thus providing improved offloading services to user equipments (UEs). In multi-UAV enabled MEC systems, the offloading strategy and UAVs' trajectories affect the fairness of both UEs and UAVs, which affects the UE experience and UAVs' existence durations. Therefore, we investigate fairness-aware offloading and trajectory optimization in the system. To ensure fairness of energy consumptions (ECs) for both UEs and UAVs, we minimize the weighted sum of the maximum EC among UEs and the maximum EC among UAVs subject to the task delay, the offloading strategy and UAVs' trajectories constraints. Despite the non-convexity of the original formulated joint optimization problem, we transform the problem into two sub-problems and solve them one by one. Finally, an iterative optimization algorithm is proposed to alternately optimize the offloading strategies and the UAVs' trajectories. Simulation results show that the proposed algorithm can effectively reduce both the maximum EC among UEs and the maximum EC among UAVs and ensure the fairness of both the UEs and UAVs. INDEX TERMS Fairness, multi-access edge computing, multiple UAVs, trajectory optimization, offloading optimization.

Network partition detection and recovery with the integration of unmanned aerial vehicle

“…DEVIPS optimizes the locations of stop points using Differential Evolution (DE), which avoids special crossover and mutation operators and employs 2‐D search space for individuals. Diao et al 49 focused on enhancing the performance of multi‐access edge computing (MEC) systems by employing multiple UAVs. The proposed approach targets trajectory optimization to assure equitable energy consumption (EC) for UAVs and user equipments (UEs).…”

A Multi-UAV Assisted Non-orthogonal Multiple Access based Relay System for Minimal Receiving Rate Maximization

“…Seid et al [12] proposed model-free deep reinforcement learning based collaborative computation offloading and resource allocation scheme to learn effective computational offloading strategies to minimize task execution latency and energy consumption. Diao et al [13] ensured fairness of energy consumptions(ECs) with other UAVs and user equipments at the same time as the computing services provided by the UAVs. The system minimized the weighted sum of the maximum EC between UAVs and the maximum EC between UAVs which can reduce the difference in energy consumption between the UEs and UAVs to ensure the fairness of both them.…”

A comprehensive survey on aerial mobile edge computing: Challenges, state-of-the-art, and future directions

“…Recently, there has been some work providing solutions to the above-mentioned challenges for multi-UAV aerial MEC systems. For the user association, [54,55,56] tackle this problem from different perspectives. Specifically, J. Zhang et al [54] investigate the computation efficiency maximization for a multi-UAV-enabled aerial MEC system, where the user association, CPU frequency allocation, power and spectrum resources, as well as trajectory scheduling of multiple UAVs are jointly optimized by an iterative algorithm with a double-loop struc-ture.…”

“…Then, existing algorithms, such as the branch-and-bound (B&B) and cutting plane methods, can be applied to solve the ILP problem. Different from [54], X. Diao et al [55] propose a greedy-based offloading strategy variable rounding (GOSVR) algorithm to obtain a near-optimal integer solution for the user association. Then, the UAVs' trajectories are optimized to minimize the weighted sum of the maximum energy consumption among users and among UAVs.…”

“…Specifically, if there is a no-fly zone or obstacle in UAV's flying range, the UAV needs to keep a safe distance with them [54]. Moreover, the multiple UAVs are required to keep a safe distance with each other to avoid collision [55]. To avoid the collision between UAVs and obstacles, in [51], it is assumed that multiple UAVs fly at different fixed altitudes, and thus they could not interfere with each other.…”