Robust Computation Offloading and Trajectory Optimization for Multi-UAV-Assisted MEC: A Multiagent DRL Approach

Abstract

For multiple Unmanned-Aerial-Vehicles (UAVs) assisted Mobile Edge Computing (MEC) networks, we study the problem of combined computation and communication for user equipments deployed with multi-type tasks. Specifically, we consider that the MEC network encompasses both communication and computation uncertainties, where the partial channel state information and the inaccurate estimation of task complexity are only available. We introduce a robust design accounting for these uncertainties and minimize the total weighted energy consumption by jointly optimizing UAV trajectory, task partition, as well as the computation and communication resource allocation in the multi-UAV scenario. The formulated problem is challenging to solve with the coupled optimization variables and the high uncertainties. To overcome this issue, we reformulate a multi-agent Markov decision process and propose a multi-agent proximal policy optimization with Beta distribution framework to achieve a flexible learning policy. Numerical results demonstrate the effectiveness and robustness of the proposed algorithm for the multi-UAV-assisted MEC network, which outperforms the representative benchmarks of the deep reinforcement learning and heuristic algorithms.

Trajectory-Aware Offloading Decision in UAV-Aided Edge Computing: A Comprehensive Survey

“…Figure 5 illustrates a detailed taxonomy of the optimization algorithms used for trajectory-aware offloading decision making in UAV-aided MEC. , AO [86], PDD [20], JSORT [87], BCD [88], DQN [15], DDQN [89], DDPG [90], MADDPG [91], MAPPO [92], MO-AVC [13], and GNN-A2C [93].…”

“…In the context of trajectory-aware offloading in UAV-aided MEC, the SCA is employed to iteratively solve complex optimization problems, such as resource allocation, trajectory planning, and offloading decision-making processes [94]. In this technique, the non-convex problem is broken down into a series of convex subprob- [57], AO [86], PDD [20], JSORT [87], BCD [88], DQN [15], DDQN [89], DDPG [90], MADDPG [91], MAPPO [92], MO-AVC [13], and GNN-A2C [93].…”

“…MAPPO is an extension of the PPO algorithm, designed to handle scenarios with multiple interacting agents. In [ 92 ], MAPPO was used as a vigorous offloading strategy in an edge computing network. The scenario involves the collaboration of numerous UAVs to serve several user equipment (UE) devices.…”

“… Classification of trajectory-aware offloading decision algorithms: SCA [ 57 ], AO [ 86 ], PDD [ 20 ], JSORT [ 87 ], BCD [ 88 ], DQN [ 15 ], DDQN [ 89 ], DDPG [ 90 ], MADDPG [ 91 ], MAPPO [ 92 ], MO-AVC [ 13 ], and GNN-A2C [ 93 ]. …”

Robust Offloading for Edge Computing-Assisted Sensing and Communication Systems: A Deep Reinforcement Learning Approach

“…However, due to the unpredictability of computation types and the complexity of tasks, there is an inherent uncertainty in the computation processes of edge networks [ 8 , 9 ]. To address these challenges, a robust design scheme has been developed for the MEC network, which specifically accounts for the uncertainties associated with task complexity.…”