AirFed: A Federated Graph-Enhanced Multi-Agent Reinforcement Learning Framework for Multi-UAV Cooperative Mobile Edge Computing

TL;DR

AirFed tackles the challenge of coordinating multiple UAVs for cooperative mobile edge computing under dynamic IoT workloads. It introduces a dual-layer dynamic Graph Attention Network to capture spatial-temporal relations, a dual-Actor single-Critic constrained multi-agent RL framework for joint continuous trajectory and discrete offloading, and a reputation-based decentralized federated learning mechanism with gradient-sensitive adaptive quantization to reduce communication overhead. The approach achieves a 42.9% reduction in weighted cost, exceeds 99% deadline satisfaction, reaches 94.2% IoT device coverage, and cuts communication overhead by 54.5% relative to strong baselines, while maintaining robustness across varying system scales. These results demonstrate AirFed’s practical potential for scalable, QoS-aware, large-scale UAV-MEC deployments in real-world environments.

Abstract

Multiple Unmanned Aerial Vehicles (UAVs) cooperative Mobile Edge Computing (MEC) systems face critical challenges in coordinating trajectory planning, task offloading, and resource allocation while ensuring Quality of Service (QoS) under dynamic and uncertain environments. Existing approaches suffer from limited scalability, slow convergence, and inefficient knowledge sharing among UAVs, particularly when handling large-scale IoT device deployments with stringent deadline constraints. This paper proposes AirFed, a novel federated graph-enhanced multi-agent reinforcement learning framework that addresses these challenges through three key innovations. First, we design dual-layer dynamic Graph Attention Networks (GATs) that explicitly model spatial-temporal dependencies among UAVs and IoT devices, capturing both service relationships and collaborative interactions within the network topology. Second, we develop a dual-Actor single-Critic architecture that jointly optimizes continuous trajectory control and discrete task offloading decisions. Third, we propose a reputation-based decentralized federated learning mechanism with gradient-sensitive adaptive quantization, enabling efficient and robust knowledge sharing across heterogeneous UAVs. Extensive experiments demonstrate that AirFed achieves 42.9% reduction in weighted cost compared to state-of-the-art baselines, attains over 99% deadline satisfaction and 94.2% IoT device coverage rate, and reduces communication overhead by 54.5%. Scalability analysis confirms robust performance across varying UAV numbers, IoT device densities, and system scales, validating AirFed's practical applicability for large-scale UAV-MEC deployments.

Figures (7)

• Figure 1: System model showing multi-UAV cooperative edge computing serving 50 requests across three application domains with heterogeneous spatial distributions.
• Figure 2: Overall architecture of AirFed framework. The framework integrates dual-layer GATs for spatial-temporal modeling, dual-Actor single-Critic for hybrid decision making, and decentralized federated learning featuring reputation-based aggregation and gradient-sensitive quantization.
• Figure 3: Convergence analysis of AirFed and baseline approaches across key performance metrics.
• Figure 4: Quality of Service analysis of AirFed and baseline approaches.
• Figure 5: Ablation analysis of AirFed key components.