Unmanned Aerial Vehicles (UAVs) exhibit significant potential in enhancing the wireless communication coverage and service quality of Mobile Edge Computing (MEC) systems due to their superior flexibility and ease of deployment. However, the rapid growth of tasks leads to transmission queuing in edge networks, while the uneven distribution of user nodes and services causes network load imbalance, resulting in increased user waiting delays. To address these issues, we propose a multi-UAV collaborative MEC network model based on Non-Orthogonal Multiple Access (NOMA). In this model, UAVs are endowed with the capability to dynamically offload tasks among one another, thereby fostering a more equitable load distribution across the UAV swarm. Furthermore, the integration of NOMA is strategically employed to alleviating the inherent queuing delays in the communication infrastructure. Considering delay and energy consumption constraints, we formulate a task offloading strategy optimization problem with the objective of minimizing the overall system delay. To solve this problem, we design a delay-optimized offloading strategy based on the Twin Delayed Deep Deterministic Policy Gradient (TD3) algorithm. By jointly optimizing task offloading decisions and UAV flight trajectories, the system delay is significantly reduced. Simulation results show that, compared to traditional approaches, the proposed algorithm achieves a delay reduction of 20.2%, 9.8%, 17.0%, 12.7%, 15.0%, and 11.6% under different scenarios, including varying task volumes, the number of IoT devices, UAV flight speed, flight time, IoT device computing capacity, and UAV computing capability. These results demonstrate the effectiveness of the proposed solution and offloading decisions in reducing the overall system delay.

The demand for low-latency computing from the Internet of Things (IoT) and emerging applications challenges traditional cloud computing. Mobile Edge Computing (MEC) offers a solution by deploying resources at the network edge, yet terrestrial deployments face limitations. Unmanned Aerial Vehicles (UAVs), leveraging their high mobility and flexibility, provide dynamic computation offloading for User Equipments (UEs), especially in areas with poor infrastructure or network congestion. However, UAV-assisted MEC confronts significant challenges, including time-varying wireless channels and the inherent energy constraints of UAVs. We put forward the Lyapunov-based Deep Deterministic Policy Gradient (LyDDPG), a novel computation offloading algorithm. This algorithm innovatively integrates Lyapunov optimization with the Deep Deterministic Policy Gradient (DDPG) method. Lyapunov optimization transforms the long-term, stochastic energy minimization problem into a series of tractable, per-timeslot deterministic subproblems. Subsequently, DDPG is utilized to solve these subproblems by learning a model-free policy through environmental interaction. This policy maps system states to optimal continuous offloading and resource allocation decisions, aiming to minimize the Lyapunov-derived “drift-plus-penalty” term. The simulation outcomes indicate that, compared to several baseline and leading algorithms, the proposed LyDDPG algorithm reduces the total system energy consumption by at least 16% while simultaneously maintaining low task latency and ensuring system stability.

With the development of the Internet of Things (IoT), Unmanned Aerial Vehicle-Mobile Edge Computing (UAV-MEC) plays a critical role in enhancing network performance and data processing capabilities. However, significant challenges that impede system performance in UAV-MEC systems include task scheduling inefficiency, privacy leakage, and insufficient trust evaluation. To address these issues, we propose a novel framework that synergistically integrates a UAV trust evaluation model, Differential Privacy Federated Learning (DPFL), and the Deep Deterministic Policy Gradient (DDPG) algorithm. The core innovation lies in creating a security-aware DRL agent, where the trust model guides the DDPG’s decisions to inherently balance efficiency with security, while the entire collaborative learning process is protected by DPFL. By incorporating trust evaluation and privacy protection mechanisms, the proposed method achieves secure task scheduling with minimal latency under the condition of high trust values, while considering UAV energy constraints and successfully optimizing UAV trajectories. Simulation results demonstrate that, compared to traditional Deep Reinforcement Learning (DRL) algorithms, the proposed algorithm significantly reduces the cost by 4.12% and exhibits higher stability and convergence. Specifically, under varying conditions such as user counts, random task sizes, UAV computational capacities, and levels of UAV intrusion power, the costs decreased by up to 9.56%, 5.49%, 3.77%, and 4.43%, respectively. Overall, the proposed scheme exhibits significant improvements in user privacy protection, system efficiency, and adaptability.