A Secure Task Offloading Scheme for UAV-Assisted MEC with Dynamic User Clustering and Cooperative Jamming: A Method Combining K-Means and SAC (K-SAC)

In the unmanned aerial vehicle (UAV) assisted edge computing system, the broadcast characteristics of the UAV signal, the high mobility of the UAV, and the limited airborne energy make the task offloading strategy face challenges such as increased risk of information disclosure, limited computing resources, and the trade-off between energy consumption and flight time. To address these issues, we propose a K-means in-depth reinforcement learning algorithm based on Soft Actor-Critic (SAC). The proposed method first leverages the K-means clustering algorithm to determine the optimal deployment of ground jammers based on the final distribution of mobile users. Then, building upon the SAC framework, the Cross-Entropy Method (CEM) global sampling strategy is incorporated into the action output phase to form the K-SAC algorithm. This algorithm aims to maximize system rewards, which holistically balance task offloading delay, energy consumption, and secure offloading rate. Consequently, it jointly optimizes the optimal hovering positions of auxiliary UAVs and the task offloading ratio for each user, leading to an overall performance improvement in system security and efficiency. Finally, compared with current schemes, the system benefits achieved by the proposed scheme were 9.83% higher on average in different computing task sizes, 13.67% higher on average in different task complexities, and 14.63% higher on average in different interference powers.