Secure Task Offloading for Multi-layer Aerial-Offshore MEC Networks With Imperfect Eavesdropper Information

This letter proposes a secure edge computing scheme for multi-layer aerial-offshore networks where source unmanned aerial vehicles (SUAVs) with limited onboard resources offload computation-intensive tasks to either a high-altitude platform or a shore-based station. To safeguard wireless transmissions against multiple surface eavesdroppers with uncertain locations, a mobile jammer UAV (JUAV) dynamically emits artificial noise. The JUAV trajectory, power allocation, and offloading decisions are jointly optimized to minimize the total system latency subject to secrecy rate, mobility, and power constraints. An iterative algorithm combining successive convex approximation and the cross-entropy method is developed to solve the resulting mixed-integer nonlinear programming problem. Simulation results demonstrate that the proposed scheme reduces latency by 16.6%-28% compared to baselines while incurring only an 11% performance penalty for security protection.