Joint Optimization of Computation Offloading and Power Allocation Based on DRL in Secure LEO Satellite Edge Computing System

Low earth orbit (LEO) satellites can be equipped with mobile edge computing (MEC) servers to form LEO satellite edge computing system to provide global services for ground users. In order to solve eavesdropping attacks in LEO satellite edge computing system, this paper designs a secure LEO satellite edge computing system, which uses deep reinforcement learning (DRL) to obtain offloading decisions and convex optimization algorithm to obtain sub-optimal power allocation scheme. Simulation results show that the proposed algorithm has excellent convergence effect and is superior to other comparative algorithms.