Robust Secrecy-Throughput Efficiency Maximization for UAV-Assisted Mobile Edge Computing under Imperfect CSI

Abstract

UAV-assisted mobile edge computing (MEC) is a promising solution for computation-intensive and securitysensitive applications, especially in infrastructure-limited environments. However, most existing secure UAV-MEC designs assume perfect channel state information (CSI), which is unrealistic in practice due to channel estimation errors, quantization noise, and feedback delays. In this paper, we study robust secrecy-throughput efficiency (STE) maximization for a dual-UAV secure MEC system under imperfect CSI, where a computing UAV (UAV-C) collects computation tasks from multiple ground users and a jamming UAV (UAV-J) interferes with a passive eavesdropper. To account for CSI uncertainty, we adopt bounded multiplicative error sets and formulate a worst-case robust optimization problem that jointly optimizes user transmit power, task offloading allocation, and UAV trajectories. To solve the resulting non-convex fractional program, we develop an iterative algorithm based on Dinkelbach’s method, block coordinate descent (BCD), and successive convex approximation (SCA). Simulation results show that the proposed robust design achieves a provable worst-case STE guarantee under channel uncertainty, while incurring only marginal averageperformance loss compared with the perfect-CSI baseline.