Secure Multi-UAV AirComp With Joint Transceiver and Trajectory Optimization

Abstract

This letter considers a secure multi-uncrewed aerial vehicle (UAV) enabled over-the-air computation (AirComp) system, where multiple UAVs cooperatively transmit data to a ground fusion center (FC) via AirComp under eavesdropping threats. To achieve reliable and secure aggregation, we jointly optimize the UAV transmit power, FC denoising factor, and UAV trajectories to minimize the mean square error (MSE) at the FC while enforcing the eavesdropper (EVE) MSE, UAV power, and mobility constraints. The formulated problem is non-convex due to coupled variables and mobility constraints. An efficient algorithm combining alternating optimization and successive convex approximation (SCA) is proposed to obtain a stationary solution. Numerical results verify the effectiveness of the proposed scheme and its superiority over benchmark schemes.