Abstract
This letter investigates a secure uncrewed aerial vehicle (UAV)-enabled over-the-air computation (AirComp) network, where a UAV acts as a mobile fusion center (FC) to aggregate data from multiple ground sensors. At the same time, an eavesdropper (EVE) attempts to intercept the transmitted signals. To ensure secure and efficient computation, we aim to jointly optimize the UAV's receiver coefficients, the sensors' transmitter scaler factors, and the UAV trajectory to minimize the mean square error (MSE) at the UAV, while satisfying the MSE constraint at the EVE, the maximum transmit power constraint, and the UAV trajectory constraint. Given the non-convexity of the formulated problem, we propose an efficient algorithm that leverages alternating optimization combined with the successive convex approximation (SCA) technique. Numerical results validate the superiority of our proposed scheme over benchmark approaches.