Abstract
—In near-field communication (NFC) systems formed by high-frequency bands and extremely large-scale antenna arrays, signal leakage creates vulnerabilities that can be exploited by potential and dynamic eavesdroppers (Eves). Most existing privacy protection approaches rely on prior information, such as Eve's location or channel state information (CSI), which is difficult to obtain in practice. Furthermore, Eve can actively adjust its position to enhance its eavesdropping capabilities. Therefore, we propose a privacy protection approach for NFC systems, which can defend against potential dynamic Eves without prior information about Eve. Specifically, the base station (BS) employs an active obfuscation strategy, injecting deceptive signals while transmitting legitimate signals. To balance privacy protection performance and communication performance, a dynamic multi-objective optimization problem is formulated, which jointly optimizes the transmit power allocation and the spatial focusing position of the deceptive signal. Subsequently, we propose the multi-objective conditioned network (MOCN) algorithm, which learns a conditional policy to obtain the Pareto front. The simulation results show that, the proposed approach improves privacy protection performance by approximately 200% compared to the benchmark algorithm.