中文摘要：

针对无线数字通信系统中人工噪声方法可被多天线窃听者破解的问题，该文提出有限字符输入下物理层安全传输的一个充分条件，并以此为指导设计了一种类符号人工噪声方法。分析表明，人工噪声方法下，有限字符输入信号和窃听者无噪接收信号之间的等效信道是一个离散有噪无损信道（Discrete Noisy Lossless Channel, DNLC）。由于DNLC输入信号的可逆性为窃密提供了必要条件，窃听者通过增加天线可使窃听信道容量达到合法用户的信道容量上限，致使系统的保密互信息为零，因此破坏输入信号的可逆性是有限字符输入下物理层安全传输的一个充分条件。类符号人工噪声方法满足这一充分条件，可以保证物理层安全传输，仿真结果也表明了该方法的有效性。

英文摘要：

Addressing the problem that the artificial noise method can be cracked by the eavesdropper with multiple antennas in wireless communication systems, a sufficient condition is proposed for secure physical layer transmission with finite alphabet inputs. Under this guideline, a signal-like artificial noise method is designed to ensure the system security transmission. Analysis reveals that the equivalent channel between the finite alphabet input and the eavesdropper’s noise-free output is a Discrete Noisy Lossless Channel （DNLC）. Since the reversibility of the input under a DNLC provides the necessary condition for eavesdropping, the eavesdropper can augment its antennas to successfully squeeze out the secure information, nullifying the system’s secrecy mutual information. As a result, destroying the reversibility of the input signal becomes a sufficient condition for the secure physical layer transmission with finite alphabet inputs. The signal-like artificial noise method satisfies the sufficient condition, which can ensure the secure physical layer transmission. Simulation results demonstrate the efficacy of this method.