Physical Layer Security Technology For Multiple Antenna System With Finite Alphabet Inputs

The wireless physical layer security aims at exploiting the spatial characteristic difference between the legitimate channel and eavesdropper’s channel, utilizes or creates the advantage of legitimate channel quality, and provides the information theoretical security for the communication system, which can assist the encryption system on the top layer of communication protocol. With the fast development and application of multiple-antenna technology, such as Multiple Input Multiple Output(MIMO), the spatial spectrum resource has been fully developed to provide more spatial freedom for the physical layer security, which can ensure both the reliability and efficacy for wireless communications. Furthermore, considering the wide application of wireless digital communications, where the digital modulated signals have finite alphabet characteristics, the dissertation mainly studies the physical layer security technology of the multiple antenna system with finite alphabet inputs.The finite alphabet characteristic of wireless digital communications incurs possible eavesdropping attacks. To eliminate the security problem, the dissertation bases on the information theory to optimize the system’s secrecy capacity with finite alphabet inputs from two aspects: One is developing the spatial characteristic difference, such as fading, spatial freedom, and noise, in order to minimize the eavesdropper’s channel capacity; the other is the linear precoding over MIMO channels in order to maximize the legitimate channel capacity. The main research contains:1. On the basis of information theory, the security of beamforming aided by artificial noise in the finite alphabet input system is analyzed. The dissertation points out that the channel between the finite alphabet input signal, i.e., secret signal, and eavesdropper’s noise-free received signal can be equivalent to a Discrete Noisy Lossless Channel(DNLC), which fails the secure method. From the viewpoint of geometrics, the eavesdropper’s noise-free received signal follows the distribution of hyper-plane. Its statistical characteristic can be figured out, leading to secrecy leakage. As a result, the dissertation indicates that a sufficient secure transmission condition for the finite alphabet input system is to destroy the eavesdropper’s DNLC architecture. According to the condition, the secret signal with randomly modulated amplitude/phase and signal-like artificial noise are designed to ensure the system security.2. An endogenic and masked space hopping secure transmission method is proposed based on the MIMO system. By establishing the mapping between transmitted symbols and receiving antenna indices, the method creates the space hopping pattern in a real-time way. It gets rid of dispensing with the space hopping pattern and greatly enlarges its application area. More importantly, the method develops the redundant wireless channel resource between the transmitter and receiver. By switching the spatial channel quickly and randomly, it effectively hides the finite alphabet characteristics of the secret signal and destroys the eavesdropper’s DNLC architecture, which ensures the system security.3. By extending the spatial modulation into the MIMO system, a spatial modulation physical layer security transmission method is proposed, which exploits the redundant wireless channel resource between the transmitter and receiver for the purpose of information transmission and ensuring security. The method divides the transmitted information into the two parts of channel modulation and signal modulation information, and establishes the mapping between digital modulated symbols and receiving antenna indices. During the transmission of signal modulation information, the receiving antenna indices are mapped to channel modulation information, which enhances the transmission efficiency. In addition, due to the randomness of channel modulation symbols, the method randomly switches wireless channels and destroys the eavesdropper’s DNLC architecture, which ensures the system security.4. The linear precoder of MIMO channels is designed to maximize the legitimate average mutual information, and improve the reliability of communication system. Since the mutual information is non-convex over the precoding matrix with finite alphabet inputs, the existing methods of linear precoder design is extraordinarily complicated. The dissertation improves the Space Time Linear Constellation Precoding(ST-LCP) method based on the channel diversity gain, and combines the method of maximizing the minimum distance between output signals to design a combined precoding algorithm. The algorithm can approximate the theoretically maximal mutual information under various channel conditions and Signal to Noise Ratio(SNR). Furthermore, it avoids recursion or exhaustive search, which is of much lower computational complexity.