IEEE Access (Jan 2021)
Antenna Grouping-Based Blind Artificial Noise for TDD Secure Transmission
Abstract
The artificial noise (AN-) aided techniques have been considered as promising candidates to enhance the physical layer security. However, the effect of AN on an eavesdropper (Eve) can be severely degraded as the number of antennas at Eve increases. To resolve this problem, we propose an antenna grouping-based blind artificial noise (AG-BAN) to prevent Eve from being aware of the existence of AN in multiple-input, single-output, and multi-antenna eavesdropper (MISOME) and multiple-input, multiple-output, and multi-antenna eavesdropper (MIMOME) channels. In the proposed AG-BAN, antennas at a legitimate transmitter (Alice) are classified into two antenna groups, which are used to transmit the data and AN, respectively. In the antenna group for transmitting the AN, the null-space-based AN scheme is employed, which makes the AN be nullified at a legitimate receiver (Bob). Any reference signals are not transmitted from the antenna group assigned to transmit the AN. As a result, Eve cannot perceive the AN, which makes the AG-BAN be effective even if there is a sufficient number of antennas at Eve. In the proposed AG-BAN, the estimated secrecy capacity at Alice is used for the antenna grouping criterion, which is derived based on the Eve’s channel statistics without the instantaneous channel information between Alice and Eve. Furthermore, to reduce the computational complexity of the exhausted search based optimal antenna grouping, the proposed AG-BAN employs one-by-one antenna search procedures. The simulation and numerical results show that the proposed AG-BAN scheme effectively improves the secrecy capacity compared to the conventional AN schemes and achieves the near-identical secrecy capacity compared to the exhausted search based optimal AG-BAN scheme.
Keywords