Intelligent Reflecting Surface Assisted Physical Layer Security: A Deep Learning Approach

Physical layer security (PLS) is a lightweight, keyless, and information-theoretical security approach, which helps to enable confidential communications in wireless networks. However, the application of PLS in 5G and beyond wireless communications is rare due to its secrecy capacity and coverage limits. To make the PLS more appealing in 5G and beyond systems, new techniques emerged to address the limitations. One of the most widely adopted PLS paradigms is intelligent reflector surface (IRS), which works on a propagation environment reconfiguration concept. The core of IRS-assisted PLS is to deliberately enhance the quality of legitimate channels, while degrading eavesdropper channels simultaneously. In this work, IRS-assisted PLS is overviewed first, followed by a discussion on potential solutions with unknown instantaneous wiretap CSI. Several IRS-assisted PLS beamforming technologies are introduced with a case study presented to demonstrate enhanced security performance due to joint application of IRS and deep learning approach. Finally, open issues are identified as future research directions.