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.

中文翻译：

---

智能反射表面辅助物理层安全：深度学习方法


物理层安全 (PLS) 是一种轻量级、无密钥的信息理论安全方法，有助于在无线网络中实现机密通信。然而，由于其保密能力和覆盖范围限制，PLS 在 5G 及以后的无线通信中的应用很少。为了使 PLS 在 5G 及其他系统中更具吸引力，出现了解决这些限制的新技术。最广泛采用的 PLS 范例之一是智能反射面 (IRS)，它基于传播环境重新配置概念。 IRS辅助PLS的核心是刻意提高合法信道的质量，同时降低窃听信道的质量。在这项工作中，首先概述了 IRS 辅助的 PLS，然后讨论了未知瞬时窃听 CSI 的潜在解决方案。介绍了几种 IRS 辅助 PLS 波束成形技术，并提供了案例研究，以证明 IRS 和深度学习方法的联合应用所带来的增强的安全性能。最后，将未解决的问题确定为未来的研究方向。