Physical Layer Security in Near-Field Communications for 6G: A Survey

Abstract

Extremely large-scale antenna arrays (ELAAs) and terahertz-band operation are pushing the boundary of the electromagnetic near field outward to tens or even hundreds of meters, bringing spherical-wavefront propagation into the operating regime of everyday sixth-generation (6G) links. This shift introduces a new spatial degree of freedom -- range, alongside angle -- that fundamentally changes how confidentiality can be enforced at the physical layer. Where far-field physical layer security (PLS) can only discriminate users by direction, near-field PLS can additionally discriminate by distance, enabling secrecy even when a legitimate receiver and an eavesdropper share the same line of sight. This survey consolidates the fast-growing literature on near-field PLS, organizing it around five thrusts: (i) beamfocusing-based secure transmission, which exploits range-dependent energy concentration; (ii) artificial-noise, directional-modulation, and wavefront-domain techniques; (iii) reconfigurable-intelligent-surface (RIS/STAR-RIS)-assisted near-field security; (iv) convergence with non-orthogonal multiple access (NOMA), rate-splitting multiple access (RSMA), and integrated sensing and communication (ISAC); and (v) emerging enablers, namely movable/fluid antennas and learning-based optimization. We also summarize fundamental secrecy-performance-limit results that quantify exactly how much the near-field range dimension improves on far-field secrecy capacity. A companion Chinese-language survey published in the Journal of Electronics & Information Technology in late 2025 covers similar ground; this article is intended as a complementary, English-language synthesis that additionally incorporates results published through early 2026, and it is organized around a threat-model-first taxonomy rather than a scenario-first one. We close by identifying open problems in channel modeling, security-performance-energy tradeoffs, hybrid near/far-field coexistence, and the near-total absence of experimental (as opposed to simulation-only) validation in this literature.