Physical Layer Security (PLS)

• Physical Layer Security is an information-theoretic framework that leverages wireless channel randomness to secure communications independently of computational assumptions.
• It employs technologies like reconfigurable intelligent surfaces and massive MIMO to dynamically control channel features for optimal secrecy rates and key extraction.
• Challenges such as pilot contamination, CSI spoofing, and jamming are addressed through adaptive channel control and sensing-driven strategies.

Physical Layer Security (PLS) is an information-theoretic paradigm that exploits intrinsic randomness, reciprocity, and spatial decorrelation in wireless channels and/or radio-frequency hardware to guarantee confidentiality, authentication, and integrity—independently of computational assumptions. Unlike classical cryptography, which relies on algorithmic hardness, PLS derives security from the physical properties of the wireless medium, enabling direct protection of data, keys, and access credentials in practical communication systems. In next-generation networks, new channel features and technologies—such as reconfigurable intelligent surfaces (RIS), massive MIMO (mMIMO), and joint sensing/communication—dramatically expand the operational space for PLS, while introducing novel challenges regarding channel control, feature selection, adversarial attacks, and robust adaptation amid non-stationarity and hardware impairment (Kihero et al., 2022).

1. Fundamental Theory and Security Metrics

PLS builds on the wiretap channel model, formalizing information-theoretic secrecy via the secrecy capacity: $C_s = [I(X;Y_b) - I(X;Y_e)]^+$ where $X$ is the transmitted signal, $Y_b$ and $Y_e$ are the observations at the legitimate receiver (Bob) and eavesdropper (Eve), respectively, and $I(\cdot;\cdot)$ denotes mutual information (Kihero et al., 2022). The system achieves perfect secrecy if Bob's channel provides strictly more information about $X$ than Eve's. Alternative key metrics include secrecy outage probability (probability that $C_s<R_0$ for a target rate $R_0$), coherence parameters (coherence time $T_c$, bandwidth $B_c$), and entropy rate $H(h) = -\int p(h) \log_2 p(h) dh$, which quantifies the randomness available per channel use.

PLS leverages:

• Channel-generated shared randomness for secret key generation.
• Adaptive beamforming, coding, or artificial noise aligned with instantaneous channel state.
• Channel- or environment-induced spatial/temporal decorrelation to guarantee irreproducibility of legitimate channel observations by Eve.

2. Channel Features Enabling Enhanced PLS

2.1 Reconfigurable Intelligent Surfaces (RIS)

RISs are planar metasurfaces comprising $N$ individually-controlled elements, each applying a phase shift $\theta_i$ to impinging electromagnetic waves. The composite channel can be expressed as: $H_{\text{total}} = H_d + G\,\Phi\,H_r$ where $H_d$ is the direct MIMO channel, $H_r$ (Tx-RIS), $G$ (RIS-Rx), and $$\Phi = \text{diag}(e^{j\theta_1},...,e^{j\theta_N})$$ (Kihero et al., 2022). By tuning $\{\theta_i\}$, RIS affords fine-grained environmental control—reshaping multipath profiles, inducing rapid spatial/temporal decorrelation, and maximizing constructive interference at Bob while imposing destructive patterns at Eve.

2.2 Massive MIMO (mMIMO)

A typical mMIMO channel ($H \in \mathbb{C}^{M \times K}$, with $M \gg K$) exhibits:

• High spatial resolution yielding angular-domain randomness (AoA/AoD).
• Element-dependent statistics from spherical wavefronts and cluster-based visibility regions.
• Rich null-space structure, exploited for spatially targeted artificial noise injection. Increasing $M$ strengthens spatial uniqueness and the secrecy rate, offering resilience to eavesdropping in beamspace (Kihero et al., 2022).

2.3 Sensing-Enabled Randomness

Joint radar-communication or RF sensing exploits measured environment parameters—scatterer positions, velocities, material properties—to introduce novel entropy sources (range/Doppler profiles, angular clustering, temporal mobility). This increases key extractable randomness and diversity, improving both secrecy rate and key agreement reliability (Kihero et al., 2022).

3. Criteria and Metrics for Channel Feature Selection

Selecting appropriate channel features for PLS involves quantifying five orthogonal properties:

• Randomness (high entropy): measured via coherence time ($T_c \approx 0.423/f_D$, where $f_D$ is maximum Doppler shift), bandwidth ($B_c \approx 1/5\sigma_\tau$), and spatial correlation decay ($\rho(d)$).
• Spatial/Temporal Uniqueness (decorrelation): Ensures Eve’s channel is statistically independent at distinct locations, governed by coherence distance ($d_c \approx \lambda/2$ in rich scattering).
• Reciprocity: Necessary for secure key extraction, leveraging time-symmetric CIRs ($h_{AB}(t)\approx h_{BA}(t)$ within coherence time).
• Accessibility/Observability: Low estimation error, high reliability in feature acquisition.
• Irreproducibility: Resistance to Eve's emulation or camouflage of legitimate channel states.

Alongside these, practical channel features must be robust against feature-specific estimation errors and susceptible to spatially selective control.

4. Channel Control and Sensing for PLS

4.1 Channel Control Mechanisms

• Baseband Diversity (e.g., Cyclic Delay Diversity): Increases delay spread, inducing frequency-selective randomness.
• Reconfigurable Antennas (RA): Dynamically alter radiation patterns to stimulate independent fading.
• mMIMO Beamwidth Control: Selects clusters with maximal entropy or spatial uniqueness.
• RIS-based Control: Realizes programmable LoS/NLoS paths and environment shaping for tailored secrecy rates.

4.2 Sensing-Driven Adaptation

Sensing technologies (radar, LiDAR, computer vision) supplement pilot-based CSI estimation, providing environmental side-information. This enables informed adaptation—such as beam steering away from scatterers near Eve—and improves metrics relating to uniqueness and irreproducibility (Kihero et al., 2022).

5. Adversarial Attacks Targeting Channel Characteristics

PLS is fundamentally vulnerable to attacks on channel estimation and physical-channel integrity. Principal attack vectors include:

• Pilot Contamination: Eve transmits identical pilots, distorting Alice’s channel estimate ($\hat{H}$), boosting Eve’s mutual information and reducing secrecy capacity.
• CSI Inference (Snooping): Eve leverages MU-MIMO frame structure to glean legitimate precoding and derive channel knowledge.
• Channel Spoofing (Camouflage): Eve manipulates her own RIS or RF front-end to synthesize a channel nearly indistinguishable from the legitimate one ($h_{AE} \approx h_{AB}$), capturing shared randomness during key extraction.
• Jamming / Artificial Noise Attacks: Eve injects interference to degrade Bob’s SINR and exploits blind source separation (ICA) to recover information, eroding secrecy capacity ($C_s$).
• Mathematical Effects: Under jamming, Bob’s SINR is

$$\text{SINR}_b = \frac{P |h_{AB}|^2}{\sigma^2 + P_j |h_{JB}|^2}$$

impacting achievable information rates and secrecy (Kihero et al., 2022).

6. Research Directions and Open Challenges

PLS continues to evolve with the introduction of programmable, high-entropy, and context-rich channel features. Key research avenues include:

• Channel-Feature Integrity: Robust PLS frameworks that resist CSI spoofing, RIS-mediated reciprocity attacks, and feature perturbations.
• Mobility and Non-Stationarity: Key extraction and secrecy adaptation methods tolerant to rapid channel aging (V2X, high-speed train scenarios).
• Beam-Squint Robustness: Ultra-wideband systems face AoA/AoD reciprocity loss; new squint-compensating PLS designs are required.
• Cross-Layer Security Intelligence: Fusion of physical-, MAC-, and application-layer data (e.g., radio environment maps) using machine learning for dynamic feature selection.
• Joint Sensing-Control PLS: Co-optimization of RIS phase profiles and sensing schedules to counter real-time attacks and maximize secrecy rates under operational constraints.

Rigorous future work will focus on the intersection of physical randomness, environmental programmability, and multi-domain adaptation to ensure scalable, attack-resilient confidentiality in diverse wireless use cases.

---

Foundation text: (Kihero et al., 2022) (Revisiting the Wireless Channel from Physical Layer Security Perspective)