Sensing-Assisted Secure Communication

• Sensing-assisted secure communication is an integrated ISAC framework that unifies radar sensing and wireless data transmission with embedded physical-layer security features.
• The approach employs diverse waveform designs—sensing-centric, communication-centric, and joint design—to balance resource allocation and optimize secrecy rates.
• Advanced optimization methods and hardware enablers like intelligent surfaces and active STAR-RIS address CSI uncertainties and conflicting system objectives.

Sensing-assisted secure communication is an emerging paradigm that unifies the functionalities of wireless communication and environmental sensing while embedding physical-layer security (PLS) features within a single hardware and spectral resource pool. In such architectures, the system leverages real-time sensing insights—such as the location or angular signatures of potential eavesdroppers—to optimize communication security, often outperforming classical approaches that consider sensing and security as separate modules. This integration introduces new technical challenges, including conflicting resource allocation between communication, sensing, and security objectives, as well as complex multi-objective optimization under hardware constraints. The field has matured rapidly, with foundational models, advanced algorithmic designs, and new threat models receiving considerable attention in both theoretical and practical directions.

1. Fundamental Design Principles

The core of sensing-assisted secure communication lies in the integrated sensing and communication (ISAC) framework, where a single waveform supports both radar sensing and wireless data transmission (Wei et al., 2021). ISAC waveforms can be classified as:

• Sensing-centric: Traditional radar/probing signals (e.g., pulse or frequency-modulated) are modulated to embed information, using schemes like pulse interval or index modulation. Communication rates are limited by radar-specific constraints (e.g., pulse repetition frequency) and in practice, applicable mainly in LoS scenarios.
• Communication-centric: Standardized communication signals (preambles, pilots) are exploited for dual use, relying on their favorable correlation properties for both data transfer and environment probing. For example, IEEE 802.11p/802.11ad waveforms can facilitate high-rate data while supporting basic ranging.
• Joint design: The waveform is optimized in a multi-objective fashion (e.g., joint SINR/SCNR optimization, secrecy-rate maximization), capturing the interplay and trade-offs between communication quality and radar estimation, subject to secrecy constraints. Formulations incorporate constraints such as:

$$\text{Secrecy Rate} = [\log_2(1 + \Gamma_\text{LU}) - \log_2(1 + \Gamma_E)]^+,$$

where $\Gamma_\text{LU}$ is the legitimate user's SINR and $\Gamma_E$ that of the (potentially adversarial) target. Optimization is typically non-convex with fractional and rank-1 matrix constraints.

This triad of design approaches underpins much of the modern ISAC/PLS literature and forms the starting point for security-oriented enhancements.

2. Security Challenges and Optimization Approaches

The integration of ISAC and PLS introduces unique, often contradictory challenges (Wei et al., 2021, Ren et al., 2022, Aman et al., 8 May 2025):

• Contradictory Objectives: Maximizing radar SCNR generally requires directing power at a target, but if the target is also an eavesdropper, this enhances their interception capability.
• Optimization Complexity: Multi-objective designs, with intersecting SINR and SCNR constraints plus physical-layer secrecy requirements, create highly non-convex formulations. Approaches such as semi-definite relaxation (SDR), S-procedure-based LMI transformations, and convex approximations (e.g., Bernstein-type inequalities) have been developed to obtain tractable solutions (Ren et al., 2022).
• CSI Uncertainty: Robustness to imperfect channel state information (CSI) at the eavesdropper is handled with deterministic (bounded error) models or probabilistic (Gaussian error) models. In both cases, the objective often becomes maximizing the worst-case secrecy rate or minimizing secrecy outage probability.
• Resource Allocation: The inherent dual-use of resources necessitates joint optimization of time allocation (for sensing vs. secure communication), beamforming, and artificial noise (AN) power. The resulting Pareto trade-offs are typically derived via global monotonic optimization or block coordinate descent (Xu et al., 2023).

3. Leveraging Sensing for Security Enhancement

ISAC's sensing capabilities provide enablers for advanced PLS strategies by conferring situational awareness unavailable in traditional settings:

• Eavesdropper Localization and Tracking: Sensing estimates of angular or range parameters (e.g., via Capon beamforming or approximate maximum likelihood methods) allow the system to shape beams and allocate AN or nulls toward eavesdropper directions (Su et al., 2022). The quality of such estimates is quantified by Cramér–Rao bounds (CRB).
• Beamforming and Artificial Noise (AN): With knowledge (even statistical) of eavesdropper locations, beamformers and AN can be adaptively designed to minimize the SINR at adversaries without sacrificing communication or sensing performance for legitimate users (Ren et al., 2022, Zou et al., 2023). The optimal allocation can be determined by iterative algorithms that jointly maximize secrecy rate and minimize estimation CRBs.
• Ambiguity Function (AF) Engineering: The exploitation of OFDM subcarrier power allocation enables the intentional creation of “ghost” targets in the ambiguity function, confusing unauthorized sensing receivers, while the legitimate receiver (with full waveform knowledge) applies reciprocal filtering to suppress artifacts at a modest SNR cost (Han et al., 2 Oct 2025).

4. Analytical Frameworks: Secrecy–Distortion Trade-offs

A substantial body of research uses information-theoretic models to characterize the fundamental trade-offs in secure ISAC (Günlü et al., 2022, Günlü et al., 2023, Welling et al., 24 Aug 2024):

• Secrecy–Distortion Region: For state-dependent broadcast (or wiretap) channels with feedback, the set of achievable tuples $(R_1, R_2, D_1, D_2)$ or $(R, D_1, D_2)$ quantifies the simultaneous rates at which public/secure messages and state (environmental) estimates can be obtained. Achievability is characterized by single-letter mutual information expressions and distortion bounds, often with auxiliary random variables to represent layered coding and one-time pad operations.
• Action-dependent Models: The inclusion of transmitter actions—e.g., beamforming vectors as action sequences affecting propagation states—extends the model to encompass adaptive physical-layer tactics. In such models, the secrecy-distortion region reflects the impact of active adaptation to feedback and environmental sensing (Welling et al., 24 Aug 2024).
• Joint vs. Separated Design: Joint communication–sensing–security designs (i.e., waveform and strategy co-design) strictly outperform separation-based counterparts, as rigorously demonstrated via binary examples (multiplicative Bernoulli state models, BEC/BSC mixtures), reflecting larger achievable secrecy-distortion regions.

5. Emerging Architectures and Hardware Enablers

Sensing-assisted secure communication is now harnessing advanced hardware solutions and network topologies:

• Intelligent Surfaces (IOS, RIS, STAR-RIS): Deployment of intelligent omni-surfaces (IOS) or reconfigurable intelligent surfaces (RIS), including simultaneous transmitting and reflecting RIS (STAR-RIS), enables dynamic manipulation of propagation characteristics. These surfaces allow joint reflection and refraction to enhance both communication and sensing in vehicular scenarios (Meng et al., 2022, Meng et al., 2022, Noh et al., 24 Jul 2025). Optimization of their phase shifts and power splitting ratios further improves throughput, beam alignment, and security in low-power and high-mobility environments.
• Active STAR-RIS Systems: By cooperating multiple amplification-enabled STAR-RISs, systems can achieve full-space coverage in NLoS environments, maximize sum-rate, and strictly enforce sensing and security constraints through jointly designing BS beamforming and surface coefficients, often via alternating optimization and convex relaxation (Noh et al., 24 Jul 2025).
• Quantum Techniques and Optical Secure Communication: New protocols leverage quantum entanglement—where message modulation is embedded in one half of an entangled photon pair and retrieved through exclusive measurement of the partner, while classical jamming beams mask the channel from unauthorized detectors (Sternberg et al., 22 Mar 2024). This physical-layer approach ensures that only the intended receiver can reliably access the message, while eavesdroppers are blinded by noise.

6. Defenses, New Security Metrics, and Threat Models

Modern ISAC security frameworks address an expanded set of threats (Aman et al., 8 May 2025, Xu et al., 28 Jun 2025, Ren et al., 2023):

• Sensing Security and LPI: Security considerations are extended to the sensing modality, where the objective is low probability of intercept (LPI) for the radar functionality. Adversaries may combine power detection (PD) and cyclostationary analysis (CA) to reveal system presence. Approaches such as enforcing noise-like cyclic spectrum and power equalization across subcarriers are designed to thwart such multi-mode attacks (Xu et al., 28 Jun 2025).
• Data Confidentiality, Covert Communication, and Sensing Spoofing: Beyond traditional eavesdropping, attacks may aim to detect the mere presence of a transmission (covertness) or to spoof the sensing process (inserting artificial echoes). Countermeasures involve randomizing transmission, optimizing hypothesis testing metrics (likelihood ratio tests), injecting authentication watermarks, and cross-layer co-design.
• Combined Communication–Sensing–Security Optimization: Modern frameworks optimize for joint metrics—such as secrecy rate, sensing estimation accuracy, and power consumption—using robust optimization, Pareto boundary characterization, and consensus (decentralized ADMM) algorithms, especially in networked ISAC settings (Xu et al., 18 Oct 2025).

7. Future Research Directions

Open questions and recommended directions include (Wei et al., 2021, Aman et al., 8 May 2025):

• Finite Blocklength and Low-Latency Regimes: Quantifying secrecy and estimation performance under practical code lengths and strict delay constraints typical of URLLC and V2X networks.
• Adaptive and Learning-based Security: AI-enabled beamforming/jamming, on-the-fly adaptation to environmental state and threat dynamics.
• Privacy-Preserving Sensing and Multi-functional ISAC: Protecting the privacy of radar parameters (e.g., in shared spectrum military scenarios) and generalizing security guarantees to both data and environmental information.
• System-level Protocols and Cross-layer Integration: Coordinating physical-layer security measures with MAC, networking, and application-layer policies, including support for reconfigurable surfaces and spectrum sharing.
• Robustness and Complexity: Ensuring secure ISAC is viable under hardware impairments, model mismatches, and with scalable, distributed optimization suitable for real-time implementation.

---

Sensing-assisted secure communication thus represents a mature, multi-faceted discipline in next-generation wireless network design, justified both theoretically via secrecy-distortion characterizations and practically by robust, joint waveform and hardware innovations. Substantial gains in spectral efficiency, resource savings, and security performance have been demonstrated, and future research is expected to push the boundaries on scalability, adaptability, and resilience against even more sophisticated adversaries.