Physical Layer Security for NOMA Systems: Requirements, Issues, and Recommendations

Saeid Pakravan, Jean-Yves Chouinard, Xingwang Li, Ming Zeng, Wanming Hao, Quoc-Viet Pham, Octavia A. DobreIEEE Internet of Things Journal 2023 · DOI

TL;DR

This paper provides a comprehensive survey of Physical Layer Security (PLS) within Non-Orthogonal Multiple Access (NOMA) networks, a key candidate for 6G. It explores how NOMA’s unique features—like interference cancellation and power domain multiplexing—can be leveraged alongside emerging technologies like Reconfigurable Intelligent Surfaces (RIS) and Ambient Backscatter (AmB) to protect data from both internal and external eavesdroppers.

What Problem Does This Paper Address?

As we move toward 6G, the demand for spectral efficiency has led to the rise of Non-Orthogonal Multiple Access (NOMA). Unlike traditional methods that give each user their own "lane" (frequency or time slot), NOMA allows multiple users to share the same resource block by stacking them in the power domain.

However, this efficiency comes with a massive security headache. Because signals are superimposed, a legitimate user (or an eavesdropper) receives everyone’s data. While high-level encryption exists, it is computationally expensive and vulnerable to future quantum attacks. The authors address the challenge of implementing Physical Layer Security (PLS)—which uses the inherent randomness of the wireless channel (noise, fading, and interference)—to ensure that even if a signal is intercepted, it cannot be decoded by unauthorized parties.

Methods

This is a systematic review and state-of-the-art analysis. The authors categorize the existing body of research into four primary scenarios:

  1. Eavesdropper Type: Distinguishing between internal eavesdroppers (legitimate users who try to peek at others' data) and external eavesdroppers (unauthorized outsiders).
  2. Relay-Assisted Systems: Evaluating how Decode-and-Forward (DF) and Amplify-and-Forward (AF) relays impact secrecy.
  3. RIS-Aided NOMA: Analyzing how "smart mirrors" (Reconfigurable Intelligent Surfaces) can proactively shape the wireless environment to favor legitimate users over attackers.
  4. Ambient Backscatter (AmB): Investigating security in ultra-low-power IoT devices that communicate by reflecting existing RF signals.

The paper evaluates these systems using standard PLS metrics: Secrecy Outage Probability (SOP), Ergodic Secrecy Capacity (ESC), and Secrecy Coverage Region (SCR).

Datasets

As this is a comprehensive survey and theoretical analysis paper, it does not utilize a specific training dataset or introduce a new experimental benchmark. Instead, it synthesizes results from over 100 cited research works, comparing their mathematical models and simulation outcomes.

Key Insights and Ideas

  • The SIC Vulnerability: In NOMA, users perform Successive Interference Cancellation (SIC). This means a "strong" user must decode a "weak" user's signal to remove it. This inherent feature makes internal eavesdropping a unique and difficult challenge in NOMA compared to traditional systems.
  • Interference as a Shield: While interference is usually the enemy, NOMA can use it as a deterrent. The non-orthogonal signals make it harder for an external eavesdropper to distinguish the intended user’s signal from the "noise" of other users.
  • RIS as a Game Changer: Reconfigurable Intelligent Surfaces allow operators to "program" the environment. By adjusting the phase shifts of an RIS, a system can create a "constructive" signal at the legitimate receiver and a "destructive" (null) signal at the eavesdropper’s location.
  • Hardware Impairments: Interestingly, the paper notes that hardware imperfections like In-phase and Quadrature-phase Imbalance (IQI) can sometimes improve security by adding a layer of unpredictable distortion that only the legitimate receiver (with proper channel state information) can effectively mitigate.

Key Results

  • Full-Duplex vs. Half-Duplex: Research shows that Full-Duplex Relays (FDR) generally provide a lower Secrecy Outage Probability than Half-Duplex Relays (HDR) because they can simultaneously receive data and transmit "Artificial Noise" (AN) to confuse eavesdroppers.
  • Power Allocation (PA) is Critical: The most effective way to bolster PLS in NOMA is through optimized power allocation. By carefully choosing the power coefficients ($\alpha_k$), the Base Station can ensure the legitimate user’s signal is strong enough to decode while keeping it below the decodable threshold for an eavesdropper.
  • Artificial Noise (AN): Injecting noise specifically designed to land in the "null space" of the legitimate user's channel is a highly effective strategy for degrading an eavesdropper's signal-to-noise ratio (SNR).

Limitations and Future Work

The authors highlight several "open frontiers" for NOMA security:

  • STAR-RIS: Most current research looks at RIS that only reflect. Future work needs to address Simultaneous Transmitting and Reflecting (STAR) surfaces, which provide 360-degree coverage but also 360-degree eavesdropping risks.
  • Correlated Fading: Most models assume channels are independent. In reality, if an eavesdropper is physically close to a legitimate user, their channels are correlated. New mathematical tools (like copula theory) are needed to model this.
  • AI and Machine Learning: The paper suggests that Deep Reinforcement Learning (DRL) will be essential for real-time, adaptive security in dynamic 6G environments, while Federated Learning (FL) could allow devices to train security models without sharing raw, sensitive data.
  • Transmission Disturbances: Current studies often ignore practical issues like partial Channel State Information (CSI) and random phase noise, which become much more significant in the Terahertz (THz) bands planned for 6G.

Summary updated

BibTeX
@article{pakravan2023physical,
  author = {Saeid Pakravan and Jean-Yves Chouinard and Xingwang Li and Ming Zeng and Wanming Hao and Quoc-Viet Pham and Octavia A. Dobre},
  title = {{Physical Layer Security for NOMA Systems: Requirements, Issues, and Recommendations}},
  journal = {IEEE Internet of Things Journal},
  year = {2023},
  doi = {10.1109/jiot.2023.3296319},
}
Figure