Physical-Layer Security in AmBC-NOMA Networks With Random Eavesdroppers

Xinyue Pei, Xingwei Wang, Min Huang, Yingyang Chen, Xiaofan Li, Theodoros A. TsiftsisIEEE Internet of Things Journal 2024

TL;DR

This paper tackles the challenge of securing Ambient Backscatter Communication Non-Orthogonal Multiple Access (AmBC-NOMA) networks against unknown, randomly distributed eavesdroppers. By introducing an artificial noise injection and an "Eve-exclusion zone," the authors develop analytical models and simulations to demonstrate enhanced physical-layer security, offering a crucial step towards more robust and private IoT networks.

What Problem Does This Paper Address?

The proliferation of Internet of Things (IoT) devices in next-generation mobile networks (6G) presents significant challenges in terms of connectivity and spectrum efficiency. While Non-Orthogonal Multiple Access (NOMA) offers a promising solution for massive connectivity, and Ambient Backscatter Communication (AmBC) provides an energy-efficient way for devices to communicate without dedicated spectrum, combining them (AmBC-NOMA) creates new vulnerabilities. Specifically, the broadcast nature of these networks makes them susceptible to eavesdropping. Traditional encryption methods are often impractical in high-density IoT environments due to latency and key management issues. Physical-layer security (PLS) offers an alternative by leveraging the inherent randomness of wireless channels. However, a key challenge in AmBC-NOMA networks is the presence of eavesdroppers (Eves) whose locations and numbers are unknown, posing a significant security threat. Existing research often assumes fixed Eve locations, which is unrealistic for passive eavesdroppers. This paper addresses this gap by investigating PLS in AmBC-NOMA networks with randomly distributed Eves, a more practical and challenging scenario.

Methods

The authors propose a novel AmBC-NOMA system that incorporates two key security enhancements:

  • Artificial Noise (AN) Injection: The base station (BS) injects AN into the transmitted signal. This AN is designed to mislead eavesdroppers without significantly impacting legitimate users. The effectiveness of AN cancellation for backscattered signals is quantified by an attenuation factor, $\eta$.
  • Eve-Exclusion Area: A circular "protected zone" around the BS is defined, within which no eavesdroppers are assumed to be present. Legitimate users are also positioned within this zone to further enhance security.

The paper then employs stochastic geometry, specifically modeling eavesdroppers as a homogeneous Poisson point process (HPPP), to analyze the system's performance. This approach allows for the mathematical modeling of random locations and numbers of eavesdroppers. The core of the analysis involves deriving theoretical expressions for:

  • Outage Probability (OP): This metric quantifies the probability that a legitimate user or the backscatter device (BD) fails to achieve its required signal-to-interference-plus-noise ratio (SINR).
  • Intercept Probability (IP): This metric measures the probability that an eavesdropper successfully decodes the transmitted information.

The authors derive exact expressions for OP and IP for the near user (UN), far user (UF), and the BD. They also derive asymptotic expressions for high signal-to-noise ratio (SNR) regimes to gain further insights into the system's behavior and to analyze diversity orders.

Datasets

Not discussed in the paper. The paper relies on theoretical analysis and Monte Carlo simulations for evaluation.

Key Insights and Ideas

  • Random Eavesdroppers are a Significant Challenge: The paper highlights that assuming fixed eavesdropper locations is unrealistic and that modeling them as randomly distributed (using HPPP) is crucial for a more accurate security assessment.
  • AN and Eve-Exclusion Zone are Effective: Both AN injection and the Eve-exclusion zone are shown to significantly improve the physical-layer security of AmBC-NOMA networks.
  • Reliability-Security Trade-off: The study confirms the inherent trade-off between reliability (measured by OP) and security (measured by IP). As transmit SNR increases, OP decreases, but IP generally increases, necessitating careful parameter tuning.
  • Impact of Parameters: The paper provides valuable insights into how various system parameters, such as transmit SNR, reflection efficiency ($\beta$), power allocation coefficients ($\alpha_N, \alpha_F$), power fraction ($\theta$), Eve-exclusion area radius ($r_p$), Eve distribution density ($\lambda_E$), and backscattered AN cancellation efficiency ($\eta$), influence both OP and IP.
  • Eve-Exclusion Zone Radius is Critical: Enlarging the Eve-exclusion area radius ($r_p$) is identified as a particularly effective strategy for enhancing PLS, as it directly reduces the probability of eavesdropping.
  • BD Performance is Linked to Users: The outage performance of the BD is shown to be jointly affected by the performance of both the near and far users, with the near user's performance having a greater impact.
  • Asymptotic Behavior Reveals Error Floors: At high SNRs, the OP converges to non-zero floor values, indicating zero diversity order in this regime. This suggests that beyond a certain SNR, further power increases yield diminishing returns in terms of reliability.

Key Results

  • Theoretical Expressions: Exact and asymptotic analytical expressions for OP and IP for all users (UN, UF, BD) are derived.
  • Validation: The theoretical results are validated through extensive Monte Carlo simulations, demonstrating their accuracy.
  • Performance Trends:
    • OP generally decreases with increasing transmit SNR, while IP increases.
  • Increasing reflection efficiency ($\beta$) increases IP and OP for UN/UF, but has a more complex effect on BD's OP.
  • Increasing power fraction ($\theta$) allocated to the desired signal decreases OP but increases IP.
  • Increasing Eve-exclusion area radius ($r_p$) and decreasing Eve distribution density ($\lambda_E$) significantly reduce IP.
  • The IP is not affected by the backscattered AN cancellation efficiency ($\eta$) as Eves cannot cancel AN, but OP is.
  • User Performance Order: For OP, UF generally performs best, followed by UN, and then BD. For IP, the order is reversed, with BD having the highest IP and UF the lowest.

Limitations and Future Work

  • Idealized Assumptions: The paper assumes ideal channel state information (CSI) for legitimate users and perfect AN cancellation for direct links. Real-world scenarios might involve imperfect CSI and residual AN.
  • Single-Cell Scenario: The analysis is conducted for a single-cell network. Future work could extend this to multi-cell environments.
  • Passive Eavesdroppers: The study focuses on passive eavesdroppers. The security implications of active eavesdroppers could be explored.
  • Simplified User Pairing: The paper considers a typical NOMA user pair. Investigating more complex user grouping strategies could be a future direction.
  • Parameter Optimization: While the paper analyzes the impact of various parameters, explicit optimization strategies for achieving the best trade-off between reliability and security could be a valuable extension.

Summary updated

BibTeX
@article{pei2024physical,
  author = {Xinyue Pei and Xingwei Wang and Min Huang and Yingyang Chen and Xiaofan Li and Theodoros A. Tsiftsis},
  title = {{Physical-Layer Security in AmBC-NOMA Networks With Random Eavesdroppers}},
  journal = {IEEE Internet of Things Journal},
  year = {2024},
}
Figure