Abstract
This paper investigates the stability region of a two-user broadcast channel (BC) characterized by bursty data arrivals and stringent security constraints. We focus on a scenario where one receiver ($D_1$) requires its data to remain secret from the other ($D_2$). To enhance its service rate under this constraint, $D_1$ is equipped with full-duplex (FD) capabilities, allowing it to transmit a jamming signal while receiving its own data. We characterize the stability region for various decoding schemes and analyze the impact of imperfect self-interference cancellation. Our findings reveal that the stability region is highly sensitive to the quality of self-interference cancellation and provide insights into the trade-offs between throughput and secrecy in random traffic environments.
Introduction and System Model
Traditional physical layer security research often assumes "infinitely backlogged" users with a constant stream of data. However, real-world network traffic is inherently "bursty," with data arriving at random intervals. If the arrival rate exceeds the secure processing capacity, queues grow indefinitely, leading to system instability.
This study addresses a Two-User Broadcast Channel (BC) where a single source ($S$) transmits distinct messages to two receivers, $D_1$ and $D_2$, as illustrated in Figure 1.
The primary challenge is ensuring that $D_1$’s data remains confidential from $D_2$. To achieve this, $D_1$ employs a full-duplex radio to transmit a jamming signal ($h_{12}$) intended to degrade $D_2$'s reception of $M_1$. This setup introduces a complex interplay between the desired signal reception ($h_1$), the jamming of the eavesdropper ($h_{12}$), and the resulting self-interference ($g$) at $D_1$.
Methodology
The analysis centers on defining the "stability region"—the set of arrival rates $(\lambda_1, \lambda_2)$ for which the system queues remain finite.
Stochastic Dominance
Because the service rate of one queue depends on whether the other queue is empty (a coupled system), the authors utilize the "dominant system" approach. By constructing a hypothetical system where queues continue to transmit "dummy" packets even when empty, the interaction is decoupled, enabling the mathematical derivation of the stability region's boundaries.
Full-Duplex Jamming and Decoding
The model incorporates a full-duplex receiver $D_1$ that acts as its own "friendly jammer." The residual self-interference (SI) after cancellation is modeled by a coefficient $g$. The paper evaluates three decoding scenarios:
- TIN/TIN: Both receivers treat the unintended signal as additive noise.
- SD/TIN: $D_1$ uses Successive Decoding (SD) to remove $D_2$’s interference, while $D_2$ treats $D_1$’s signal as noise.
- TIN/SD: $D_1$ treats $D_2$ as noise, but $D_2$ employs SD to attempt to strip away interference and potentially compromise secrecy.
Secrecy is quantified using Signal-to-Interference-plus-Noise Ratio (SINR) thresholds. A transmission is considered secure only if $D_1$ successfully decodes the message while $D_2$ fails to do so.
Key Insights and Results
The research yields several critical insights into the behavior of secure broadcast channels:
- Sensitivity to Self-Interference: The stability region is remarkably sensitive to the self-interference cancellation coefficient $g$. Even minor degradations in cancellation performance (e.g., moving from 60dB to 30dB isolation) can cause the sustainable traffic rates to shrink significantly.
- The Cost of Secrecy: Numerical simulations demonstrate a non-linear trade-off between the arrival rates. As the demand for secure data ($\lambda_1$) increases, the required jamming power rises, which in turn reduces the available throughput for the second user ($\lambda_2$) due to increased interference and potential self-interference.
- Saturated Throughput Optimization: The authors identified a threshold for the secure user's service rate ($\mu_{th}$). Below this threshold, $D_1$ can achieve its required rate without jamming. Above it, jamming power must scale rapidly to maintain secrecy, causing a sharp decline in $D_2$’s maximum throughput.
Conclusion and Future Directions
This work provides a foundational mathematical framework for analyzing secrecy in bursty broadcast networks. It highlights that full-duplex jamming is a powerful tool for security but is heavily dependent on the efficacy of hardware-level self-interference cancellation.
Future research could extend this model by incorporating instantaneous Channel State Information (CSI) for more dynamic power control, scaling the analysis to $N$-user networks, and adopting more realistic, non-linear models for hardware impairments in full-duplex systems.
Summary updated