Secure Communications for the Two-User Broadcast Channel With Random Traffic

Parthajit Mohapatra, Nikolaos Pappas, Jemin Lee, Tony Q. S. Quek, Vangelis Angelakis

This paper, titled "Secure Communications for the Two-user Broadcast Channel with Random Traffic," bridges the gap between information-theoretic secrecy and the practical reality of bursty network traffic. It provides a mathematical framework for understanding how much data a network can handle when one user’s privacy must be protected from another.

TL;DR

The authors characterize the "stability region"—the set of sustainable data arrival rates—for a two-user broadcast channel where one receiver requires secrecy. By equipping the secure receiver with full-duplex jamming capabilities, they demonstrate how to maintain privacy in bursty traffic environments, though they highlight a critical sensitivity to self-interference.

What Problem Does This Paper Address?

Most classical research in physical layer security assumes "infinitely backlogged" users—nodes that always have a mountain of data ready to send. In reality, network traffic is "bursty" (random arrivals). If data arrives faster than it can be securely processed and transmitted, queues grow infinitely, and the system becomes unstable.

The challenge here is a Two-User Broadcast Channel (BC): a single transmitter sends different data to Receiver 1 ($D_1$) and Receiver 2 ($D_2$). The catch? $D_1$’s data must remain a secret from $D_2$. To achieve this, $D_1$ uses a full-duplex (FD) radio to transmit a "jamming" signal to confuse $D_2$ while simultaneously trying to listen to its own data. This creates a complex three-way tension between throughput, secrecy, and the self-interference caused by $D_1$ jamming itself.

Methods

The authors employ several key theoretical and analytical techniques:

  • Stochastic Dominance: Because the two queues at the transmitter interact (the service rate of one depends on whether the other is empty), the authors use the "dominant system" approach. They construct hypothetical systems where queues send "dummy" packets to decouple this interaction, allowing them to define the boundaries of the stability region.
  • Full-Duplex Jamming Model: Unlike traditional "friendly jammers" (a third party), the receiver itself acts as the jammer. They model the residual self-interference (SI) using a coefficient $g$, where $g=0$ represents perfect cancellation.
  • Decoding Variations: The paper derives stability regions for three distinct scenarios:
  • TIN/TIN: Both receivers treat the other's signal as noise.
  • SD/TIN: $D_1$ uses Successive Decoding (SD) to cancel $D_2$’s interference, while $D_2$ treats $D_1$ as noise.
  • TIN/SD: $D_1$ treats $D_2$ as noise, but $D_2$ is "smarter" and tries to use SD to strip away interference.
  • SINR-based Secrecy: Secrecy is defined by the Signal-to-Interference-plus-Noise Ratio (SINR). For a transmission to be "secure," $D_1$ must decode it successfully (SINR above a threshold), while $D_2$ must fail to decode it (SINR below a threshold).

Datasets

Not discussed in the paper. As a theoretical work in communication theory, the authors use numerical simulations based on mathematical models (e.g., Rayleigh fading channels) rather than empirical datasets.

Key Insights and Ideas

  • The Full-Duplex Double-Edged Sword: Jamming is essential for secrecy when the channel to the "eavesdropping" receiver ($D_2$) is strong. However, if self-interference cancellation isn't nearly perfect, the jammer ($D_1$) effectively "blinds" itself, causing its own queue to become unstable.
  • Secrecy Sensitivity: The stability region is remarkably sensitive to the self-interference cancellation coefficient. A small drop in cancellation performance (e.g., moving from 60dB to 30dB isolation) can lead to a massive "shrinkage" of the sustainable traffic rates.
  • Successive Decoding Dynamics: If the unintended receiver ($D_2$) has the hardware to perform successive decoding, the "secure" user ($D_1$) is forced to jam much more aggressively, which further taxes the system's overall stability.

Key Results

  • Stability Region Characterization: The authors provide the first closed-form mathematical expressions for the stability regions of a secure BC with bursty traffic.
  • Throughput Trade-offs: Numerical results show that as the arrival rate for the secure user ($\lambda_1$) increases, the penalty on the second user’s throughput increases non-linearly due to the required jamming power.
  • Saturated Throughput Optimization: The authors solved an optimization problem to maximize $D_2$’s throughput while guaranteeing a minimum service rate for $D_1$. They found that below a certain service threshold (e.g., $\mu_{th} \approx 0.125$ in their simulations), $D_1$ doesn't need to jam at all. Once that threshold is crossed, jamming power must scale up rapidly, causing $D_2$’s throughput to plummet.

Limitations and Future Work

  • Channel State Information (CSI): The model assumes the transmitter only knows the statistics of the channels, not the instantaneous state. Real-time CSI could potentially allow for more sophisticated beamforming or power control.
  • Two-User Limit: The analysis is restricted to two users. Scaling this to a broadcast channel with $N$ users and multiple secrecy constraints would significantly increase the complexity of the queue interactions.
  • Hardware Assumptions: The full-duplex model is idealized. Future work could incorporate more realistic, non-linear models of self-interference and hardware impairments.

Summary updated

Abstract

In this paper, we study the stability region of the two-user broadcast channel (BC) with bursty data arrivals and security constraints. It is assumed that one of the receivers has a secrecy constraint, i.e., its packets need to be kept secret from the other receiver, which is defined based on signal to interference noise ratio. The receiver with secrecy constraint has full-duplex capability to send a jamming signal for improving its service rate. The stability region of the two-user BC with secrecy constraint is characterized for the general decoding case. Then, assuming two different decoding schemes, the respective stability regions are derived. The full-duplex operation of receiver results in self-interference, and the effect of imperfect self-interference cancelation on the stability region is also investigated. The stability region of the BC with a secrecy constraint, where the receivers do not have full duplex capability can be obtained as a special case of the results derived in this paper. In addition, the paper considers the problem of maximizing the saturated throughput of the queue for which there is no secrecy constraint under minimum service guarantees for the other queue. The results provide new insights on the effect of the secrecy constraint on the stability region of the BC. It is found that the stability region with secrecy constraint is sensitive to the degree of self-interference cancelation.
BibTeX
@article{mohapatra2018secure,
  author = {Parthajit Mohapatra and Nikolaos Pappas and Jemin Lee and Tony Q. S. Quek and Vangelis Angelakis},
  title = {{Secure Communications for the Two-User Broadcast Channel With Random Traffic}},
  journal = {IEEE Trans. Inf. Forensics Secur.},
  year = {2018},
  doi = {10.1109/tifs.2018.2818076},
}
Figure