Performance Analysis of Finite Blocklength Transmissions Over Wiretap Fading Channels: An Average Information Leakage Perspective

Milad Tatar Mamaghani, Xiangyun Zhou, Nan Yang, A. Lee Swindlehurst, H. Vincent Poor

In the evolving landscape of 5G and 6G, communication is shifting from high-speed streaming toward "short-packet" bursts. These bursts, common in IoT sensors, autonomous vehicles, and industrial controllers, operate in the Finite Blocklength (FBL) regime. Unlike traditional theory, which assumes infinite blocklengths to guarantee security, FBL transmissions face unique vulnerabilities where coding gains drop and the trade-off between reliability and secrecy becomes much tighter.

A recent study published in IEEE Transactions on Communications by Milad Tatar Mamaghani and colleagues provides a rigorous framework for analyzing and optimizing security in these low-latency scenarios.

The Wiretap Channel Model

The research focuses on a classic physical-layer security scenario known as the wiretap channel. In this model, a transmitter (Alice) attempts to send a secret message to a legitimate receiver (Bob) while a passive eavesdropper (Eve) attempts to intercept the communication.

The classic wiretap channel model featuring Alice (transmitter), Bob (legitimate receiver), and Eve (eavesdropper).

In real-world deployments, Eve is often passive and her exact location or channel state information (CSI) is unknown. To counter this, the authors analyze Artificial Noise (AN) beamforming, where Alice uses multiple antennas to beam the secret message toward Bob while simultaneously "spraying" noise into the null-space of Bob's channel to confuse Eve.

A New Metric: Average Information Leakage (AIL)

Traditional security metrics like "secrecy capacity" or binary "outage" probabilities are often insufficient for short packets. The authors introduce Average Information Leakage (AIL) as a more practical metric. AIL measures the statistical dependence between the original message and Eve’s observation, quantifying how much information Eve can actually exploit.

To make this metric computationally feasible, the researchers employed Laplace’s Approximation Theorem. This mathematical approach allowed them to derive closed-form expressions for AIL, turning complex integrals into manageable formulas that can be used for real-time system optimization.

Validation of Analytical Framework

A critical part of the study was verifying the accuracy of these new analytical tools. The researchers used Monte-Carlo simulations with 100,000 independent channel realizations to test their formulas against simulated data.

Comparison between simulation results and analytical approximations for the CDF of the system metrics across different SNR levels (ρ).

As shown in the validation plots, the analytical approximations (red squares) match the simulation results (blue lines) almost perfectly across a wide range of Signal-to-Noise Ratios (SNR). This confirms that the derived formulas are highly reliable for predicting system performance in various fading environments.

Key Insights and Design Implications

  • The Power of Noise: Unlike standard beamforming, which reaches a "leakage floor" at high power, Artificial Noise can drive AIL to near-zero levels as transmit power increases, making it essential for short-packet security.
  • The Reliability Dividend: The study found that by allowing a miniscule, controlled amount of information leakage, systems can achieve massive gains in transmission reliability.
  • Optimal Design: For non-adaptive systems, the optimal blocklength was found to grow linearly with the number of information bits. However, by adaptively changing blocklength and power based on Bob's channel, the "Average Secrecy Throughput" can be significantly boosted.

Future Directions

While this framework provides a robust foundation, the authors note that future work should address scenarios where Alice has imperfect knowledge of Bob's channel or where Eve possesses a large antenna array capable of nullifying artificial noise. They also suggest that Deep Reinforcement Learning (DRL) could be a powerful tool for managing these complex optimizations in dynamic, real-time environments.

Summary updated

BibTeX
@article{mamaghani2024performance,
  author = {Milad Tatar Mamaghani and Xiangyun Zhou and Nan Yang and A. Lee Swindlehurst and H. Vincent Poor},
  title = {{Performance Analysis of Finite Blocklength Transmissions Over Wiretap Fading Channels: An Average Information Leakage Perspective}},
  journal = {IEEE Trans. Wireless Communications},
  year = {2024},
  doi = {10.1109/twc.2024.3400601},
}
Figure