On the capacity of the underwater acoustic channel with dominant noise sources

Mustafa A. Kishk, Ahmed M. Alaa

TL;DR

This paper derives an upper bound for the channel capacity of underwater acoustic (UWA) communications operating in shallow waters, where impulsive noise and generalized fading dominate. By modeling the noise as a Generalized Gaussian (GG) distribution, the authors prove that impulsive noise actually increases theoretical channel capacity compared to standard Gaussian noise. Furthermore, they show that this non-Gaussian noise behavior can be exploited to achieve secure communications, even if an eavesdropper has a better signal-to-noise ratio (SNR).

What Problem Does This Paper Address?

In standard wireless communications, engineers usually assume background noise is perfectly Gaussian (AWGN). However, in shallow underwater acoustic channels, this assumption falls apart. The underwater environment is filled with dominant, impulsive noise sources (like snapping shrimp, breaking waves, or machinery).

Previous experimental studies have shown that this underwater noise is better modeled by a Generalized Gaussian (GG) distribution, while the signal fading is best captured by an $\alpha-\mu$ distribution. The problem is that the fundamental performance limits—specifically, the maximum data rate (channel capacity) and secure data rate (secrecy capacity)—for this specific combination of GG noise and $\alpha-\mu$ fading were previously unknown. This paper attempts to mathematically bound these limits.

Methods

The authors take an information-theoretic approach to bound the capacity of the Additive White Generalized Gaussian Noise (AWGGN) channel:

  • Differential Entropy: They calculate the differential entropy of the GG noise distribution, which is parameterized by a shape parameter $\beta$ (where $\beta=2$ is standard Gaussian, and lower values are more impulsive).
  • Capacity Bounding: Using the established information theory principle that Gaussian noise is the "worst-case" noise (it maximizes entropy for a given variance), they prove that the capacity of the AWGGN channel is bounded by the standard AWGN capacity plus a constant gap that depends entirely on $\beta$.
  • Sphere-Packing Interpretation: They provide a geometric proof using the "sphere-packing" concept, comparing the volume of GG noise spheres to AWGN spheres to explain the capacity increase.
  • Fading Integration: They extend the AWGGN bounds to fading environments by averaging the conditional capacity over the $\alpha-\mu$ fading distribution, yielding an analytical solution involving Meijer-G functions.
  • Secrecy Capacity: They apply these capacity bounds to calculate point-to-point secrecy rates against eavesdroppers experiencing different noise conditions.

Datasets

Not discussed in the paper. This is a purely theoretical and analytical study, though it builds upon noise and fading statistical models that were validated by prior experimental literature (such as acoustic data collected off the coast of Rio de Janeiro, mentioned in the introduction).

Key Insights and Ideas

  • Impulsive noise is actually good for capacity: Counterintuitively, because standard Gaussian noise represents the worst-case scenario (highest uncertainty), the more "impulsive" or deterministic the noise becomes ($\beta < 2$), the higher the theoretical channel capacity.
  • The Capacity Gap: The capacity of an underwater AWGGN channel is simply the standard AWGN capacity plus a constant, positive gap $f(\beta)$. As $\beta$ decreases (noise becomes more impulsive), this gap grows, offering significant capacity improvements over standard Gaussian channels.
  • Geometric Compression: When noise spheres are subjected to GG noise rather than standard Gaussian noise, they effectively get compressed. This smaller noise volume means a receiver can pack more distinguishable signal spheres into the same space without them overlapping, directly translating to higher data rates.
  • Secrecy through Noise Shape: In a secure communication scenario, if the legitimate receiver and the eavesdropper are at different depths, they might experience different noise shapes ($\beta$). A legitimate receiver can achieve a positive secrecy rate even if the eavesdropper has a stronger signal (higher SNR), provided the eavesdropper's noise is more Gaussian.

Key Results

  • Closed-form Capacity Bound: Derived an upper bound for the AWGGN channel capacity: $C_{AWGGN} \leq C_{AWGN} + f(\beta)$, where $f(\beta)$ is a specific logarithmic function of the shape parameter $\beta$.
  • Ergodic Fading Capacity: Derived the ergodic capacity bound for the UWA channel under $\alpha-\mu$ fading, providing a generalized formula that covers a wide range of underwater settings (depth, frequency, etc.).
  • Secrecy Rate Advantage: Demonstrated numerically that if a legitimate receiver experiences noise with $\beta = 1.5$ and an eavesdropper experiences $\beta = 0.8$, secure communication is possible even when the legitimate receiver's SNR is worse than the eavesdropper's SNR (e.g., the eavesdropper is at $-5$ dB, and the legitimate receiver is below $-5$ dB).

Limitations and Future Work

The paper is purely theoretical and focuses on deriving mathematical upper bounds rather than exact capacity expressions or practical coding schemes that achieve these bounds. While the authors do not explicitly outline a "Future Work" section, natural next steps based on this research would involve designing specific modulation and forward error correction schemes tailored to exploit the $f(\beta)$ capacity gap in practical underwater acoustic modems, as well as validating the secrecy capacity findings in physical underwater testbeds.

Summary updated

Abstract

This paper provides an upper-bound for the capacity of the underwater acoustic (UWA) channel with dominant noise sources and generalized fading environments. Previous works have shown that UWA channel noise statistics are not necessary Gaussian, especially in a shallow water environment which is dominated by impulsive noise sources. In this case, noise is best represented by the Generalized Gaussian (GG) noise model with a shaping parameter β. On the other hand, fading in the UWA channel is generally represented using an α-μ distribution, which is a generalization of a wide range of well-known fading distributions. We show that the Additive White Generalized Gaussian Noise (AWGGN) channel capacity is upper bounded by the AWGN capacity in addition to a constant gap of 1/2 log(β <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> πe <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1-2/β</sup> Γ(3/β)/2(Γ(1/β)) <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sup> bits. The same gap also exists when characterizing the ergodic capacity of AWGGN channels with α-μ fading compared to the faded AWGN channel capacity. We justify our results by revisiting the sphere-packing problem, which represents a geometric interpretation of the channel capacity. Moreover, UWA channel secrecy rates are characterized and the dependency of UWA channel secrecy on the shaping parameters of the legitimate and eavesdropper channels is highlighted.
BibTeX
@inproceedings{kishk2014capacity,
  author = {Mustafa A. Kishk and Ahmed M. Alaa},
  title = {{On the capacity of the underwater acoustic channel with dominant noise sources}},
  booktitle = {ISTT},
  year = {2014},
  url = {https://openreview.net/forum?id=0Pder7aX1YJ},
}

Loading related content...

Figure