Modeling Cellular Networks in Fading Environments with Dominant Specular Components

Ahmad AlAmmouri, Hesham ElSawy, Ahmed Kamal Sultan-Salem, Marco Di Renzo, Mohamed-Slim Alouini

Abstract

Stochastic geometry (SG) has been widely accepted as a fundamental tool for modeling and analyzing cellular networks. However, the fading models used with SG analysis are mainly confined to the simplistic Rayleigh fading, which is extended to the Nakagami-m fading in some special cases. However, neither the Rayleigh nor the Nakagami-m accounts for dominant specular components (DSCs) which may appear in realistic fading channels. In this paper, we present a tractable model for cellular networks with generalized two-ray (GTR) fading channel. The GTR fading explicitly accounts for two DSCs in addition to the diffuse components and offers high flexibility to capture diverse fading channels that appear in realistic outdoor/indoor wireless communication scenarios. It also encompasses the famous Rayleigh and Rician fading as special cases. To this end, the prominent effect of DSCs is highlighted in terms of average spectral efficiency.

Introduction

For over a decade, Stochastic Geometry (SG) has served as the backbone for modeling cellular networks, typically treating base station locations as a Poisson Point Process (PPP). While this approach effectively captures spatial randomness, the accompanying fading models—which describe signal fluctuations—have often remained overly simplistic. Most studies rely on Rayleigh fading, assuming no direct line-of-sight (LOS) between the transmitter and receiver.

In modern 5G and millimeter-wave (mmW) networks, this assumption is increasingly inaccurate. Realistic environments often feature multiple Dominant Specular Components (DSCs) arising from LOS paths or strong reflections from metallic objects like vehicles and light posts. This paper bridges the gap by integrating the Generalized Two-Ray (GTR) fading model into a tractable SG framework, providing a more accurate tool for calculating network spectral efficiency.

The Generalized Two-Ray (GTR) Model

Unlike traditional models, the GTR model accounts for two DSCs plus a diffuse (scattered) component. It is characterized by four primary parameters:

  • $K$: The specular-to-diffuse power ratio.
  • $\Delta$: The peak-to-average specular power ratio.
  • $2\sigma^2$: The diffuse power.
  • $f_\alpha$: The distribution of the phase difference between the two specular components.

By adjusting these parameters, the GTR model can simplify into Rayleigh (zero specular components) or Rician (one specular component) fading, making it a versatile "one-size-fits-all" framework for diverse wireless environments.

Analytical Framework and Tractability

The core challenge in applying SG to complex fading models is finding the Laplace Transform (LT) of the aggregate interference. The authors derived an exact LT for GTR fading and proposed a computationally efficient lower bound. This allows for the analysis of different phase correlation scenarios, such as GTR-U (uniformly distributed phase), GTR-T (truncated uniform distribution), and GTR-V (Von Mises distribution).

Impact of Specular Components on Spectral Efficiency

A key insight of the study is that the fading characteristics of interfering signals are just as critical as those of the intended signal. While a strong LOS component (high $K$) in the useful link improves performance, the same characteristic in interfering links can be detrimental.

Average spectral efficiency vs. specular-to-diffuse ratio K for various fading models.

As shown in Figure 1, Rayleigh fading on interfering links actually results in the highest spectral efficiency because the interference is more "spread out." Conversely, as the specular-to-diffuse ratio ($K$) increases for Rician or GTR models, the spectral efficiency drops because the interferers become more dominant and consistent.

The Role of Phase Correlation

The research also highlights how the correlation between the phases of the two specular components affects performance. In the GTR-T model, the parameter $p$ represents the degree of phase correlation.

Impact of phase correlation parameter p on the average spectral efficiency.

Figure 2 demonstrates that high phase correlation (low $p$) can lead to significant destructive interference, causing a sharp drop in average spectral efficiency. As the phases become less correlated (increasing $p$), the spectral efficiency improves, eventually approaching the levels of more traditional models.

Conclusion

This work demonstrates that assuming Rayleigh fading in network modeling can lead to significant inaccuracies in environments with strong reflections or LOS paths. By providing a unified, tractable framework using GTR fading, the authors enable more realistic performance evaluations for next-generation cellular networks. Future work may extend this to distance-dependent fading parameters and multi-tier heterogeneous network architectures.

Summary updated

Abstract

Stochastic geometry (SG) has been widely accepted as a fundamental tool for modeling and analyzing cellular networks. However, the fading models used with SG analysis are mainly confined to the simplistic Rayleigh fading, which is extended to the Nakagami-m fading in some special cases. However, neither the Rayleigh nor the Nakagami-m accounts for dominant specular components (DSCs) which may appear in realistic fading channels. In this paper, we present a tractable model for cellular networks with generalized two-ray (GTR) fading channel. The GTR fading explicitly accounts for two DSCs in addition to the diffuse components and offers high flexibility to capture diverse fading channels that appear in realistic outdoor/indoor wireless communication scenarios. It also encompasses the famous Rayleigh and Rician fading as special cases. To this end, the prominent effect of DSCs is highlighted in terms of average spectral efficiency.
BibTeX
@inproceedings{alammouri2016modeling,
  author = {Ahmad AlAmmouri and Hesham ElSawy and Ahmed Kamal Sultan-Salem and Marco Di Renzo and Mohamed-Slim Alouini},
  title = {{Modeling Cellular Networks in Fading Environments with Dominant Specular Components}},
  booktitle = {ICC},
  year = {2016},
  url = {https://openreview.net/forum?id=_aNZ6RfyPQc},
}
Figure