Unified Stochastic Geometry Modeling and Analysis of Cellular Networks in LOS/NLOS and Shadowed Fading

Ben Liang, Imene Trigui, Sofiène Affes, Ben Liang 0001

TL;DR

This paper introduces a unified mathematical framework using stochastic geometry to analyze cellular networks under realistic, complex fading conditions. By shifting the analytical focus from the standard Cumulative Distribution Function (CDF) to the Moment Generating Function (MGF) of the Signal-to-Interference-plus-Noise Ratio (SINR), the authors provide a "Swiss Army knife" toolkit for calculating coverage, data rates, and error probabilities across diverse channel models like 5G and D2D.

What Problem Does This Paper Address?

For decades, researchers have relied on Rayleigh fading to model wireless channels because it is mathematically "tractable"—it makes the equations easy to solve. However, Rayleigh fading is a poor fit for the next generation of networks (5G, femtocells, and Device-to-Device communication). These modern scenarios involve complex Line-of-Sight (LOS) components, non-Line-of-Sight (NLOS) paths, and heavy shadowing that Rayleigh simply cannot capture.

While more sophisticated models like $\kappa$-$\mu$ and $\eta$-$\mu$ exist, they are notoriously difficult to use in Stochastic Geometry—the mathematical field used to model the random locations of base stations. Previous attempts to bridge this gap often resulted in infinite series or required computationally expensive simulations, offering little in the way of intuitive design guidelines for engineers.

Methods

The authors tackle this by moving away from the ubiquitous CDF-based analysis. Instead, they exploit the Moment Generating Function (MGF) of the SINR.

Key technical highlights include:

  • Unified Modeling: They use the shadowed $\kappa$-$\mu$ distribution as a "master" model. By tuning its parameters, it can simplify into almost any common fading model, including Rayleigh, Rician, Nakagami-$m$, and Hoyt.
  • Stochastic Geometry: They model base station locations as a homogeneous Poisson Point Process (PPP).
  • Mathematical Heavy Lifting: The paper derives exact expressions for the MGF using advanced functions like the Humbert function and Appell hypergeometric functions.
  • Performance Integration: For the first time, they integrate three critical metrics—coverage probability, ergodic achievable rate, and bit error probability (BEP)—into a single analytical framework.

Datasets

As a theoretical contribution, this paper does not use external training datasets. Instead, the "data" consists of:

  • Monte Carlo Simulations: Extensive simulations were used to validate the accuracy of the derived closed-form mathematical expressions.
  • Parameter Sweeps: The model was tested across a broad range of physical parameters, including base station density ($\lambda$), path-loss exponents ($\alpha$), and various fading/shadowing intensities ($m, \kappa, \mu, \eta$).

Key Insights and Ideas

  • The MGF Advantage: The authors demonstrate that the MGF is a more powerful starting point than the CDF for complex fading. It allows for a cleaner dissociation between the "tagged" user's link and the aggregate interference from other base stations.
  • Asymmetric Fading: The framework allows the direct link (BS to user) and the interference links to experience different types of fading. This is crucial for urban environments where the serving BS might have a clear LOS, but interferers are blocked by buildings.
  • Density Invariance: The paper confirms that in "interference-limited" regimes (where noise is negligible), the average data rate is independent of base station density. This suggests that simply adding more base stations won't improve spectral efficiency unless interference management techniques are also used.
  • Shadowing Nuance: Interestingly, the results show that heavy shadowing on dominant signal components can sometimes improve SINR by reducing the "fluctuation" of interference, though it generally hurts the desired signal strength.

Key Results

  • Perfect Match: The analytical formulas show an "excellent agreement" with Monte Carlo simulations across all tested scenarios.
  • Unified Formulas: The paper provides a single set of equations that can calculate performance for Rayleigh, Rice, and Nakagami-$m$ channels simultaneously by simply changing input variables.
  • High-SIR Approximations: For engineers looking for quick insights, the authors derived simplified closed-form expressions for the high Signal-to-Interference Ratio (SIR) regime, showing exactly how parameters like the number of clusters ($\mu$) scale with error rates.

Limitations and Future Work

  • Mathematical Complexity: While the formulas are "closed-form," they involve specialized hypergeometric functions. While these are standard in software like MATLAB or Mathematica, they are more complex to implement than simple exponential functions.
  • Single-Tier Focus: The current model focuses on a single-tier network. The authors note that extending this to multi-tier heterogeneous networks (HetNets)—where macrocells and small cells overlap—is a logical next step.
  • MIMO Systems: The analysis assumes single-antenna systems. Future work could extend this framework to Multiple-Input Multiple-Output (MIMO) configurations, which are standard in 5G.

Summary updated

Abstract

Statistical characterization of the signal-tointerference-plus-noise ratio (SINR) via its cumulative distribution function is ubiquitous in a vast majority of technical contributions in the area of cellular networks, since it boils down to averaging the Laplace transform of the aggregate interference, a benefit accorded at the expense of confinement to the simplistic Rayleigh fading. In this paper, to capture diverse fading channels that arise in realistic outdoor/indoor wireless communication scenarios, we tackle the problem differently. By exploiting the moment generating function of the SINR, we succeed in analytically assessing cellular networks performance over the shadowed κ-μ, κ-μ, and η-μ fading models. These channel models offer high flexibility by capturing diverse fading channels, including Rayleigh, Nakagami-m, Rician, and Rician shadow fading distributions. These channel models have been recently promoted for their capability to accurately model dense urban environments, future femtocells, and device-to-device shadowed channels. In addition to unifying the analysis for different channel models, this paper integrates the coverage, the achievable rate, and the bit error probability, which are largely treated separately in the literature. The developed model and analysis are validated over a broad range of simulation setups and parameters.
BibTeX
@article{liang2017unified,
  author = {Ben Liang and Imene Trigui and Sofiène Affes and Ben Liang},
  title = {{Unified Stochastic Geometry Modeling and Analysis of Cellular Networks in LOS/NLOS and Shadowed Fading}},
  journal = {IEEE Trans. Communications},
  year = {2017},
  doi = {10.1109/tcomm.2017.2732444},
}

Loading related content...

Figure