Abstract
Traditional stochastic geometry models for wireless networks often rely on the Poisson Point Process (PPP), assuming an infinite spatial domain. However, real-world deployments like Unmanned Aerial Vehicle (UAV) swarms are inherently finite and bounded. This paper presents the first exact analytical framework for calculating coverage probability in finite 3D networks confined within a cylindrical volume. By employing a Binomial Point Process (BPP) and a novel geometric-probabilistic decomposition, the authors characterize inter-nodal distances and aggregate interference under Nakagami-$m$ fading. Validation against Monte Carlo simulations confirms the framework's precision, offering a robust alternative to simplified infinite-space approximations for designing reliable 3D robotic networks.
The Challenge of Finite 3D Networks
In wireless communication research, Stochastic Geometry is the primary tool for predicting network performance. For years, the "gold standard" has been the Poisson Point Process (PPP), which assumes an infinite number of nodes scattered across an infinite plane.
However, real-world deployments—like a swarm of Unmanned Aerial Vehicles (UAVs) monitoring a specific area—are finite and bounded. When a network is confined to a 3D shape like a cylinder, the standard mathematical assumptions break down. The spatial independence assumed in infinite models disappears, and the distance to a "serving" node becomes tightly coupled with the interference caused by its neighbors. For over a decade, finding a precise analytical way to characterize coverage in these finite 3D spaces remained an unresolved challenge due to the sheer complexity of the geometry.
A Geometric-Probabilistic Framework
The authors tackle this intractability using a Binomial Point Process (BPP) model, which assumes a fixed number of nodes ($N$) uniformly distributed within a finite volume. Their technical innovation is a four-step framework:
- Distance Distribution: They derive the exact Probability Density Function (PDF) for the distance between any two random points in a 3D cylinder. They achieve this by treating the horizontal (2D disk) and vertical (1D line) components as independent, calculating their individual squared distance distributions, and then using convolution to find the total 3D Euclidean distance.
- Order Statistics: They determine the distribution of the "serving link" (the distance to the nearest neighbor) by applying order statistics to the previously derived distance PDF.
- Laplace Transform Modeling: To handle aggregate interference from all other nodes, they use Laplace transforms. They specifically leverage the derivative properties of these transforms to incorporate Nakagami-$m$ fading, a versatile model for signal fluctuations.
- The Tower Law: Finally, they combine these elements using the tower law of expectation to produce a closed-form integral expression for coverage probability.
Key Results and Validation
The paper validates this mathematical framework using extensive Monte Carlo simulations across two distinct representative geometries: a "squat" cylinder (wide and short) and a "slender" cylinder (narrow and tall). These simulations serve as the ground truth to verify the analytical formulas.
Near-Perfect Accuracy
The analytical curves generated by the framework showed near-perfect agreement with Monte Carlo simulation results across all tested geometric configurations. This confirms that exact 3D analytical characterization is possible by decomposing the geometry into tractable components.
Superiority over PPP
The proposed BPP model significantly outperformed the conventional PPP approximation. The PPP model fails to capture the reduced interference at the boundaries—the "boundary effect"—where nodes near the edge experience less interference because no transmitters exist beyond the boundary. The authors' model, by contrast, aligns perfectly with the simulated ground truth.
Volume vs. Connectivity
A key insight from the results is that for a fixed number of drones, increasing the height of the cylinder (increasing volume) consistently degrades coverage. This occurs because the signal strength drops faster due to increased distance than the interference drops, highlighting a clear trade-off in 3D deployment design.
Future Work
While this framework provides a rigorous foundation for analysis, the authors acknowledge areas for expansion:
- Non-Uniformity: Exploring non-uniform distributions, such as drones clustering around specific targets.
- Channel Complexity: Incorporating altitude-dependent channels common in complex air-to-ground urban environments.
- Optimization: Using this mathematical foundation to optimize UAV placement, trajectory design, and swarm coordination in real-time.
Summary updated