Joint Spatial-Propagation Modeling of Cellular Networks Based on the Directional Radii of Poisson Voronoi Cells

TL;DR

This paper introduces a joint spatial-propagation model for cellular networks that accounts for the correlation between cell sizes and signal propagation conditions, using the novel concept of directional radii in Poisson Voronoi cells.

Contribution

It proposes the notion of directional radius of Voronoi cells and analyzes its distribution, linking cell shape variability to large-scale signal propagation in cellular network modeling.

Findings

JSP model with PPP closely matches coverage of lattice models.

Network performance is sensitive to variance in large-scale path loss.

Directional radii provide new insights into cell boundary behavior.

Abstract

In coverage-oriented networks, base stations (BSs) are deployed in a way such that users at the cell boundaries achieve sufficient signal strength. The shape and size of cells vary from BS to BS, since the large-scale signal propagation conditions differ in different geographical regions. This work proposes and studies a joint spatial-propagation (JSP) model, which considers the correlation between cell radii and the large-scale signal propagation (captured by shadowing). We first introduce the notion of the directional radius of Voronoi cells, which has applications in cellular networks and beyond. The directional radius of a cell is defined as the distance from the nucleus to the cell boundary at an angle relative to the direction of a uniformly random location in the cell. We study the distribution of the radii in two types of cells in the Poisson Voronoi tessellations: the zero-cell, which contains the origin, and the typical cell. The results are applied to analyze the JSP model. We show that, even though the Poisson point process (PPP) is often considered as a pessimistic spatial model for BS locations, the JSP model with the PPP achieves coverage performance close to the most optimistic one -- the standard triangular lattice model. Further, we show that the network performance depends critically on the variance of the large-scale path loss along the cell boundary.