Performance Analysis of Clustered LoRa Networks

TL;DR

This paper analyzes the uplink performance of LoRa networks with clustered node distributions, considering noise, and identifies optimal node densities for maximizing spectral efficiency.

Contribution

It introduces a new topology modeling LoRa nodes with a Poisson cluster process and derives exact and approximate coverage probability expressions considering noise.

Findings

Optimal active LoRa node number maximizes spectral efficiency

Performance improves with increased node density around receivers

Noise consideration significantly affects coverage probability

Abstract

In this paper, we investigate the uplink transmission performance of low-power wide-area (LPWA) networks with regards to coexisting radio modules. We adopt long range (LoRa) radio technique as an example of the network of focus even though our analysis can be easily extended to other situations. We exploit a new topology to model the network, where the node locations of LoRa follow a Poisson cluster process (PCP) while other coexisting radio modules follow a Poisson point process (PPP). Unlike most of the performance analysis based on stochastic geometry, we take noise into consideration. More specifically, two models, with a fixed and a random number of active LoRa nodes in each cluster, respectively, are considered. To obtain insights, both the exact and simple approximated expressions for coverage probability are derived. Based on them, area spectral efficiency and energy efficiency are obtained. From our analysis, we show how the performance of LPWA networks can be enhanced through adjusting the density of LoRa nodes around each LoRa receiver. Moreover, the simulation results unveil that the optimal number of active LoRa nodes in each cluster exists to maximize the area spectral efficiency.