Detection of Gaussian signals via hexagonal sensor networks

TL;DR

This paper presents a method for sensor networks arranged in a hexagonal pattern to estimate Gaussian signals, such as fire spread or pollution, using local measurements and consensus algorithms to improve accuracy.

Contribution

It introduces a novel approach for Gaussian signal estimation in hexagonal sensor networks, including local parameter estimation and error compensation through consensus.

Findings

Each sensor can estimate Gaussian parameters locally.

The method is robust to measurement errors.

A consensus algorithm effectively fuses local estimates into a global one.

Abstract

This paper considers a special case of the problem of identifying a static scalar signal, depending on the location, using a planar network of sensors in a distributed fashion. Motivated by the application to monitoring wild-fires spreading and pollutants dispersion, we assume the signal to be Gaussian in space. Using a network of sensors positioned to form a regular hexagonal tessellation, we prove that each node can estimate the parameters of the Gaussian from local measurements. Moreover, we study the sensitivity of these estimates to additive errors affecting the measurements. Finally, we show how a consensus algorithm can be designed to fuse the local estimates into a shared global estimate, effectively compensating the measurement errors.