Localization of a Passive Source with a Sensor Network based Experimental Molecular Communication Platform

TL;DR

This work introduces a macroscale molecular communication platform that uses passive evaporation of ethanol as the transmitter and a 24-node sensor network for TX localization. It leverages a Gaussian plume dispersion model as the system model and develops a Sensor Network-Based Clustered Localization Algorithm (SNCLA) that fuses wind-velocity estimates, transmitter mass, and sensor concentrations to localize the transmitter via nonlinear equations solved with two-node pairs. Experimental results show SNCLA achieves better localization under higher wind speeds and reveal non Gaussian channel characteristics, with sensed signals following a log-normal distribution and additive noise following a Student's t distribution. The framework provides a practical, scalable approach to passive-source localization in macroscale MC and offers insights into dispersion, sensor response, and non-Gaussian noise behavior, with potential extensions to larger-scale deployments.

Abstract

In a practical molecular communication scenario such as monitoring air pollutants released from an unknown source, it is essential to estimate the location of the molecular transmitter (TX). This paper presents a novel Sensor Network-based Localization Algorithm (SNCLA) for passive transmission by using a novel experimental platform which mainly comprises a clustered sensor network (SN) with $24$ sensor nodes and evaporating ethanol molecules as the passive TX. In SNCLA, a Gaussian plume model is employed to derive the location estimator. The parameters such as transmitted mass, wind velocity, detection time, and actual concentration are calculated or estimated from the measured signals via the SN to be employed as the input for the location estimator. The numerical results show that the performance of SNCLA is better for stronger winds in the medium. Our findings show that evaporated molecules do not propagate homogeneously through the SN due to the presence of the wind. In addition, our statistical analysis based on the measured experimental data shows that the sensed signals by the SN have a log-normal distribution, while the additive noise follows a Student's t-distribution in contrast to the Gaussian assumption in the literature.

Figures (12)

• Figure 1: Experimental platform.
• Figure 2: The deployment of the sensor nodes and TX.
• Figure 3: Block diagram of the SNCLA.
• Figure 4: Measurement circuit of the sensor board.
• Figure 5: Estimated points using SNCLA for all clusters.