IoT Wallet: Machine Learning-based Sensor Portfolio Application

TL;DR

This paper introduces an IoT sensor wallet application integrating machine learning to estimate soil moisture from LoRa beacon signals, with features like user-specific data views, control, notifications, and a case study demonstrating its effectiveness.

Contribution

It presents a novel IoT sensor wallet system that incorporates machine learning for soil moisture estimation using LoRa signals, enhancing sensor data analysis and user interaction.

Findings

Successful estimation of soil moisture from LoRa signal strength

System supports customizable user roles and notifications

Case study validates the machine learning approach

Abstract

In this paper an application for building sensor wallet is presented. Currently, given system collects sensor data from The Things Network (TTN) cloud system, stores the data into the Influx database and presents the processed data to the user dashboard. Based on the type of the user, data can be viewed-only, controlled or the top user can register the sensor to the system. Moreover, the system can notify users based on the rules that can be adjusted through the user interface. The special feature of the system is the machine learning service that can be used in various scenarios and is presented throughout the case study that gives a novel approach to estimate soil moisture from the signal strength of a given underground LoRa beacon node.