Leveraging IoT and Machine Learning for Improved Monitoring of Water Resources – A Case Study at Upper Ewaso Nyiro River (KENYA).

Abstract

This report outlines the development of an IoT sensor system (Project Ewaso), capable of water-level monitoring in rivers channels, with an aim of ensuring equitable distribution of water and sourcing of data to quantify unsustainable water usage and catchment area destruction. The paper also defines an application scenario in a specific hydrological region of the Ewaso Nyiro basin in Kenya, highlighting the characteristics of data collection and processing used. Fixed position node systems are described along with web-based data acquisition platform developments integrated with IoT techniques to retrieve data. The developed architecture utilizes the LoRaWAN - LoRa protocol to send real-time data packets from nodes deployed to a server that displays, decodes and stores the data. From the server, data can be transferred to a time series database, where it can be accessed and displayed through different customizable queries and graphical representation allowing future use in prediction machine learning systems. All these characteristics are presented along with evidence of the deployment of different devices and of the IoT network infrastructure.