Water quality monitoring of Sungai Pusu using IoT technology and machine learning approaches

Abstract

Water quality monitoring is crucial for maintaining healthy ecosystems and ensuring the availability of clean water resources. Traditional monitoring methods, relying on laboratory tests, are often insufficient due to their time-consuming nature and lack of real-time data. This research aims to develop a cost-effective, real-time water quality monitoring system using Internet of Things (IoT) technology and machine learning (ML) algorithms. The primary objectives of this study are to characterize water quality in the Sungai Pusu at IIUM Gombak campus, investigate the effectiveness of IoT-enabled sensors and ML algorithms in continuous water quality monitoring, and design an integrated water quality monitoring system. The methodology involved developing an IoT-based device using Arduino UNO microcontroller and multiple sensors to collect water quality data. Two datasets were generated: the first with four parameters (pH, turbidity, temperature, and total dissolved solids) and the second including an additional dissolved oxygen sensor. Data was collected from three distinct water sources: potable water, flowing river water, and stagnant puddle water. Nine different classification algorithms were applied to analyze the collected data. Results showed that the developed system successfully classified water quality conditions with up to 98% accuracy. For the first dataset, Random Forest algorithm performed best with 98.1% accuracy, while for the second dataset, K-Nearest Neighbors (KNN) algorithm achieved 97.2% accuracy. This research demonstrates the potential of combining IoT and ML technologies for efficient, real-time water quality monitoring. The developed system offers a scalable and cost-effective solution for continuous assessment of water quality parameters, which can significantly improve water management in rivers and streams.