Heavy Metal Detection and Prediction in Water Using Smart IoT-Enabled System

Abstract

The contamination of water bodies by heavy metals is known to pose significant risks to public health and environmental sustainability, particularly in rapidly industrialized regions such as Indonesia. Conventional laboratory-based methods are costly, time-consuming, and unsuitable for real-time large-scale monitoring. Therefore, this study proposed an AI-enhanced edge-IoT system for real-time detection and prediction of heavy metal contamination in water. The system integrated an anodic stripping voltammetry (ASV) sensor array with an ESP32-based edge computing unit, enabling on-device preprocessing, anomaly detection, and predictive analytics to use lightweight machine learning models. The platform also supported the multi-parameter sensing of lead (Pb), cadmium (Cd), mercury (Hg), and arsenic (As), combined with Wi-Fi/GSM communication, MQTT-based cloud infrastructure, and an interactive monitoring dashboard. Experimental validation was conducted using 120 water samples collected from industrial, agricultural, and urban environments across West Java, Indonesia. Additionally, the proposed system achieved strong agreement with ICP-OES reference measurements, with coefficients of determination (R²) exceeding 0.97 for all target metals. The limits of detection were 0.8 μg/L (Pb), 0.5 μg/L (Cd), 0.3 μg/L (Hg), and 1.1 μg/L (As). The embedded AI model showed reliable short-term prediction of contamination trends and early anomaly detection at the edge level. The proposed method offered a cost-effective, scalable, and field-deployable solution for continuous water quality monitoring, supporting compliance with Indonesian Government Regulation No. 22/2021 and WHO drinking-water standards.