Abstract
<jats:p>In sustainable aquaculture, maintaining optimal water quality is essential, as sudden environmental changes can lead to disease outbreaks, reduced productivity, and significant fish mortality. This paper presents an edge-assisted Internet of Things (IoT)-Cyber Physical System (CPS) framework for intelligent water quality monitoring and real-time anomaly detection. The proposed framework integrates IoT-enabled water quality sensors, an ESP32 edge device, and a Raspberry Pi 4 Model B to perform real-time data acquisition, preprocessing, and anomaly detection using the Isolation Forest algorithm. While the ESP32 is responsible for collecting and transmitting sensor data, the Raspberry Pi executes the computationally intensive anomaly detection process and supports intelligent early-warning generation with low communication latency. Four critical water quality parameters, namely temperature, dissolved oxygen (DO), pH, and turbidity, were selected from the publicly available IoTMLCQ dataset. After data preprocessing and feature normalization, the proposed framework was evaluated using confusion matrix analysis, receiver operating characteristic (ROC) analysis, and standard classification metrics. Experimental results demonstrate that the proposed model achieved an accuracy of 97.25%, a precision of 66.67%, a recall of 66.45%, an F1-score of 66.56%, and an ROC-AUC of 98.44%, outperforming One-Class Support Vector Machine and Local Outlier Factor under identical experimental conditions. The proposed low-cost edge-assisted architecture provides timely anomaly detection and intelligent early warning, offering a practical solution to reduce fish losses, improve aquaculture farm management, and support sustainable smart aquaculture.</jats:p>