IoT sensor diagnostics through anomalous data detection and classification

Le résumé fourni par la source

Reliable monitoring of environmental variables is essential for effective decision-making in agricultural IoT systems. However, sensors deployed in harsh rural conditions often experience progressive degradation, generating anomalous data that can lead to incorrect operation and reduced crop yield. This work presents an algorithm to detect and classify anomalous sensor data and to estimate a Sensor Quality Indicator [SQI] to facilitate timely sensor replacement. The method integrates two complementary detection procedures: verification of physical limits defined by sensor manufacturers and statistical monitoring using a Six Sigma-based approach on sliding windows. Detected anomalies are classified according to a domain-specific typology, including patterns of outliers, noise, peaks, stepped values, stuck values, and extreme values. The SQI quantifies the severity and frequency of the anomalies to assess sensor reliability. The algorithm was validated in six climate-controlled experimental chambers for Capsicum chinense. The results demonstrate the effectiveness of ICS-based anomaly detection, classification, and degradation assessment for agricultural IoT implementations.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.