Data Quality in IoT Temperature Sensor Systems: Demonstrated on Time-Dependent Temperature Fluctuations

The growing integration of the Internet of Things (IoT) in Industry 4.0 is primarily attributed to its low cost and high adaptability. Our research addresses the expansive application of IoT temperature sensor systems within industrial environments characterized by harsh conditions and rapid temperature fluctuations. Typically, these sensors lack awareness of the data quality they measure and provide. The article mitigates this limitation by introducing data quality models that enhance the accuracy and timeliness of temperature sensor data. These models interpret temperature readings affected by thermal lag and delayed response times. Timeliness is determined using three distinct methods that reflect data volatility, while accuracy is assessed by applying the Savitzky-Golay filter, which provides temperature transients. Both data quality models precede a knowledge-based approach, accumulating necessary information for the sensor system through calibration. These quality categories are combined into a unified Quality of Sensing (QoS) parameter, utilizing fuzzy logic to abstract the quality dimensions, thereby having profound implications for the IoT sensor system. This methodology has been encapsulated within a comprehensive framework and validated using an industrial-grade IoT temperature sensor in a real-world measurement setup. By transcending the conventional reliance on raw sensor outputs, our framework empowers the extraction of intrinsic quality information associated with each temperature reading, enhancing the resilience and reliability of IoT temperature sensor systems.