Performance Evaluation of Machine Learning Models for Anomaly Detection in Sensor Tampering Scenarios

Anomaly detection in sensor data in IoT infrastructure is a rising trend among researchers. The rapid growth of IoT applications in various sectors, such as smart cities and healthcare, have resulted in heightened security threats, with attackers exploiting weaknesses through tactics like tampering, scanning, and malicious operations. This study focuses on addressing the issue of IoT sensor tampering in office environments. For that performance of four different models, two of which are classical machine learning algorithms, and two are deep learning configurations, have been compared to accurately detect sensor tampering in IoT systems. Among the four different models, KNN got the highest accuracy rate of 93.43% with a minimum false positive rate.