| Résumé | With the ever-growing reliance on IoT-enabled sensors to age in place, a need arises to protect them from malicious actors and detect malfunctions. In an IoT smart home, it is reasonable to hypothesize that sensors near one another can exhibit linear or nonlinear correlations. If substantiated, this property can be beneficial for constructing relationship trends between the sensors and, consequently, detecting attacks or other anomalies by measuring the deviation of their readings against these trends. In this work, we confirm the presence of correlations between co-located sensors by statistically analyzing two public smart-home datasets and a dataset we collected from our experimental setup. Additionally, we leverage the sliding window approach and supervised machine learning to develop a contextual-anomaly-detection model. This model reaches a true positive rate of 89.47% and a false positive rate of 0%. Our work not only substantiates the correlations but also introduces a novel anomaly-detection technique to enhance security in IoT smart homes. |
|---|
