A clustering algorithm for detecting differential deviations in the multivariate time-series IoT data based on sensor relationship

Internet-of-things (IoT) applications involve a large number of sensors reporting data as a set of time series. Often these data are related to each other based on the relationship of the sensors in the actual application. Any small deviations could indicate a change in the operation of the IoT system and potential problems with the IoT application's goals. It is often difficult to detect such deviations with respect to the relationship between the sensors. This paper presents the clustering algorithm that can efficiently detect all the deviations small or large in the complex and evolving IoT data streams with the help of sensor relationships. We have demonstrated with the help of experiments that our algorithm can efficiently handle high-dimensional data and accurately detect all the deviations. In this paper, we have presented two more algorithms, anomaly detection and outlier detection, that can efficiently categorize the deviations detected by our proposed clustering algorithm into anomalous or normal deviations. We have evaluated the performance and accuracy of our proposed algorithms on synthetic and real-world datasets. Furthermore, to check the effectiveness of our algorithms in terms of efficiency, we have prepared synthetic datasets in which we have increased the complexity of the deviations to show that our algorithm can handle complex IoT data streams efficiently.