A New Data Model for Behavioral Based Anomaly Detection in IoT Device Monitoring

The paper presents the new model of data for Internet of Things (IoT) devices monitoring and anomaly detection. The model bases mostly on behavioral description of current state of device, however it contains also some additional information. Raw input data, coming from the external simulation software, are aggregated on two levels of detail: raw variable values preprocessing and time-based aggregation. It was shown, that a sample data following this model, data that contains anomalies, can be analyzed with standard anomaly detection methods and results of this application are very satisfactory. The data used in the paper are also publicly available.