Context-Aware Anomaly Detection for the Distributed Data Validation Network in Industry 4.0 Environments

In the Industry 4.0 context, especially when considering large factories producing costly goods, monitoring sensor values is important to ensure high quality. This reduces large costs for mending faulty products or recall of those. Different approaches are used to ensure efficient monitoring and validation of sensor values. The Distributed Data Validation Network (DDVN) can remove single points of failure. Still, not every anomaly in the validation procedure means that errors or attacks have occurred. Other reasons like maintenance procedures, updates of firmware, or changed materials can lead to False-Positive (FP) or False-Negative (FN) detection of errors. To reduce these, we incorporate context information in the validation procedure. Further, we show how the appropriate context information is selected and used on a real machine data set.