Joint Fault and Intrusion Detection For Industrial Internet of Things Sensing Environments
This on-going project aims to develop an Industrial Internet of Things (IIoT) testbed for undergraduate research and education. As Industry 4.0 becomes increasingly adopted in the mainstream factory environment, the burden of managing the physical sensing is placed upon the factory's network, presenting great information security risk as these IIoT devices are often significantly less secure than typical network nodes. With this research we provide a methodology for the implementation of insecure sensor devices in fault detection and isolation (FDI) while preserving the integrity of the environmental data in the event of network intrusion. This methodology will first determine if network intrusions influence data integrity through statistical analysis. Threshold and machine learning detection methods will be used to discover intruded datapoints and remove or remedy them before they enter the FDI systems. This research plans to show that defending against these intrusions is possible in the worst-case scenario, where only the telemetry data is available to detect from. Protecting the input into the FDI system is crucial. As reliance upon FDI systems, allowing them greater control over the factory environment, increases, the potential production risk due to intrusion increases in tandem. The current version of the testbed can retrieve single data telemetry from thermocouple and humidity sensors and display the output through a webpage and downloadable csv file.