Assessing the Integrity of Mobile Platform-Based Sensor Data Pipelines

This paper describes the development and evaluation of a framework for assessing the integrity of sensor data flows within mobile sensor-reliant systems such as UAVs, robots, and autonomous vehicles. These platforms rely on internal pipelines for collecting, processing, and distributing sensor inputs to their command and control components. If the integrity of this sensor data feed is compromised then the platform could pose a safety risk. Hence, the new framework is designed to be integrated into the platform so it can detect compromised sensor data pipelines. It ingests raw data directly from sensor input groups and predicts the corresponding values which should emerge from the pipeline under normal conditions. It uses multi-output prediction models for prediction. Predicted and actual output value are compared to determine if the data pipeline has been compromised. If properly integrated, this framework could increase cyber-physical system resilience.