Sensor Data Integrity Verification for Real-time and Resource Constrained Systems

Abstract

Sensors are used in multiple applications that touch our lives and have become an integral part of modern life. They are used in building intelligent control systems in various industries like healthcare, transportation, consumer electronics, military, etc. Many mission-critical applications require sensor data to be secure and authentic. Sensor data security can be achieved using traditional solutions like cryptography and digital signatures, but these techniques are computationally intensive and cannot be easily applied to resource constrained systems. Low complexity data hiding techniques, on the contrary, are easy to implement and do not need substantial processing power or memory. In this applied research, we use and configure the established low complexity data hiding techniques from the multimedia forensics domain. These techniques are used to secure the sensor data transmissions in resource constrained and real-time environments such as an autonomous vehicle. We identify the areas in an autonomous vehicle that require sensor data integrity and propose suitable water-marking techniques to verify the integrity of the data and evaluate the performance of the proposed method against different attack vectors. In our proposed method, sensor data is embedded with application specific metadata and this process introduces some distortion. We analyze this embedding induced distortion and its impact on the overall sensor data quality to conclude that watermarking techniques, when properly configured, can solve sensor data integrity verification problems in an autonomous vehicle.