Accelerated Reliability and Efficacy Testing of Intrusion Detection Systems for Self-Driving Vehicles

Jeremy Straub


Self-driving vehicles are poised to provide a wealth of benefits ranging from reducing traffic congestion and travel times to increasing safety. The command, control and coordination systems of these vehicles, however, represent targets to nefarious individuals. Problematically, these systems must operate in numerous conditions in a multitude of locations. It is impractical to test every possible set of inputs that may be presented to the system; however, operations (or graceful failure) must be assured. This paper considers accelerated testing as a prospective solution to validate the system’s safety, reliability and efficacy. It discusses prior and current testing techniques for these and similar systems. Then an accelerated testing regime for self-driving vehicles and multi-vehicle systems is presented and assessed. The assessment focuses on identifying classes of issues that can and cannot be readily detected via accelerated testing. Finally, areas of current and future work related to self-driving vehicles are also discussed.