A Framework for Evaluating Security in the Presence of Signal Injection Attacks
share
Sensors are embedded in security-critical applications from medical devices to nuclear power plants, but their outputs can be spoofed through signals transmitted by attackers at a distance. To address the lack of a unifying framework for evaluating the effect of such transmissions, we introduce a system and threat model for signal injection attacks. Our model abstracts away from specific circuit-design issues, and highlights the need to characterize the response of Analog-to-Digital Converters (ADCs) beyond their Nyquist frequency. This ADC characterization can be conducted using direct power injections, reducing the amount of circuit-specific experiments to be performed. We further define the concepts of existential, selective, and universal security, which address attacker goals from mere disruptions of the sensor readings to precise waveform injections. As security in our framework is not binary, it allows for the direct comparison of the level of security between different systems. We additionally conduct extensive experiments across all major ADC types, and demonstrate that an attacker can inject complex waveforms such as human speech into ADCs by transmitting amplitude-modulated (AM) signals over carrier frequencies up to the GHz range. All ADCs we test are vulnerable and demodulate the injected AM signal, although some require a more fine-tuned selection of the carrier frequency. We finally introduce an algorithm which allows circuit designers to concretely calculate the security level of real systems, and we apply our definitions and algorithm in practice using measurements of injections against a smartphone microphone. Overall, our work highlights the importance of evaluating the susceptibility of systems against signal injection attacks, and introduces both the terminology and the methodology to do so.
READ FULL TEXT