SCREAM: Sensory Channel Remote Execution Attack Methods

Sensory channel threats on embedded systems are an often overlooked attack vector. Because many computing systems focus on digital communication, much of the security research for embedded systems has focused on securing the communication channels between devices. This project explores the concept of sensory channel attacks and demonstrates that an attack on an embedded device purely through sensory channel input can achieve arbitrary code execution. Unlike previous research on sensory channel attacks, this work does not require the device to have preloaded malware. We demonstrate that our attacks were successful in two separate, realistic applications with up to a 100.00% success rate. Finally, we propose a possible defense to these attacks and suggest future avenues of research in this field.