Modeling and Mitigating Side Channels in Optical and Embedded Sensing Systems

PASSCODE: YanPhD

Cyber-physical systems (CPS) such as mobile and wearable devices, Internet of Things, and autonomous vehicles are becoming ubiquitous in public and private spaces. While CPS depends on sensors to process physical information, the increasingly complex sensor hardware and the lack of low-level data protection and privacy controls create security challenges that are caused by the fundamental problem of sensor side channels. Such sensor side channels are challenging to prevent due to the undefined interactions between physical signals, sensor semiconductors, and downstream software.

Using the example of camera sensing and common embedded sensors, my thesis explains how to characterize the causality, limits, and mitigations of sensor side channels through physics modeling and simulation. I will first explain how camera images can not only leak sensitive optical information, but also leak room audio unwittingly modulated in pixels. Then I will demonstrate how the electromagnetic leakage from smart home camera circuits allows eavesdroppers to reconstruct real-time, high-quality camera videos even through walls. Besides information leakage, I will explain how sensor side channels also allow adversaries to inject false information into CPS infrastructure, and demonstrate the potential of utilizing these channels to enable novel multimodal sensing functionalities for enhancing the security of emerging technologies.

CO-CHAIRS: Professors Kevin Fu & Mingyan Liu