Research

Efficient Computing for Machine Learning at the Edge

We make machine learning computing faster and more efficient for embedded applications with scarce compute resources. We develop various design-time and run-time techniques in circuit, architecture, and algorithm levels, such as neural network model reduction and energy-quality scalable computing.

Beyond Approximate Computing: Enabling Full-System Energy-Quality Scalability in Embedded Systems

National Science Foundation

NSF AI Institute for Edge Computing Leveraging Next Generation Networks

National Science Foundation

SynthNet (ASP-DAC 2022)

A high-throughput yet energy-efficient combinational logic neural network

SECO (ISLPED 2019)

A scalable accuracy approximate exponential function via cross-level optimization

SAADI (ASP-DAC 2019)

A scalable accuracy approximate divider for dynamic energy-quality scaling

Usable Security and Privacy for Mobile and Embedded Systems

We make the Internet-of-Things and human-computer interaction (VR/AR) more trustworthy through usability-aware security and privacy. We develop techniques for ensuring a high level of security and privacy that are easy to understand and use for the users.

Kaleido (USENIX Security 2021)

Real-time privacy control for eye-tracking systems

VoltKey (UbiComp 2019)

Continuous secret key generation based on power line noise for zero-involvement pairing and authentication

CamPUF (DAC 2018)

Physically unclonable function based on CMOS image sensor fixed pattern noise

Embedded Systems Applications

We solve societal problems in areas such as healthcare, agriculture, and environment using embedded computing and networking technologies. We utilize advanced Internet-of-Things and machine learning techniques for monitoring, data analytics, and decision-making.