Hardware for Internet of Things (IoT)

Barry Johnson

Benton Calhoun

Brad Campbell

Brian Park

Cong Shen

Haiying Shen

Kun Qian

Kyusang Lee

Madhur Behl

Mircea Stan

Qiaochu Zhang

Qing (Cindy) Chang

Shangtong Zhang

Somayeh Asadi

Steve Bowers

Yixin Sun

Zongli Lin

Search publications by keyword, author, year, or journal to learn more about the research emerging from the Link Lab.

Publications

QIAOCHU ZHANG

This paper introduces an adaptive sampling and active learning approach to accelerate analog mixed-signal (AMS) circuit optimization by using machine learning regression models to efficiently identify promising parameter combinations while minimizing expensive circuit simulations. Experimental results show that the method discovers higher-quality Pareto-optimal designs at lower computational cost than prior techniques, with particularly strong advantages for complex AMS circuits and, in some cases, outperforming expert-designed circuit implementations.

PUBLICATION: An Active Learning Framework for Analog Circuit Multi-objective Customization

Zhu, Mutian; Hassanpourghadi, Mohsen; Zhang, Qiaochu; Chen, Mike Shuo-Wei; Levi, A.F.J.; Gupta, Sandeep, ACM Transactions on Design Automation of Electronic Systems, 2026.

BRAD CAMPBELL

This work introduces a privacy-preserving federated learning framework for IoT applications that protects user data by separating shared and personalized model components while limiting the impact of privacy protections on model performance. Evaluations across multiple real-world IoT datasets demonstrate that the approach maintains high accuracy with only modest performance loss, substantially outperforming existing locally differentially private federated learning methods.

PUBLICATION: Atlas: Ensuring Accuracy for Privacy-Preserving Federated IoT Applications

Gao, Jiechao; Tang, Mingyue; Wang, Wenpeng; Routh, Tushar; Campbell, Bradford, Proceedings of the ACM/IEEE 16th International Conference on Cyber-Physical Systems (with CPS-IoT Week 2025), 2025

BEN CALHOUN

This research presents a compact, low-power communication converter that efficiently bridges I2C and SPI interfaces for electronic textile systems without requiring data buffering or internal clock generation. The design significantly reduces chip size and energy consumption compared with existing commercial solutions while maintaining reliable high-speed performance, making it well suited for distributed wearable and smart textile applications.

PUBLICATION: A Compact, Power-Efficient, and On-the-Fly I2C-to-SPI Converter for Distributed E-Textile Systems

Faruqe, Omar; Chen, Zhenghong; Bhattacharya, Suprio; Foysal, Md. Fahim; Hasan, Samit; Wang, Jinhua; Truesdell, Daniel S.; Calhoun, Benton H., IEEE Transactions on Circuits and Systems I: Regular Papers, 2025

KUN QIAN

Radar sensors are increasingly used in IoT applications such as healthcare, smart homes, industrial automation, and transportation, but their widespread adoption is limited by the high-power consumption of both radar hardware and the artificial neural networks (ANNs) commonly used for signal processing. ANNs consume significant energy because they rely on continuous neuron activations and the von Neumann computing architecture, which requires frequent data transfers between memory and processing units, making energy efficiency a critical challenge for battery-powered and wearable IoT devices.

PUBLICATION: NeuroRadar: A Neuromorphic Radar Sensor for Low-Power IoT Systems

Kai Zheng, Kun Qian, Timothy Woodford, Xinyu Zhang, Communications of the ACM, Volume 68, Issue 9. Pages 91 - 100. 21 August 2025.