Qiangfei Xia, electrical and computer engineering at UMass Amherst, has received a three-year, $499,998 grant from the National Science Foundation (NSF) to combine on-chip sensing and computing functions for next-generation smart, energy-efficient wearable electronics.
Today’s wearable devices require sensing and computing functions to happen separately, resulting in a massive amount of data having to shuttle between the sensor module and a cloud computing platform.
As the number of nodes used in sensor networks continue to grow to enable greater functionality in wearable devices, the larger the amount of redundant data being exchanged between the separate sensor and computing units, which takes more time and uses more energy.
“This is unsustainable in terms of energy efficiency, latency, and the capacity for processing sensor data,” Xia says. “Integrating sensing, memory, and computing functionalities into one chip will allow for ubiquitous applications at low energy budget and low latency.”
With billions of sensors embedded in billions of internet-connected devices worldwide, reducing each sensor’s energy needs could have significant environmental impact. Building these systems on flexible materials will also allow next-generation devices to be “affordable and biodegradable smart-wearable electronics capable of monitoring human health continuously and adaptively,” Xia says.
The project will also demonstrate an in-sensor computing platform in which ferroelectric field-effect transistors will be used as both sensors and nonvolatile elements for in-memory computing, meaning the transistors can retain stored information even when the power is off. Collaborative partner VTT, a technical research center in Finland, will develop the flexible ferroelectric sensors.
Educational and outreach activities proposed under the grant will promote STEM careers and encourage diversity in engineering education and research.
Xia is also principle investigator of a $1.3 million award from the NSF ASCENT (Addressing Systems Challenges through Engineering Teams) program, which began in September 2020 and ends in August 2024. The project aims to develop 3D memristive neural networks for real-time machine learning applications.