Three research papers by researchers from the Department of Electrical Engineering at UNIST have been accepted for presentation at the 2024 IEEE International Solid-State Circuits Conference (ISSCC).
Known as the ‘Semiconductor Design Olympics,’ ISSCC is the foremost global forum for presentation of advances in solid-state circuits and systems-on-a-chip that has been held since 1954. Marking its 71st anniversary this year, ISSCC 2024 is set to take place in San Francisco from February 18 to 22, next year under the theme of ‘Integrated Circuits for a Better World.’ It will bring together 3,000 academic scientists, researchers, and professionals around the globe to exchange and share their research findings, as well as to discuss the future of semiconductor industries and technologies.
This year, a total of 234 research papers were selected for presentation at the 2024 ISSCC, including 49 papers from Korea. According to the recent ‘ISSCC 2024 Korean Conference,’ three cases of UNIST were acknowledged in the fields of IMMD (Imager, MEMS, Medical and Display) and DAS (Digital Architecture and System). Two papers presented by Professor Seong-Jin Kim and Professor Jae Joon Kim’s research teams were selected in the IMMD category, while the DAS category featured a paper from Professor Kyuho Lee’s research team.
From left are Professor Seong-Jin Kim, Suhyun Han, and SeongHyeok Park.
■ Revolutionary Advancements in LiDAR Sensor Technology
Professor Seong-Jin Kim’s research team made significant strides in the development of a low-power and ultra-small CMOS LiDAR sensor. This breakthrough follows two decades of research and development in the field. The newly created LiDAR sensor boasts a size reduction of 50% compared to previously announced sensors, while consuming 21 times less power. Moreover, it provides high-resolution distance images of 160×120. These improvements were achieved through the implementation of smaller, more power-efficient analog circuits, replacing the larger digital counters and time control circuits used in previous LiDAR sensors. The team also incorporated self-correcting circuits to enhance performance.
This technology holds great promise for integration into mobile metaverse devices and service robots, as it reduces the size and power consumption of LiDAR sensors, which serve as the eyes of autonomous vehicles and robots. While conventional LiDAR sensors offer accurate imaging, their high power consumption and bulkiness limit their usability in mobile environments.
The research received support from Samsung Electronics’ DS division and the Korea Research Foundation.
Professor Kyuho Lee (Second row, center) and his research team, including JuEun Lee (First row, center).
■ Ultra-Low Power Spatial Information Processing AI Processor
Professor Kyuho Lee’s team accomplished a groundbreaking feat by developing the world’s first ultra-low power spatial information processing artificial intelligence system-on-chip semiconductor. This semiconductor exhibits a remarkable 2,044-fold reduction in power consumption compared to commercial products. Acting as the “brain” of a machine, this AI SoC semiconductor processes input information from lidar sensors to infer location and surrounding conditions in real time. It efficiently handles vast amounts of data, including long-distance and 360-degree surround spatial information. Compared to the commercial mobile computing platform Jetson Tx2, energy consumption has been reduced by an impressive 2,044 times, while supporting calculation speeds of up to 48 FPS.
“We have introduced the world’s first low-power system-on-chip semiconductor that combines point cloud deep neural network (PNN) technology with SLAM (Simultaneous Localization and Mapping), a location estimation algorithm. This technology leverages the time-space relationship of 3D information to predict inefficient operations and optimize memory usage,” noted the research team.
Professor Jae Joon Kim (Inside the circle) and his research team, including YouJang Pyeon (Second from the right).
■ Convergence Sensor for Various Diagnostic Applications
Professor Jae Joon Kim’s lab has developed an innovative patch sensor technology known as the ‘convergence sensor.’ This sensor incorporates an analog artificial intelligence circuit within its signal processing chip, enabling autonomous diagnostic and sensing functions. Capable of diagnosing arrhythmia and detecting harmful gases through bio-signals, this sensor stands out as a versatile platform technology. It allows for the attachment and detachment of sensor elements, functioning like stickers that can be combined to create the desired sensing capabilities. The technology, named ‘Adhesive Imposer,’ is the result of collaboration with Professor Heungjoo Shin and Professor Hoon Eui Jeong in the Department of Mechanical Engineering at UNIST.
“By integrating analog circuit-based artificial neural network engines in a single sensor chip, capable of processing multiple biosignals and environmental sensor signals simultaneously, even a small sensor patch can discern various events such as arrhythmia and harmful gas detection within the human body,” explained the research team.
Professor Jae Joon Kim’s lab has been dedicated to studying signal processing semiconductor chips and systems for various sensors, including bio-signals and environmental sensors. ISSCC has previously recognized their work in wearable device technology for bio-signal detection.
These research endeavors were made possible with support from the Ministry of Trade, Industry and Energy, and the Ministry of Science and ICT.
