Archives: 연구 > 연구성과

[Professor Song Min Kim and first author Kang Min Bae, from left to right]

On the 28th, School of EE professor Song Min Kim’s research team has announced that they have succeeded in creating the world’s first mmWave backscatter system for massive IoT connectivity.

The research, (OmniScatter: extreme sensitivity mmWave backscattering using commodity FMCW radar), led by Kang Min Bae as first author, was presented at ACM MobiSys 2022 this June, and was presented with the best paper award. This is meaningful as it marks the second consecutive year in which the best paper award was presented to a paper belonging to a research group at KAIST’s School of Electrical Engineering.

The backscatter technology described by this research team can greatly reduce the maintenance cost as it operates on ultra-lower power of less than 10 μW, being able to run on a single battery for more than 40 years.

By enabling connectivity on a scale that far exceeds the network density required by next gen communication technologies such as 5G and 6G, this system may serve as a great potential for serving as a stepping stone for the upcoming hyperconnected era.

“mmWave backscattering is a dreamlike technology that can run IoT devices on a large scale, which can drive massive communications at ultra-low power compared to any other technology,” said Professor Song Min Kim. “We hope that this technology will be actively used for the upcoming era of Internet of Things,” he added.

The research was made possible by the funding from Samsung Future Technology Development Project and the Institute for Information & Communication Technology Planning & Evalution.

[Fig 1. Tags used for massive IoT communications (as depicted by red triangles). Over 1100 tags are able to communicate simultaneously without any conflicts]

KAIST EE Prof. Jang, Min Seok and his research team succeed in observing strongly confined mid-infrared light propagating on monocrystalline gold with a scattering-type scanning near-field optical microscope (s-SNOM). 

It is highly applicable in next-generation optoelectronic devices development, with increasing the interaction between light and matter by confining the light in an atomically flat nanostructure. The study will be useful in advancing high-efficient nanophotonics and quantum computing.

[From left, Prof. Jang, Min Seok and Research Prof. Sergey Menabde]

KAIST announced on the July 18th that a joint research effort has succeeded in implementing a new platform for strongly confined light propagation in a low-dimensional material. This finding is expected to contribute to the development of next-generation optoelectronic devices development based on strong light-matter interactions.

Stacking atomically flat two-dimensional materials results in van der Waals crystals, which exhibit properties different from the original two-dimensional materials. Phonon-polaritons are the composite quasi-particles resulting from the polar dielectric ions’ oscillations coupling with electromagnetic waves. In particular, phonon-polaritons forming in van der Waals crystals placed on highly conductive metals have a high degree of compression. This is because the charges in the polariton crystals reflect in the underneath metal due to the image charge effect, and thus produce a new kind of polariton called image phonon-polaritons.

Light propagating as image phonon-polaritons may induce strong light-matter interactions, but their propagation is limited on rough metal surfaces and thus suffers low feasibility.

Prof. Jang said, “This work illustrates very well the advantages of image polaritons, especially image phonon-polaritons. Image phonon-polaritons exhibit low loss and strong light-matter interactions useful in the development of next-generation optoelectronic devices.” He then added that he hopes to advance the commercialization of high-efficiency nanophotonic devices, including in metasurfaces, optical switches, and optical sensors.

This work, first-authored by Research Prof. Sergey Menabde, was published in Science Advances on the July 13th. It has been supported by Samsung Science & Technology Foundation and the National Research Foundation of Korea, as well as Korea Institute of Science and Technology, Japanese Ministry of Education, Culture, Sports, Science and Technology, and the Villum Foundation of Denmark.

Fig. 1 Nanotip used in measuring the image phonon-polaritons traveling in h-BN in super-high resolution

[Prof. Myoungsoo Jung, Miryeong Kwon , Donghyun Gouk, from left]

Title: EE Professor Yoo, Chang Dong and  Kweon, In So ’s Research Team Give Oral Presentation on Self-supervised Learning at ECCV 2022

KAIST EE Prof. Yoo, Chang Dong and Kweon, In So’s team conducted a joint research and proposed a self-supervised learning method that is remarkably robust and performs well with even only a small volume of labeled data.

<(From left) EE Professors Yoo, Chang Dong and Kweon, In So and Researchers Chaoning Zhang and Kang Zhang>

ECCV began in 1990 and has since focused on introducing the latest findings in artificial intelligence and machine learning research on vision and signal processing. It has long been a renowned conference on computer vision and deep learning, and its 2022 rendition gathered 5,803 submissions, only 1,650 (28%) of which have had the honor of being accepted, and merely a select 158 (2.7%) of the accepts given the opportunity for an oral presentation.

The team’s findings titled “Decoupled Adversarial Contrastive Learning for Self-supervised Adversarial Robustness” earned the oral presentation honor and will be presented on Oct. 23, 2022 in Tel Aviv, Israel.

While artificial intelligence is making progress in various domains, it has yet to win full trust from humans. Reliable learning should encompass learning from little data as well as robust learning, and attempts at this objective have been made with combining self-supervised learning and adversarial learning. This work utilizes distillation methods to efficiently put together the two and proposed an adversarial learning framework capable of self-supervised learning without labels.

The paper outlining these findings has been selected as an ECCV Oral Presentation (acceptance rate 2.7%) work. The work is a joint endeavor by Professors Yoo, Chang Dong and Kweon, In So, and their team, and it promises exciting opportunities for providing high-performance services based on robust artificial intelligence learning from little data.

This research is supported by IITP by MSIT.

[Prof. Sung-Ju Lee, Jaemin Shin (KAIST PhD Candidate),  Prof. Yunxin Liu of  Tsinghua University , Prof. Yuanchun Li  of  Tsinghua University, from left]

The research team led by Prof. Sung-Ju Lee of KAIST has published a paper “FedBalancer: Data and Pace Control for Efficient Federated Learning on Heterogeneous Clients” at ACM MobiSys (International Conference on Mobile Systems, Applications, and Services) 2022. Founded in 2003, MobiSys has been a premier conference on Mobile Computing and Systems. This year, 38 out of 176 submitted papers have been accepted to be presented at the conference. 

Jaemin Shin (KAIST PhD Candidate) was the lead author, and this work was in collaboration with Tsinghua University in China (Professors Yuanchun Li and Yunxin Liu participated).

Federated Learning is a recent machine learning paradigm proposed by Google that trains on a large corpus of private user data without collecting them. The authors developed a systematic federated learning framework that accelerates the global learning process of federated learning. The new framework actively measures the contribution of each training sample of clients and selects the optimal samples to optimize the training process. The authors also included an adaptive deadline control scheme with varying training data, and achieved 4.5 times speedup in global learning process without sacrificing the model accuracy.

Prof. Lee stated that “Federated learning is an important technology used by many top companies. With the accelerated training time achieved by this research, it has become even more attractive for real deployments. Moreover, our technology has shown to work well in different domains such as computer vision, natural language processing, and human activity recognition.”

(Prof. Yong Man Ro, Sangmin Lee, Sungjune Park,  from left)

Ph.D. Candidate Sangmin Lee and Sungjune Park (Prof. Yong Man Ro’s Lab) won the 1st place in the Ad-hoc Video Search (AVS) section of the 11th Video Browser Showdown (VBS 2022).

VBS is the international video retrieval competition held annually, and this year VBS 2022 is the 11th competition.

This year’s competition was held at Vietnam Phú Quốc for two days from June 6th to 7th, with 16 finalized video search teams from around the world.

The Ad-hoc Video Search section is to find as exact videos for given querys out of 2.5 million videos.

Sangmin Lee and Sungjune Park won the first place by constructing a multimodal search engine based on deep learning, which effectively searches target videos through the multi-modal correspondences of visual-audio-language latent representations.

The core algorithm adopted in the search engine, novel visual-audio representation learning method will be presented at CVPR 2022, the top tier conference in computer vision and AI field.

The title of the paper is “Weakly Paired Associative Learning for Sound and Image Representations via Bimodal Associative Memory”.

– Competition: 11th Video Browser Showdown 2022

– Award: Best AVS (1st place winner in Ad-hoc Video Search)

– Recipient: Sangmin Lee, Sungjune Park, Yong Man Ro (Advisory Professor)

EE Prof. Kim, SangHyeon’s team, develops display using 3D integration techniques, promising applications on next generation high resolution displays

[ Prof. Kim, SangHyeon, Ju Hyeok Park (P.H.D candidate), Dr. Dae-myeong Geum, Woo Jin Baek (P.H.D candidate), From left]

KAIST EE Prof. Kim, SangHyeon’s research team has successfully developed a 1600-PPI MicroLED display by utilizing monolithic 3D integration techniques, as announced.

(*monolithic 3D integration: dubbed the ultimate 3D integration tech, wherein after the lower-layer devices, the upper layer’s thin film is created and stacking proceeds sequentially so as to maximize the upper-lower device alignment)

(* PPI: pixels per inch)

KAIST EE Prof. Kim, SangHyeon Kim’s research team members Ju Hyeok Park and Dr. Dae-myeong Geum led the work as co-first authors, collaborating with Woo Jin Baek from the same research lab and Dr. Johnson Shieh from Jasper Display in Taiwan. Their joint work has been presented at the “semiconductor Olympics”, the 2022 IEEE Symposium on VLSI Technology & Circuits. (Paper: Monolithic 3D sequential integration realizing 1600-PPI red micro-LED display on Si CMOS driver IC)

MicroLED devices using inorganic-based III-V compound semiconductors are gaining attention as core candidates for next-generation ultra-high resolution displays that are growing rapidly in demand. MicroLEDs offer advantages over current OLED and LCD displays widely used in modern TVs and mobile devices with features such as high luminance and contrast ratio, and a longer pixel life.

(*III-V compound semiconductors: Semiconductors comprising of compounds of Group III and Group V elements in the periodic table, offering excellent charge transport and light characteristics)

A monolithic 3D integration of red light-emitting LEDs on a Si CMOS circuit board was applied to solve the issues present in existing device technology. A demonstration of high-resolution display was made successful through continuous semiconductor processes on the wafer. Through this process, the LED semiconductor display layer was designed to reduce the thickness of the active layer for light emission to 1/3 and greatly reduce the challenges of the etching process required for pixel formation. In addition, to prevent performance degradation of the lower display driving circuit, the research team was able to maintain the performance of the lower Driver IC even after the integration of the upper layer by using ultra-low temperature processes such as wafer bonding that integrates the upper III-V layer below 350 C.

By successfully implementing state-of-the-art resolution of 1600-PPI MicroLED display using a monolithic 3D integration of red LEDs, this result is expected pave way for the next-generation ultra-high resolution displays.

Image 1.

Image 2.

Title:  EE Professor Kim, SangHyeon’s Research Team Develops 3D Stackable Quantum Computing Readout Device

KAIST Builds 3D Stackable Quantum Computing Readout Device  Low-power, low-noise, high-speed device integrated in 3D operates at super-low temperatures and promises large-scale applications to quantum computing devices.

<(From left) EE Prof. Kim, SangHyeon, PhD candidate Jeong, Jae Yong, NanoFab PhD candidate Kim, Jongmin, and KBSI Prof. Park, Seung-Young>

KAIST EE Prof. Kim, SangHyeon’s research team has developed a 3D-stacked semiconductor readout device integration technology, as made public on the 16th. The team made this possible by applying the strengths of monolithic 3D integration to overcome large-scale qubit implementation based on existing quantum computing systems. Their work is a first of its kind exhibiting the 3D stackability of quantum computing readout devices after an actively pursued line of research on monolithic 3D stacking of high-speed devices following a 2021 VLSI presentation, a 2021 IEDM presentation, and a 2022 ACS Nano publication.

(*monolithic 3D integration: dubbed the ultimate 3D integration tech, wherein after the lower-layer devices, the upper layer’s thin film is created and stacking proceeds sequentially so as to maximize the upper-lower device alignment)

KAIST EE Prof. Kim, SangHyeon Kim’s research team member Jeong, Jae Yong led the work as first author, collaborating with NanoFab PhD candidate Kim, Jongmin and KBSI Prof. Park, Seung-Young. Their joint work has been presented at the “semiconductor Olympics”, Symposium on VLSI Technology. (Paper: 3D stackable cryogenic InGaAs HEMTs for heterogeneous and monolithic 3D integrated highly scalable quantum computing system)

A qubit is capable of processing twice the amount computation compared with that of a bit. Number of qubits increasing linearly results in exponential speedup of their computation. Thus, developing large-scale quantum computing is of utmost importance. IBM, for instance, presented Eagle containing 127 qubits, and the IBM roadmap outlines development of a 4,000-qubit quantum computer by 2025 and one with 10,000-qubits or more in 10 years.

Designing such large-scale quantum computers with many qubits requires implementing devices for qubit control/readout. The research team has not only proposed and implemented 3D-stacked control/readout devices but also achieved world-best cutoff frequency characteristics at cryogenic settings despite the 3D stacking.

This work has been supported by the National Research Foundation of Korea, the System Semiconductor Development Program funded by Gyeonggi-do, and the Korea Basic Science Institute.

 EE Professors Kim, Changick and Jeong, Jae-Woong Awarded on 2022 KAIST Research Day.

[From left, Prof. Kim, Changick, Prof. Jeong, Jae-Woong]

Professor Kim, Changick has been recognized with the Transdisciplinary Research Prize for his contributions to computer vision- and artificial intelligence-based monitoring  technology of anthropocene effects on the planet. Anthropocene is a scientific concept referring to the recent geological epoch distinct from previous ones, marked by unprecedented transformations in the planet’s system from human activities since the Industrial Revolution. Prof. Kim has been conducting research with satellite images, computer modeling, and deep learning tools on monitoring the compromised states of planet Earth, such as climate change and sea level rises. In addition, as part of AI-based digital study of ecology, he has cooperated closely with anthropogeography and ecology experts to detect endangered species in the DMZ; he has developed a deep network capable of counting and classifying endangered species, such as the red-crowned cranes, the white-naped cranes, and the white-fronted geese. This study is meaningful in automating and maintaining the monitoring process of endangered species in the DMZ and Cheorwon.

Professor Jeong, Jae-Woong has been awarded the KAIST Scholastic Award for proposing a new direction in the automated treatment of brain diseases and cognitive research by developing for the first time an IoT (Internet of Things) based wireless remote control system for neural circuits in the brain. The proposed direction sets a vision for one of humanity’s most difficult challenges: overcoming brain diseases. Prof. Jeong has also led the field of research in wirelessly rechargeable soft subdermal implantable devices. These works have been published in 2021 in top journals of medical engineering: Nature Biomedical Engineering and Nature Communications. Said studies were led by Prof. Jeong’s team, with international collaborators in Washington University in School of Medicine, attracting over 60 press reports across the world.