-
mmWave Wearable Antenna for Interaction with VR Devices
Authors:
Haksun Son,
Song Min Kim
Abstract:
The VR industry is one of the most promising industries for the near future, as it can provide a more immersive connection between people and the virtual world. Currently, VR devices interact with people using inconvenient controllers or cameras that perform poorly in dark environments. Interaction through millimeter-wave wearable devices has the potential to conveniently track human behavior rega…
▽ More
The VR industry is one of the most promising industries for the near future, as it can provide a more immersive connection between people and the virtual world. Currently, VR devices interact with people using inconvenient controllers or cameras that perform poorly in dark environments. Interaction through millimeter-wave wearable devices has the potential to conveniently track human behavior regardless of the lighting conditions. In this study, a millimeter-wave wearable antenna was developed, opening up the possibility for more immersive interaction with VR devices. The antenna features a low loss tangent polyester fabric to minimize dielectric losses and a smooth coating to reduce losses due to rough surfaces. The antenna operates in the 24GHz ISM band, with an S11 value of -29dB at 24.15GHz.
△ Less
Submitted 19 April, 2024;
originally announced April 2024.
-
NR-Surface: NextG-ready $μ$W-reconfigurable mmWave Metasurface
Authors:
Minseok Kim,
Namjo Ahn,
Song Min Kim
Abstract:
Metasurface has recently emerged as an economic solution to expand mmWave coverage. However, their pervasive deployment remains a challenge, mainly due to the difficulty in reaching the tight 260ns NR synchronization requirement and real-time wireless reconfiguration while maintaining multi-year battery life. This paper presents NR-Surface, the first real-time reconfigurable metasurface fully comp…
▽ More
Metasurface has recently emerged as an economic solution to expand mmWave coverage. However, their pervasive deployment remains a challenge, mainly due to the difficulty in reaching the tight 260ns NR synchronization requirement and real-time wireless reconfiguration while maintaining multi-year battery life. This paper presents NR-Surface, the first real-time reconfigurable metasurface fully compliant with the NR standard, operating at 242.7 $μ$W for a 2.1-year lifetime on an AA battery. NR-Surface incorporates (i) a new extremely low-power (14KHz sampling) reconfiguration interface, NarrowBand Packet Unit (NBPU), for synchronization and real-time reconfiguration, and (ii) a highly responsive and low-leakage metasurface designed for low-duty cycled operation, by carefully leveraging the structure and the periodicity of the NR beam management procedure in the NR standard. NR-Surface is prototyped and evaluated end-to-end with NR BS built on srsRAN to demonstrate diverse usage scenarios including multiple NR-Surface per BS, multiple UE per NR-Surface, and 3D beamforming. Around-the-corner UE evaluations showcase NR-Surface efficacy under different user mobility patterns (20.3dB gain) and dynamic blockage (22.2dB gain).
△ Less
Submitted 14 March, 2024;
originally announced March 2024.
-
EM-Network: Oracle Guided Self-distillation for Sequence Learning
Authors:
Ji Won Yoon,
Sunghwan Ahn,
Hyeonseung Lee,
Minchan Kim,
Seok Min Kim,
Nam Soo Kim
Abstract:
We introduce EM-Network, a novel self-distillation approach that effectively leverages target information for supervised sequence-to-sequence (seq2seq) learning. In contrast to conventional methods, it is trained with oracle guidance, which is derived from the target sequence. Since the oracle guidance compactly represents the target-side context that can assist the sequence model in solving the t…
▽ More
We introduce EM-Network, a novel self-distillation approach that effectively leverages target information for supervised sequence-to-sequence (seq2seq) learning. In contrast to conventional methods, it is trained with oracle guidance, which is derived from the target sequence. Since the oracle guidance compactly represents the target-side context that can assist the sequence model in solving the task, the EM-Network achieves a better prediction compared to using only the source input. To allow the sequence model to inherit the promising capability of the EM-Network, we propose a new self-distillation strategy, where the original sequence model can benefit from the knowledge of the EM-Network in a one-stage manner. We conduct comprehensive experiments on two types of seq2seq models: connectionist temporal classification (CTC) for speech recognition and attention-based encoder-decoder (AED) for machine translation. Experimental results demonstrate that the EM-Network significantly advances the current state-of-the-art approaches, improving over the best prior work on speech recognition and establishing state-of-the-art performance on WMT'14 and IWSLT'14.
△ Less
Submitted 14 June, 2023;
originally announced June 2023.
-
MCR-Data2vec 2.0: Improving Self-supervised Speech Pre-training via Model-level Consistency Regularization
Authors:
Ji Won Yoon,
Seok Min Kim,
Nam Soo Kim
Abstract:
Self-supervised learning (SSL) has shown significant progress in speech processing tasks. However, despite the intrinsic randomness in the Transformer structure, such as dropout variants and layer-drop, improving the model-level consistency remains under-explored in the speech SSL literature. To address this, we propose a new pre-training method that uses consistency regularization to improve Data…
▽ More
Self-supervised learning (SSL) has shown significant progress in speech processing tasks. However, despite the intrinsic randomness in the Transformer structure, such as dropout variants and layer-drop, improving the model-level consistency remains under-explored in the speech SSL literature. To address this, we propose a new pre-training method that uses consistency regularization to improve Data2vec 2.0, the recent state-of-the-art (SOTA) SSL model. Specifically, the proposed method involves sampling two different student sub-models within the Data2vec 2.0 framework, enabling two output variants derived from a single input without additional parameters. Subsequently, we regularize the outputs from the student sub-models to be consistent and require them to predict the representation of the teacher model. Our experimental results demonstrate that the proposed approach improves the SSL model's robustness and generalization ability, resulting in SOTA results on the SUPERB benchmark.
△ Less
Submitted 14 June, 2023;
originally announced June 2023.
-
Hawkeye: Hectometer-range Subcentimeter Localization for Large-scale mmWave Backscatter
Authors:
Kang Min Bae,
Hankyeol Moon,
Sung-Min Sohn,
Song Min Kim
Abstract:
Accurate localization of a large number of objects over a wide area is one of the keys to the pervasive interaction with the Internet of Things. This paper presents Hawkeye, a new mmWave backscatter that, for the first time, offers over (i) hundred-scale simultaneous 3D localization at (ii) subcentimeter accuracy for over an (iii) hectometer distance. Hawkeye generally applies to indoors and outdo…
▽ More
Accurate localization of a large number of objects over a wide area is one of the keys to the pervasive interaction with the Internet of Things. This paper presents Hawkeye, a new mmWave backscatter that, for the first time, offers over (i) hundred-scale simultaneous 3D localization at (ii) subcentimeter accuracy for over an (iii) hectometer distance. Hawkeye generally applies to indoors and outdoors as well as under mobility. Hawkeye tag's Van Atta Array design with retro-reflectivity in both elevation and azimuth planes offers 3D localization and effectively suppresses the multipath. Hawkeye localization algorithm is a lightweight signal processing compatible with the commodity FMCW radar. It uniquely leverages the interplay between the tag signal and clutter, and leverages the spetral leakage for fine-grained positioning. Prototype evaluations in corridor, lecture room, and soccer field reveal 6.7 mm median accuracy at 160 m range, and simultaneously localizes 100 tags in only 33.2 ms. Hawkeye is reliable under temperature change with significant oscillator frequency offset.
△ Less
Submitted 6 January, 2023;
originally announced January 2023.
