Deep Learning for Counting People from UWB Channel Impulse Response Signals

The use of higher frequency bands compared to other wireless communication protocols enhances the capability of accurately determining locations from ultra-wideband (UWB) signals. It can also be used to estimate the number of people in a room based on the waveform of the channel impulse response (CIR) from UWB transceivers. In this paper, we apply deep neural networks to UWB CIR signals for the purpose of estimating the number of people in a room. We especially focus on empirically investigating the various network architectures for classification from single UWB CIR data, as well as from various ensemble configurations. We present our processes for acquiring and preprocessing CIR data, our designs of the different network architectures and ensembles that were applied, and the comparative experimental evaluations. We demonstrate that deep neural networks can accurately classify the number of people within a Line of Sight (LoS), thereby achieving an 99% performance and efficiency with respect to both memory size and FLOPs (Floating Point Operations Per Second).

Correlation between the Surface Undulation and Luminescence Characteristics in Semi-Polar 112¯2 InGaN/GaN Multi-Quantum Wells

Surface undulation was formed while growing InGaN/GaN multi-quantum wells on a semi-polar m-plane (1-100) sapphire substrate. Two distinct facets, parallel to 112¯2 and 011¯1, were formed in the embedded multi-quantum wells (MQWs). The structural and luminescence characteristics of the two facets were investigated using transmission electron microscopy equipped with cathodoluminescence. Those well-defined quantum wells, parallel and slanted to the growth plane, showed distinct differences in indium incorporation from both the X-ray yield and the contrast difference in annular darkfield images. Quantitative measurements of concentration in 011¯1 MQWs show an approximately 4 at% higher indium incorporation compared to the corresponding 112¯2 when the MQWs were formed under the same growth condition.

Low-Loss Pogo Pin Probe Card with a Coupling Isolation Structure up to 50 GHz

A design for a millimeter wave RF probe card that removes resonance is proposed. The designed probe card optimizes the position of the ground surface and the signal pogo pins to resolve the resonance and signal loss issues that occur when connecting a dielectric socket and a PCB. At millimeter wave frequencies, the height of the dielectric socket and pogo pin matches the length of half a wavelength, allowing the socket to act as a resonator. When the leakage signal from the PCB line is coupled to the 2.9 mm high socket with pogo pins, resonance at a frequency of 28 GHz is generated. The probe card uses the ground plane as a shielding structure to minimize this resonance and radiation loss. The importance of the signal pin location is verified via measurements in order to address the discontinuity caused by field polarity switching. A probe card fabricated using the proposed technique exhibits an insertion loss performance of -8 dB up to 50 GHz and eliminates resonance. A signal with an insertion loss of -3.1 dB can be transmitted to a system-on-chip in a practical chip test.

Ultrasmooth Organic Films Via Efficient Aggregation Suppression by a Low-Vacuum Physical Vapor Deposition

Organic thin films with smooth surfaces are mandated for high-performance organic electronic devices. Abrupt nucleation and aggregation during film formation are two main factors that forbid smooth surfaces. Here, we report a simple fast cooling (FC) adapted physical vapor deposition (FCPVD) method to produce ultrasmooth organic thin films through effectively suppressing the aggregation of adsorbed molecules. We have found that thermal energy control is essential for the spread of molecules on a substrate by diffusion and it prohibits the unwanted nucleation of adsorbed molecules. FCPVD is employed for cooling the horizontal tube-type organic vapor deposition setup to effectively remove thermal energy applied to adsorbed molecules on a substrate. The organic thin films prepared using the FCPVD method have remarkably ultrasmooth surfaces with less than 0.4 nm root mean square (RMS) roughness on various substrates, even in a low vacuum, which is highly comparable to the ones prepared using conventional high-vacuum deposition methods. Our results provide a deeper understanding of the role of thermal energy employed to substrates during organic film growth using the PVD process and pave the way for cost-effective and high-performance organic devices.

Next-Generation Imaging Techniques: Functional and Miniaturized Optical Lenses Based on Metamaterials and Metasurfaces

A variety of applications using miniaturized optical lenses can be found among rapidly evolving technologies. From smartphones and cameras in our daily life to augmented and virtual reality glasses for the recent trends of the untact era, miniaturization of optical lenses permits the development of many types of compact devices. Here, we highlight the importance of ultrasmall and ultrathin lens technologies based on metamaterials and metasurfaces. Focusing on hyperlenses and metalenses that can replace or be combined with the existing conventional lenses, we review the state-of-art of research trends and discuss their limitations. We also cover applications that use miniaturized imaging devices. The miniaturized imaging devices are expected to be an essential foundation for next-generation imaging techniques.