ETRI Journal最新文献

As computing systems become increasingly larger, high-performance computing (HPC) is gaining importance. In particular, as hyperscale artificial intelligence (AI) applications, such as large language models emerge, HPC has become important even in the field of AI. Important operations in hyperscale AI and HPC are mainly linear algebraic operations based on tensors. An AB21 supercomputing AI processor has been proposed to accelerate such applications. This study proposes a XEM accelerator to accelerate linear algebraic operations in an AB21 processor effectively. The XEM accelerator has outer product-based parallel floating-point units that can efficiently process tensor operations. We provide hardware details of the XEM architecture and introduce new instructions for controlling the XEM accelerator. Additionally, hardware characteristic analyses based on chip fabrication and simulator-based functional verification are conducted. In the future, the performance and functionalities of the XEM accelerator will be verified using an AB21 processor.

The photochemical acid generation is refined from the first principles of quantum electrodynamics. First, we briefly review the formulation of the quantum theory of light based on the quantum electrodynamics framework to establish the probability of acid generation at a given spacetime point. The quantum mechanical acid generation is then combined with the deprotection mechanism to obtain a probabilistic description of the deprotection density directly related to feature formation in a photoresist. A statistical analysis of the random deprotection density is presented to reveal the leading characteristics of stochastic feature formation.

LED-to-LED visible light communication (VLC), which uses LEDs not only for the transmitter but also for the receiver, is being studied as an efficient optical wireless communication technology that uses LED lighting infrastructure. In this paper, we investigate microLED-to-LED VLC, which uses microLEDs as both the transmitter and receiver. In particular, we conducted a study to improve the performance of microLED-to-microLED VLC. For this, we measured the performance depending on the transmitter and receiver LED color combination. In addition, the effects of zero bias and reverse bias at the receiver LED were investigated. We also investigated the improvement in the reverse bias when applying a transimpedance amplifier to the receiver LED. Finally, we experimentally demonstrated a data rate of 360 kbps in the microLED-to-microLED VLC.

This paper presents a membrane computation error-minimized mixed-mode spiking neural network (SNN) crossbar array. Our approach involves implementing an embedded dummy switch scheme and a mid-node pre-charge scheme to construct a high-precision current-mode synapse. We effectively suppressed charge sharing between membrane capacitors and the parasitic capacitance of synapses that results in membrane computation error. A 400 × 20 SNN crossbar prototype chip is fabricated via a 28-nm FDSOI CMOS process, and 20 MNIST patterns with their sizes reduced to 20 × 20 pixels are successfully recognized under 411 μW of power consumed. Moreover, the peak-to-peak deviation of the normalized output spike count measured from the 21 fabricated SNN prototype chips is within 16.5% from the ideal value, including sample-wise random variations.

This paper presents the basic quantum reinforcement learning theory and its applications to various engineering problems. With the advances in quantum computing and deep learning technologies, various research works have focused on quantum deep learning and quantum machine learning. In this paper, quantum neural network (QNN)-based reinforcement learning (RL) models are discussed and introduced. Moreover, the pros of the QNN-based RL algorithms and models, such as fast training, high scalability, and efficient learning parameter utilization, are presented along with various research results. In addition, one of the well-known multi-agent extensions of QNN-based RL models, the quantum centralized-critic and multiple-actor network, is also discussed and its applications to multi-agent cooperation and coordination are introduced. Finally, the applications and future research directions are introduced and discussed in terms of federated learning, split learning, autonomous control, and quantum deep learning software testing.

In this paper, a novel dual-band wideband bandpass filter (BPF) based on transversal signal-interaction concepts with a wide upper stopband is proposed and investigated. The designed specification of two passbands can be managed and satisfied based on the independent controllable fractional bandwidth of the two passbands and the centered frequencies. The centered frequencies of dual-band BPF are, respectively, 0.79 GHz (ƒ₁) and 1.24 GHz (ƒ₂) with 3 dB fraction bandwidths of 26.54% and 11.3%. Two transmission paths consisting of coupled stub-loaded square ring resonators and anti-coupled shorted lines are used to realize signal cancellation of multiple transmission path signal transmission from Port 1 to Port 2. Eleven transmission zeros (TZs) modify harmonic suppression up to 10 ƒ₁ with stopband rejection higher than 15 dB. Butterworth lumped notch network and step impedance resonator (SIR) are also utilized to improve the selectivity and harmonic suppression. A compact filter with a circuit size of 0.08λg × 0.08λg is implemented and tested. Good agreement between simulation and measured results verifies the reliability of the designing scheme.

As technology advances, smart homes are being increasingly adopted, thus generating massive data that pose new research challenges. We propose a machine learning framework for monitoring energy consumption in smart home devices. The proposed framework involves an anomaly detection module, followed by a predictive model to forecast energy consumption patterns in a typical smart home. We employ three outlier-based techniques for anomaly detection: (1) local outlier factor, (2) connectivity-based outlier factor, and (3) cluster-based local outlier factor. Furthermore, we apply random forest, linear regression, decision tree, and the ensemble techniques of adaptive, gradient, and extreme gradient boosting to anomaly free data to develop baseline models that predict the energy consumption patterns of smart home devices. The framework is evaluated on three publicly available energy datasets collected from various smart homes. The experimental results reveal that the cluster-based local outlier factor with extreme gradient boosting achieves promising results with high prediction accuracy.

The eye socket is a cavity in the skull that encloses the eyeball and its surrounding muscles. It has unique shapes in individuals. This study proposes a new recognition method that relies on the eye socket shape and region. This method involves the utilization of an inverse histogram fusion image to generate Gabor features from the identified eye socket regions. These Gabor features are subsequently transformed into Gabor images and employed for recognition by utilizing both traditional methods and deep-learning models. Four distinct benchmark datasets (Flickr30, BioID, Masked AT & T, and CK+) were used to evaluate the method's performance. These datasets encompass a range of perspectives, including variations in eye shape, covering, and angles. Experimental results and comparative studies indicate that the proposed method achieved a significantly ($�p�<�0.001$) higher accuracy (average value greater than 92.18%) than that of the relevant identity recognition method and state-of-the-art deep networks (average value less than 78%). We conclude that this improved generalization has significant implications for advancing the methodologies employed for identity recognition.

The power relay assembly (PRA) is an essential component to ensure the safety of an electric vehicle. We propose a semiconductor-based precharge switch to overcome the shortcomings of the conventional mechanical relay, which is currently used as the precharge switch in the PRA. The gate-driving circuit uses a photocoupler instead of a gate-driving integrated circuit to reduce complexity and cost. Five devices, namely, two insulated-gate bipolar transistors (IGBTs), two silicon carbide metal–oxide–semiconductor field-effect transistors (SiC MOSFETs), and one metal–oxide–semiconductor-controlled thyristor (MCT) are tested as candidate precharge switches. Each device is analyzed under varying battery voltage and fixed measurement conditions. The IGBT and SiC MOSFET burn below 600 V, while the MCT exhibits normal operation up to 800 V. The MCT precharge switch can solve shortcomings of the conventional mechanical relay and increase the performance with a simple gate-driving circuit. Furthermore, the PRA with an MCT precharge switch can reduce the volume, weight, and cost while improving reliability.

We propose a multicarrier chaotic system that incorporates dual-index and M-ary modulation techniques. The proposed system improves the energy and spectrum efficiencies with a tradeoff in computational complexity. In addition, it applies index modulation to select the chaotic signal and permuted chaotic signals along with $�M$-ary modulated symbols. As a result, the spectral and energy efficiencies are increased in chaotic systems. Theoretical expressions are derived to calculate the bit error rate of the proposed system under additive white Gaussian noise and Rayleigh fading channels. The equations are validated through Monte Carlo simulations, whose results demonstrate a reduction in bit error rate for the proposed system. The data rate, spectral efficiency, energy efficiency, and computational complexity of the proposed system are analyzed and compared with those of existing multicarrier chaotic systems.