ETRI Journal最新文献

Low-light image enhancement has made significant advancements in recent years. However, enhancing high-contrast images that exhibit both under- and overexposure remains a major challenge. To address this issue, we propose an exposure-correction method called AGCSNet. Two gamma corrections, $�{�\gamma �}_{�1�}$ and $�{�\gamma �}_{�2�}$, were applied separately to correct for underexposure and overexposure, producing two gamma-corrected images. An illumination map was used to differentiate between the underexposed and overexposed regions in the gamma-corrected images. To mitigate the saturation anomalies caused by the gamma corrections, we introduced a new saturation-correction method with a factor $�s$. Moreover, optimal values for $�{�\gamma �}_{�1�}�,��{�\gamma �}_{�2�}$, and $�s$ can be predicted using our factor-estimation deep-learning model. We evaluated our method on eight datasets. In comparison with over 20 prior methods, our method demonstrates competitive performance with and, in some cases, surpasses state-of-the-art methods that are closely aligned with human visual perception.

Ternary content-addressable memory (TCAM) is widely used in the design of high-speed search engines such as network routers and artificial-intelligence-based applications. However, traditional TCAM designs suffer from two major drawbacks. Static random access memory (RAM)-based TCAMs do one operation at a time, causing the search operation to be suspended while the update operation is in progress, rendering them unsuitable for applications with high-frequency updates. Moreover, during the implementation of wider TCAMs, when the match results are transferred from one slice to another, the last look-up table (LUT) in the slice is always set to logic one, which results in resource wastages. This research aims to overcome the problems associated with traditional TCAM design. The proposed work used six-input (RAM64X1S) LUTs in field-programmable gate arrays by allowing both search and update operations to be performed simultaneously during the data update in a particular LUT. To overcome resource wastage, the proposed design used four RAM64X1S blocks instead of RAM64M blocks. Moreover, the proposed TCAM architecture was considerably simpler, comprising LUTs with AND slicing, thus reducing FPGA resources such as slice registers and slice logic. For TCAM sizes of 512 × 36 and 1024 × 144, the slice utilization was reduced by 17% and 29%, respectively, with their speed being increased by 17% and 26%, respectively. Moreover, the lookup rate and the update rate of the designed TCAMs also improved considerably. The proposed architecture employed high-speed single-cycle searches, making it ideal for fast search applications.

In this study, a new tri-band balanced band-pass filter (BPF) design based on a single-stepped-impedance ring resonator (SIRR) with four stepped stubs loaded on the feeding structure was proposed, analyzed, and design equations were derived. Using the SIRR properties in combination with open stepped stubs, multiple transmission zeros (TZs) were generated, and by adjusting the location of the TZs, high selectivity and an ultra-wide differential-mode (DM) stopband bandwidth were achieved. A pair of stepped stubs on the symmetry line of the filter was also integrated to enhance the common-mode (CM) rejection. Compared to other state-of-the-art differential multiband BPFs, superior performance in harmonic suppression and CM rejection was achieved while maintaining high selectivity and a simple planar design. A prototype tri-band balanced BPF with center frequencies of 1.7, 2.7, and 4.5 GHz was designed and fabricated. Results showed an ultra-wide DM upper stopband bandwidth of 11.1$�f_1$ (up to 18.9 GHz) and a wideband CM suppression of $�S_{�\mathrm{CC}�21�}$< −20 dB from 1.3 GHz to 5.9 GHz.

We investigate the electro-optical properties of indium tin oxide (ITO) transparent electrodes on polyimide (PI) and glass substrates and their impact on organic light-emitting diode (OLED) performance. At a given thickness, the ITO layer exhibits a lower sheet resistance on glass substrates compared to PI substrates. The optical transmission spectra of ITO films are also analyzed, revealing substantial variations owing to interference phenomena. The optimal ITO layer and electron transporting layer (ETL) thicknesses for maximizing the luminance are identified, with glass substrates achieving higher luminance. The luminance on PI substrates shows smaller changes owing to their refractive indices being similar to those of the organic layers. Device fabrication confirms the simulation results, showing that luminance on PI substrates is more sensitive to ETL thickness. Incorporating SiN_x and Al₂O₃ as thin-film encapsulation shifts the optimal ITO layer thickness and slightly reduces luminance. Replacing SiN_x with SiO₂ optimizes luminance, yielding better outcomes. These results emphasize the importance of optimizing encapsulation materials and structures to enhance OLED performance.

In recent decades, the rapid growth of the Internet of Things (IoT) has highlighted several network security problems. In this study, an efficient intrusion detection (ID) system is implemented by using both machine learning and data mining concepts for detecting intrusion patterns. During the initial phase, the intrusion data are collected from NSL-KDD and University of New South Wales-Network Based 15 (UNSW-NB15) datasets. The collected intrusion data are then normalized/scaled by employing a standard scaler technique. Next, the informative feature values are selected by employing the proposed optimization algorithm—that is, the Niche-Strategy-based Gorilla Troops Optimization (NSGTO) algorithm. Finally, these selected informative feature values are transferred to the Long Short-Term Memory (LSTM) model to classify the types of intrusion attacks on both datasets. In comparison to the existing ID systems, the proposed ID system based on the NSGTO-LSTM model obtains a classification accuracy of 99.98% and 99.90% on both datasets.

The information interaction and positioning technology based on commercial Wi-Fi plays a crucial role in expanding the applications for the Internet of Things and smart homes. Achieving high-precision indoor positioning through the optimal deployment of Wi-Fi access points remains a significant challenge. This study proposes two indoor positioning optimization algorithms based on the received signal strength indicator (RSSI) of commercial Wi-Fi signals. The enhanced weighted centroid positioning (EWCP) algorithm introduces a novel weighting method that effectively leverages RSSI ranging errors to improve positioning accuracy and further minimize positioning errors. Meanwhile, the adaptive iterative weighted centroid positioning (AIWCP) algorithm incorporates a detailed weighting approach during the movement of the signal transmitter, providing specific movement indications and enhancing positioning performance. Both simulation and experimental results confirm the effectiveness of these two positioning methods.

In recent years, flying ad hoc networks (FANETs) have been extensively applied owing to rapid advancements in unmanned aerial vehicles (UAVs). The continuous mobility of UAVs, coupled with their operational environment in three-dimensional space, presents significant challenges for routing in FANETs. Additionally, factors such as link stability, link quality, link bitrate, and the ability to forward packets of the next-hop node directly impact routing efficiency. Combining machine learning and location-based routing approaches helps address these challenges. This paper proposes a location-based routing protocol that integrates the Q-learning (QL) reinforcement learning algorithm with dynamic link bitrate adjustment and a penalty mechanism for QL based on retransmission signals from the link layer. Performance evaluations conducted using OMNeT++ demonstrate that the proposed algorithm has markedly enhanced network performance in terms of packet delivery ratio, network throughput, end-to-end delay, link broken count, and packet error rate compared with other well-known routing algorithms.

Recognizing workers' locations and automatically assigning warnings are crucial for preventing industrial injuries. However, existing warning systems are unsuitable for industrial environments as they rely solely on images or insufficiently leverage multi-sensor inputs. Their distance- or plane-based warning assignment strategies are limited when managing 3D spatial environments. To address these issues, we propose DeepMonitor, a novel industrial automatic warning system that incorporates a prior knowledge-based 2D-to-3D conversion and a multi-sensor, space-based warning assignment strategy. We use a mature 2D object detector to avoid the need for 3D training datasets and apply prior knowledge with multi-sensors to reduce the search space for workers' locations. To manage 3D spatial environments, warnings are assigned based on the overlap ratios between workers and zones, defined as 3D bounding boxes. We have constructed a novel dataset for industrial safety and have tested our system against existing approaches. Results demonstrate our system's superiority, achieving an F1-score 16.7% and 24.7% higher than those of the image-only and camera-geometry systems, respectively.

This paper investigates securing nonorthogonal multiple access (NOMA) by leveraging receivers equipped with full-duplex (FD) communication and energy harvesting (EH) capabilities. These receivers decode their intended information while simultaneously jamming eavesdroppers using harvested energy, aiming to achieve high security, energy efficiency, and spectral efficiency. The study analyzes the proposed NOMA scheme with EH-assisted FD receivers across various key performance metrics for a quick performance assessment. The proposed analysis is validated through simulations, which demonstrate the influence of the proposed model on multiple specifications. Furthermore, the proposed model is shown to be considerably more secure than the conventional orthogonal multiple access (OMA) with EH-assisted FD receivers, revealing the advantages of NOMA over OMA.