Recent research on embedded systems has become increasingly important due to their central role in high-performance embedded devices, including artificial intelligence, autonomous driving, and energy management technologies. Embedded systems are specialized computer systems designed to perform specific tasks while optimizing performance and minimizing memory usage. RISC-V, an open RISC-based instruction set architecture developed by the University of Berkeley in 2010, is well-suited for these systems. In addition to the base 32-bit integer instruction set, RISC-V supports extensions such as the M-extension for multiplication and division and the C-extension for instruction compression. In this paper, we propose the design of a 32-bit 5-stage pipeline RV32IMC processor aimed at high-performance embedded devices. By incorporating the RV32IMC instruction set, the proposed processor achieves enhanced computational efficiency and reduced code size, making it a strong candidate for energy-efficient, high-performance embedded applications. Furthermore, the design was validated on an Artix-7 field-programmable gate array, demonstrating the processor's feasibility and potential benefits for embedded systems.

With the development of technologies such as the Internet of Things (IoT) and autonomous vehicles, research is being conducted on embedded processors that meet high performance, low power, and memory efficiency. The "C" expansion of the RISC-V processor is required to increase memory efficiency. In this paper, we propose an RV32IC processor and compare the benchmark performance score of the RV32I processor with the code size generated by the GCC compiler. In addition, we propose memory access and combination methods to support 16-bit compression commands, and command extension methods. The proposed RV32IC processor satisfies the maximum operating frequency of 50 MHz on the Artix-7 FPGA. The performance was checked using the benchmark programs of the Dhrystone and Coremark, and the code sizes of the RV32I and RV32IC generated by the GCC compiler were compared. The proposed processor RV32IC decreased DMIPS/MHz by 2.72% and Coremark/MHz by 0.61% compared to RV32I, but Coremark's code size decreased by 14.93%.

In this study, TiN thin films, used as electrode thin films, were deposited through the Thermal Atomic Layer Deposition method using TDMAT (Tetrakis(dimethylamido)titanium) as the precursor and NH3 as the reactant. After deposition at various temperatures, the thickness, crystal structure, surface roughness, impurity content, and resistivity of the films were compared. The results showed that the optimal process temperature was 200℃, forming TiO2 anatase and TiN(200) crystal structures. The surface roughness was the lowest, with Ra=0.393nm and Rq=0.502nm, and the impurity content was 27.5% O and 6.35% C, with the resistivity also recording its minimum value. Particularly, at a long pulsed time of 11 seconds, the impurity inclusion due to self-decomposition reactions was dominant, showing excellent properties at lower temperature.

Unsupervised learning is a type of machine learning, and unlike supervised learning or reinforcement learning, a target value for input value is not given. Clustering is mainly used for such unsupervised learning. One of the representative methods of such clustering is K-means clustering. Since K-means clustering is a method of determining the number of clusters and continuing to find the central point of the data allocated to the cluster, there is a problem that the clustered group may not be the optimal cluster. In this study, particle swarm optimization algorithm, which determines the motion vector by adding various variables as well as the center point, is applied to K-means clustering. The improved K-means clustering makes it possible to move toward better outcome values even when the center of cluster no longer change. In the conventional clustering method, the center of the cluster moves to the center of the data belonging to the cluster, and clustering ends when the cluster does not change, so other characteristics other than the center value are excluded. Unlike the conventional clustering method, the improved clustering method uses a central value, an average value, and a random value as variables, and a particle swarm optimization algorithm that modifies the vector for each iteration is applied. As a result, improved clustering method derived a better result value than the existing clustering method in the group's fitness index, silhouette score.

We investigated the n-type δ-doping activation of the tunneling junctions of Si nanolayers for silicon tandem cell applications. The thin film growth of pn junction with the inclusion of phosphorus monolayer was performed by plasma-enhanced chemical vapor deposition with the implement on 6-inch wafers of p-Si microtextured substrates. The rapid thermal annealing processes with various temperatures were performed to activate the δ-doped layer. The activation was confirmed by the electron spin resonance with Lande factor g=2.006085 for the delocalized conduction electron from the phosphorus δ-doped layer at the magnetic field of 3357.5 Gauss. The tunneling junction shows the Ohmic character at the low voltage and the Schottky character at the high voltage bias.

High efficiency evaporation source is developed to perform a vacuum deposition process in which a deposition material is heated and vaporized to eject from a solid state to a gaseous state. In order to obtain a uniform thin film, conditions such as the structure of the effusion cell, the distance between the effusion cell and the substrate, nozzle size, and evaporation angle must be optimized. In this experiment, organic material Alq3 and metal Al thin film deposition process was performed using the effusion cell and thin film characteristics was analyzed.

The installation of arc circuit breakers is being strengthened to prevent accidents such as electric shock and fire caused by Arc. Among arcs, serial arcs are difficult to detect with general arc detectors because there is not much change in load current when an arc occurs. Therefore, in this paper, unlike the existing Arc Fault Circuit Interrupters method, arc detection hardware is implemented using the FFT algorithm. FFT is suitable for serial arc identification because it can efficiently analyze high-frequency signals generated outside of normal AC signals. This study explains ARC detection circuits and the 2048-FFT based on radix-2 and radix-4, and presents hardware implementation results using FPGA. The implemented system detects the arc up to the frequency range of 122,880 Hz. Through simulation and FPGA board testing, it was confirmed that ARC was detected.

In modern society, the incidence of various joint diseases such as degenerative arthritis is steadily increasing with the increase of the elderly population. For the effective treatment of these diseases, the demand for CPM medical devices that increase the range of motion of the joint is increasing, and if a patient is unable to perform exercise on his or her own, it plays a role in helping joint recovery through repetitive and continuous manual exercise to prevent joint stiffness, promote blood circulation, and improve tissue flexibility to shorten the recovery period after joint surgery to efficiently proceed with the rehabilitation process. Due to the uncertainty of the angle measurement method currently in use, it is difficult to accurately evaluate the patient's exercise amount required to improve the range of motion, which can negatively affect the reliability of CPM medical devices and the patient's willingness to rehabilitate. Therefore, a new technology using encoders and angle sensors is needed to overcome the problems arising from the method measured using only the existing Potentiometer.

Persistent memory technology has been recognized as a next-generation memory solution because of its stability over traditional volatile memory. The primary advantage of persistent memory is its non-volatile nature, allowing data retention even when the power is off. Additionally, persistent memory offers faster read speeds compared to traditional HDDs and SSDs. However, ensuring data consistency through cache flushing commands is increasingly important so that performance and power consumption issue would be challenges. This paper proposes a new cache structure to mitigate the drawbacks of cache flushing by considering the number of flushes per cache line. On top of that, a counter and decision bit to track and manage these actions are added. As a result, this architecture decreases approximately 56% of the additional memory access.

Recently, the number of pet owners has been increasing, and considering that the number of single-person households will increase further due to the low birth rate and aging population, the growth of the pet owner population is expected to continue. Accordingly, the pet-based industry is also expanding its base. In addition, as the time spent at home has increased due to COVID-19, the demand for products that help with pet owner care has increased, further maximizing the growth. Along with the growth of the pet-related industry, the size of the pet clothing and supplies market is also increasing explosively. Meanwhile, in relation to pet styling, services that simulate the expected appearance of a pet wearing a specific product are being developed, but services that assist pet owners in styling that is suitable for the physical and psychological characteristics of their pets are insufficient. Therefore, the purpose of this study is to provide a pet styling service provision device and method that can solve the problem of owners having difficulty judging whether styling is suitable for the physical and psychological characteristics of their pets by using pet fashion simulation.

Thermodynamic solution stability of aqueous copper-nickel-sulfur system was numerically analyzed to produce thin copper-nickel nano-multi-layers by pulse electro-forming. The main program for numerical analysis was written by C# language, which was composed of the data input, numerical calculation, decision and plotting sub-programs. From the thermodynamic data of 32-feasible phases of the Cu-Ni-S-H₂O system, the phase stability diagram of the Cu-Ni-S-H₂O system was constructed. It revealed the electro-forming condition of the copper and the nickel was V_SHE<0.35 for copper deposition, V_SHE<-0.24 for nickel deposition, pH=1.0 and 25℃ in the sulfide bath. The coppernickel multi-layers was well produced by electro-forming in the sulfide bath with two-wave pulse voltages of - 0.2V_SHE, -0.5 mA/cm², and 25 seconds for copper deposition and -1.7 V_SHE, -50 mA/cm² and 80 seconds for nickel deposition, at pH=1.0 and 25℃. From TEM and EDX analysis the Cu-Ni multi-layers with about 5 ㎛ thick had the copper-rich phase of about 20 nm in thick and the nickel rich phase of about 25 nm in thick, respectively.

For potential applications in large-area organic light-emitting diode (OLED) displays, solar cells, sensors, and secondary batteries, we have developed a slot-die coating process that allows control of coating width and continuous coating. In a sheet-to-sheet (S2S) type coating system, coating width adjustment and continuous coating processing are highly demanded to improve productivity and automate the coating process. To control coating width, we have proposed a shim plate with a narrower outlet width and a dual plate slot-die head (DP-SDH) with a meniscus guide. For continuous coating process, we have attempted a lip dipping method and a motorized cleaning module method. Finally, using the DP-SDH and motorized cleaning module method, we have successfully fabricated a poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS) film with a width of 150 mm, despite the head-lip width being 200 mm.

This paper proposes a method of improving cooling efficiency by applying a Peltier Element to a heat pipe of an air-cooled cooling system of a Battery Thermal Management System for high-speed cooling of a vehicle battery cell. In addition, when the temperature sensor detects the heat generation of the battery cell, the Peltier Element and cooler can be operated to quickly reduce the temperature of the cell. For optimal thermal management, we built an ATmega128A-based Battery Thermal Management System and used KiCAD tool to model and design the cooling system structure. Finally, the experiment verified the high efficiency improvement of cooling performance by comparing the difference between cooling efficiency and cooling performance at room temperature over time for vehicles adopting the existing air cooling method.

Conductive charging of electric vehicles has entered a stable stage, and various product lines have been developed and standards have been made. Inductive charging methods can be classified into various other methods such as Indirect Power Transfer, Coupled Magnetic Resonance, and Permanent Magnet Coupled Transfer. Since choosing the most effective charging method for electric vehicles is difficult, standardizing the method of wirelessly transferring power is still a challenge to be achieved in a future. SAE J2954 sets an overall standard of electric vehicle Wireless Power Transport system. This paper proposes a wireless charging module that installs two receiving coils at a certain distance under the vehicle to minimize coil alignment due to incorrect parking and other vehicle full length during static charging. Finally, we developed an ATmega128 MCU-based module prototype and tested it to verify the efficiency of lower misalignment probability in different alignment charging environments compared to conventional single VA coils.

This paper proposes a system that utilizes the Micro Controller Unit to detect and effectively respond to thermal runaway events that may occur during electric vehicle battery charging. Thermal runaway refers to a rapid fire hazard caused by the increase in internal battery temperature, which can be particularly catastrophic in indoor charging environments. The proposed system is equipped with real-time temperature sensors and communication modules to monitor battery temperature changes. When a fire is detected, the system automatically moves the battery to a fire suppression area to extinguish the fire. Finally, a prototype was developed, and the system's functionality was verified through simulations of fire scenarios.

In this study, we report on the joining of mullite ceramics using the reaction-bonded aluminum oxide method without applying any external pressure, in consideration of a possible multilayer ceramic substrate in semiconductor back-end process. For this purpose, Al/Al2O3/SiO2/Mullite powder mixture paste was applied to the joining surfaces between two parent mullite bodies of the same composition, and then sintered at 1,650 ℃ for 2 h in air, resulting in a dense and rigid mullite ceramic joints. Phase and microstructural analysis of the joined mullites showed that the reaction bonding by Al oxidation and thereafter mullite formation were completed during the heat treatment process. However, due to the difference in sintering behavior between the parent body and the joining layer, few pore of which size proportional to the joining layer thickness, were observed at some parts of the joining interface. The formation of the pore and its causes was discussed.

Previous research on solar power generation forecasting has generally relied on meteorological data, leading to lower prediction accuracy. This study, in contrast, uses actual measured power generation data to train various ANN (Artificial Neural Network) models and compares their prediction performance. Additionally, it describes the characteristics and advantages of each ANN model. The paper defines the principles of solar power generation, the characteristics of solar panels, and the model equations, and it also explains the I-V characteristics of solar cells. The results include a comparison between calculated and actual measured power generation, along with an evaluation of the accuracy of power generation predictions using artificial intelligence. The findings confirm that the LSTM (Long Short-Term Memory) model performs better than the MLP (Multi- Layer Perceptron) model in handling time-series data.

This study details the development of YARA rules and a detection program specifically designed to identify malware in HWP documents, a common target in cyber-attacks within South Korea. By thoroughly analyzing the unique structural features of HWP files, we developed precise YARA rules that were subsequently integrated into a custom detection tool. The program was rigorously tested on a dataset of benign and malicious HWP documents, demonstrating high detection accuracy and a low false-positive rate. This research offers a robust and practical solution for enhancing cybersecurity in environments where HWP files are frequently used, contributing valuable tools for the targeted detection of document-based malware.

As smartphones become more integral to daily life, security concerns, particularly regarding smishing, have risen significantly. With the increasing frequency and variety of smishing attacks, current detection methods struggle to provide effective solutions, with most commercial algorithms achieving around a 70% detection rate. This paper proposes a novel approach to enhancing smishing detection accuracy by utilizing Natural Language Processing to analyze message syntax and semantics, combined with URL Punycode conversion and whitelist techniques. The approach focuses on improving detection through comprehensive message analysis, aiming to address the limitations of existing preventive methods. The proposed system offers conceptual improvements in smishing detection strategies, providing a more robust framework for addressing evolving security challenges.

In high-precision laser machining process in semiconductor and display device manufacturing, thermal plume may arise from high heating zone around laser application point and it may significantly affect the surface contamination by micron-sized fume dust particles in laser surface machining. The present numerical study investigates this thermal plume flow by employing three-dimensional large eddy simulations and also fume dust particle dispersion through particle tracing techniques. The numerical results show the laminar to turbulent transition characteristics of buoyant thermal plume and fume dust particles rising and falling on the machining surface. The influences of laser power on surface scattering of fume dust particles are closely examined for the dust particle size ranging from 10 to 90 ㎛.

Products were typically produced using specialized equipment such as CNC machines, milling machines, and lathes in traditional manufacturing. However, modern manufacturing is increasingly attempting with technological advancements to leverage large industrial robots for machining, offering greater flexibility, efficiency, and a high degree of freedom throughout the entire production process. Additionally, the demand for industrial robots continues to rise as industries adopt smart factories. These robots are becoming larger, more precise, and faster, as they take over tasks previously requiring specialized equipment or skilled human operators. Where numerous robots are in operation in factories, ensuring a stable supply chain and maintaining operational uptime is crucial. Therefore, preparing for potential mechanical failures in each robot is necessary, and there is a growing need for technologies that enable real-time fault diagnosis and predictive maintenance. A large industrial robot used for machining was employed as a testbed for fault diagnosis in this study. The Vibration data was collected from various robot axes under both normal operating conditions and abnormal conditions, such as end-effector overloads and drive malfunctions. The collected vibration data was then preprocessed, and key features were analyzed and extracted. The extracted features were used to build a learning model, and in this study, the CNN (Convolutional Neural Network) algorithm was applied instead of k-NN (k-Nearest Neighbors) to diagnose defects occurring in the discontinuous movements of the robot, thereby improving accuracy.

Public datasets, which are freely available and often labeled, play a crucial role in training object detection models in computer vision. While public datasets are effective for developing general object detection models, they may not be ideal for specialized tasks. For specific object detection needs, it is more beneficial to create and use a dataset tailored to the target object. This paper proposes a method for extracting a target-specific dataset from public datasets to develop object detection models with superior performance for the target object. This approach not only improves detection accuracy, but also reduces training data requirements and complexity. We evaluate the performance of the proposed method using the latest object detection model YOLOv10.

This paper explores the recent advancements in speech recognition technology, focusing on the integration of artificial intelligence to improve recognition accuracy in challenging environments, such as noisy or low-quality audio conditions. Traditional speech recognition methods often suffer from performance degradation in noisy settings. However, the application of deep neural networks (DNN) has led to significant improvements, enabling more robust and reliable recognition in various industries, including banking, automotive, healthcare, and manufacturing. A key area of advancement is the use of Silent Speech Interfaces (SSI), which allow communication through non-speech signals, such as visual cues or other auxiliary signals like ultrasound and electromyography, making them particularly useful for individuals with speech impairments. The paper further discusses the development of multi-modal speech recognition, combining both audio and visual inputs, which enhances recognition accuracy in noisy environments. Recent research into lip-reading technology and the use of deep learning architectures, such as CNN and RNN, has significantly improved speech recognition by extracting meaningful features from video signals, even in difficult lighting conditions. Additionally, the paper covers the use of self-supervised learning techniques, like AV-HuBERT, which leverage large-scale, unlabeled audiovisual datasets to improve performance. The future of speech recognition technology is likely to see further integration of AI-driven methods, making it more applicable across diverse industries and for individuals with communication challenges. The conclusion emphasizes the need for further research, especially in languages with complex morphological structures, such as Korean

Electric scooters are used by many people as a means of transportation because they are easy to operate and get around quickly. Although electric scooters can move at high speeds, they lack safety measures such as seat belts and blocked car body, which can lead to serious injury in the event of an accident. If there are two people on board, the braking speed will be slower and the weight of the vehicle will increase its braking distance during sudden stops, leading to even more serious damage. Therefore, this paper proposes the illuminance sensors based multiuser boarding detection scheme for electric scooter, which is expected to decrease the risk of accidents and to lessen the likelihood of injury. From the performance evaluation results, it has been shown that the proposed scheme has higher detection ratio than existing schemes and has the detection accuracy of about 83% by means of the machine learning based foot position estimation for the sensed data.