김미현 Mi-hyun Kim , 나범철 Beom-cheol Na , 김기태 Ki-tae Kim , 전상현 Sang-hyun Jeon

DOI: JANT Vol.28(No.6) 767-774, 2024

---

Recent advancements in drone and aerial weapon systems have emphasized the importance of precise navigation systems. Such weapons require sophisticated navigation technology to accurately reach targets. However, accumulated errors in the Inertial Measurement Unit (IMU), a critical sensor in precise navigation systems, significantly impact targeting accuracy. To address this issue, this study introduces a novel Unscented Extended Kalman Filter (UEKF) algorithm, which combines the prediction step of the Unscented Kalman Filter (UKF) with the update step of the Extended Kalman Filter (EKF). Our experiments, conducted across various environments including EKF, UKF, and a no-filter baseline, demonstrate that the UEKF algorithm improves accuracy by up to 62% and achieves up to 21% faster real-time processing performance compared to traditional filter algorithms. This performance enhancement is expected to greatly advance the efficacy of various navigation weapon systems and lay the foundation for developing next-generation precision navigation devices.

손민혁 Minhyuk Son , 박건환 Geonhwan Park , 박소라 So-ra Park , 정환호 Hwanho Jeong , 이병석 Byungseok Lee , 김정자 Jungja Kim , 송영재 Youngjae Song

DOI: JANT Vol.28(No.6) 775-783, 2024

---

KASS (Korea augmentation satellite system) is a satellite navigation-based calibration system that complies with international standards for aviation users. As aviation users who utilize it should continuously ensure accuracy, integrity, availability, and continuity, the KASS service provider should continuously verify and ensure their quality. The PDAT(performance degradation monitoring tool) supports system state and service performance monitoring, operation management, and training/testing, while the PDAT verifies the suitability of KASS signals and messages in quasi-real time, calls the user performance tool using the reference station and user receiver, and compares the support information for issuing the air notification. In this study, we introduced the configuration of the PDAT and KASS message, and derived the method and results of verifying the detailed functions of the tool.

진성근 Seong-keun Jin , 정한균 Han-gyun Jung , 김주영 Ju-young Kim , 곽재민 Jae-min Kwak

DOI: JANT Vol.28(No.6) 784-789, 2024

---

In the recent automotive market, autonomous driving technology has been rapidly advancing. Beyond autonomous passenger vehicles, the development of autonomous driving for commercial vehicles and related autonomous transportation technologies is also progressing actively. Commercial vehicles exhibit unique dynamic characteristics compared to passenger vehicles due to various factors such as the type of cargo, packaging units, and weight. Therefore, road infrastructure support is essential to ensure safe autonomous operation. This paper proposes a safety evaluation methodology for the V2X communication system implemented in the “Saemangeum Autonomous Transportation Commercial Vehicle Demonstration Infrastructure”, established to support autonomous transportation with commercial vehicles. The results derived from the field tests were visualized, confirming the ability to quickly identify and address the root causes of failures. Additionally, delay measurements based on time synchronization demonstrated a precision level of ±0.1 ms even in asynchronous environments, validating the reliability and applicability of the proposed evaluation method.

김유진 Eugene Kim , 정다님 Danim Jung , 김의호 Euiho Kim , 이진실 Jinsil Lee , 고요한 Yohan Ko , 김수경 Su-kyung Kim

DOI: JANT Vol.28(No.6) 790-802, 2024

---

To ensure the integrity of satellite navigation systems, various satellite signal monitoring mechanisms are required, among which Signal Quality Monitoring (SQM) is responsible for detecting Evil Waveforms (EWF) caused by satellite hardware faults. Existing research on SQM techniques and methodologies for defining the Threat Space (TS) of EWF has primarily been developed with augmentation systems in mind. However, the next-generation Regional Navigation Satellite System (N-RNSS) will work without augmentation systems. This means that new TS definitions and SQM techniques must be created that are specifically for N-RNSS signals. This paper summarizes prior research on signal distortion models and TS definition methodologies, and proposes development plans for SQM techniques suitable for N-RNSS signals based on the assumptions and conditions of N-RNSS monitoring stations.

왕수현 Soo-hyun Wang

DOI: JANT Vol.28(No.6) 803-808, 2024

---

Maritime vessel detection is critical in various fields such as maritime safety, security, and environmental protection. However, vessel detection becomes increasingly challenging in low-visibility conditions, such as nighttime or adverse weather. To address this issue, infrared (IR) cameras are employed. Nevertheless, IR cameras face limitations due to low resolution and limited color information, making the detection of small vessels particularly difficult. This study proposes a multi-scale attention (MSA) technique based on convolutional neural network (CNN) feature maps to accurately detect vessels of various sizes. By leveraging the multi-scale feature maps of YOLOv8, MSA improves the detection of small vessels while maintaining real-time detection performance. Experimental results demonstrate that the proposed model enhances the detection precision for small and medium-sized vessels in IR-based detection, thereby maximizing real-time detection performance in maritime environments.

김수현 Soohyun Kim , 최지효 Jihyo Choi , 임상호 Sangho Lim , 손인혜 Inhye Son , 신동조 Dongcho Shin

DOI: JANT Vol.28(No.6) 809-818, 2024

---

Autofocus techniques improve image quality by compensating for phase errors in SAR (synthetic aperture radar) imagery, with PGA (phase gradient autofocus) being the most widely used. Phase errors that occur during SAR imagery formation process degrade quality of SAR imagery. However, conventional PGA techniques may not estimate the phase error properly in the presence of adjacent strong signals in the azimuth direction, such as man-made structures, and may actually degrade the image quality. In this paper, we propose a method to estimate and compensate for the phase error by selecting analytically meaningful signals during target selection for phase error estimation. The performance of the improved PGA method is quantitatively compared and validated based on the conventional PGA method using real airborne SAR data, which results in decreasing number of iteration from 5 to 3, decreasing residual error from 1658.3 mrad. to 45.79 mrad., 87% improvement in -3 dB resolution and 19.53 dB and 18.61 dB improvement in ISLR, PSLR, respectively.

이승준 Seungjun Lee , 김태호 Taeho Kim , 김웅이 Woongyi Kim , 장호상 Hosang Chang , 구성관 Sungkwan Ku

DOI: JANT Vol.28(No.6) 819-826, 2024

---

Due to the concentration of the population in metropolitan areas, congestion in ground transportation networks in urban areas is worsening. Urban air mobility(UAM), a three-dimensional transportation method, is emerging as an alternative and is being presented as a solution. Vertiports are one of the most important infrastructures for operating UAM, and in order for UAM aircraft to take off and land safely, analysis of the environment surrounding the vertiport, such as obstacles and weather, is essential. In this study, we analyzed the criteria for environmental analysis related to obstacles and weather around vertiports, and conducted an empirical analysis targeting candidate sites for vertiports in the Jeju island based on these criteria. We expect that this study will be serve as a reference guideline for reviewing the selection of candidate sites for vertiports in the future.

박상도 Sang-do Park , 김진영 Jin-young Kim , 김영륜 Young-ryun Kim , 정동원 Dongwon Jung , 전향식 Hyang-sig Jun

DOI: JANT Vol.28(No.6) 827-836, 2024

---

According to the K-UAM Concept of Operations, K-UAM navigation is based on PBN. PBN is further divided into RNAV and RNP, which are distinguished by the presence or absence of onboard performance monitoring capabilities. Given that urban air mobility operates along predefined corridors within complex urban environments, RNP-based navigation performance monitoring is essential to ensure safe operations. In this study, integrity monitoring algorithms deemed suitable were applied based on the environmental factors considered at each operational phase. A simulator was developed using realistic data, and the results confirmed that the proposed navigation requirements were met during the cruising, approach, and landing phases.

정성훈 Seong-hoon Jeong

DOI: JANT Vol.28(No.6) 837-843, 2024

---

Marine satellite data exhibits complex spatiotemporal characteristics, necessitating the development of standardized metadata for systematic management and interoperability. This study proposes a framework for designing and generating metadata in JSON, XML, RDF, and CSV formats by integrating international standards (ISO 19115, ISO/IEC 11179) and domestic standards. The research process includes defining metadata elements, schema design, generation, and validation, utilizing tools such as JSONLint, XML Editor Online, EasyRDF, and Online CSV Editor to ensure data quality and interoperability. Additionally, a standard integration logic and a format conversion engine were developed. Through case studies, the study provides a detailed methodology for metadata creation and validation, confirming improved data usability and seamless data exchange between systems. The results demonstrate that metadata standardization enhances data consistency, reliability, accessibility, and improves the efficiency of data-driven decision-making. Future research should focus on incorporating new data formats and advancing additional standardization efforts to further strengthen the scalability and consistency of metadata management.

박정환 Jung-hwan Park , 류재영 Jae-young Ryu , 김태영 Tae Young Kim , 이석환 Seokhwan Lee , 박민균 Minkyun Park , 이학태 Hak-tae Lee

DOI: JANT Vol.28(No.6) 844-854, 2024

---

Generally, commercial transport aircraft are quipped with automatic dependent surveillance - broadcast (ADS-B) out that broadcast its position and other state information at a regular interval via transponder. This system can supplement the ground radar surveillance system and also can be used to mitigate the airborne collision risk. In this paper, the process of formally registering an ADS-B out/transponder integrated equipment for a small drone is outlined. In addition, the reception range are measured and conflict detection and resolution test are performed with a small drone equipped with the ADS-B out and ground-based ADS-B receiver stations. The study experimentally confirmed that it is possible to control a small drone similar to the radar separation between commercial aircraft traffic.

곽민우 Min-woo Kwak , 배근학 Geun-hak Bae , 천지웅 Ji-woong Chun , 이재호 Jae-ho Lee , 김호 Ho Kim , 설창원 Chang-won Shul

DOI: JANT Vol.28(No.6) 855-862, 2024

---

This study addresses the procedures and results of a full-scale static structural test conducted on an octocopter-type cargo air vehicle (CAV) with a payload of 100 kg. According to FAR Part 27 standards, the two type of flight loads were generated by load analysis. And the test loads and test fixture were designed so that the load was applied to the airframe as distributed loads. Design limit load test and design ultimate load test were conducted under two types of loading conditions. Strain, displacement, and load data measured using test equipment were analyzed, confirming that the maximum deviation between the test results and predictions for major structural components was less than 9%. In addition, it was confirmed that there was no permanent deformation or damage to the airframe through, post-test structural inspection and non-destructive testing, thereby ensuring the structural integrity and safety of the aircraft before its first flight.

김민주 Min-joo Kim , 윤성호 Seong-ho Yoon , 전종진 Jong-jin Jeon

DOI: JANT Vol.28(No.6) 863-871, 2024

---

UAM is attracting attention as a next-generation transportation method to solve ground congestion in urban areas, and various routes are expected to be developed to connect vertiports in accordance with future commercialization plans. However, unlike existing aviation systems, due to the characteristic that a large number of aircraft move within urban areas at relatively low altitudes, there is a high possibility that a UAM accident will lead to damage to life and property in the city. Therefore, this study proposes a risk-based UAM corridor setting method. To this end, the risk was analyzed based on the types of damage that could occur when a UAM aircraft crashes, and an AHP survey was used to determine the weights among the factors considered during the analysis. In addition, since UAM is a transportation system that utilizes airspace, and it was judged that the results of the risk analysis can be used as information necessary for various stakeholders to use UAM corridors, a web-based 3D visualization method utilizing spatial information was proposed.

허창환 Chang-hwan Heo , 강광천 Kwang-chun Kang , 윤형근 Heung-keun Yoon

DOI: JANT Vol.28(No.6) 872-878, 2024

---

Urban air mobility (UAM) has emerged as a key solution to address urban traffic congestion and enable environmentally friendly flight at low altitudes. Operating in densely populated urban environments, UAM requires precise airspace management and surveillance systems due to its unique operational characteristics. In particular, UAM airspace involves simultaneous operations of multiple vehicles, necessitating an integrated system capable of managing both cooperative and non-cooperative aircraft. This study analyzes the surveillance environment for cooperative and non-cooperative aircraft within UAM operational airspace and derives the requirements for UAM surveillance systems. Specifically, it proposes a system design that incorporates multi-sensor data fusion and real-time response capabilities for abnormal flight scenarios, supporting seamless airspace surveillance in mediumto high-density operational environments. These findings aim to contribute to the development of safe and efficient urban airspace management systems as the UAM market matures in the future.

김준환 Jun-hwan Kim , 강지수 Ji-sue Kang , 임재진 Jae-jin Lim

DOI: JANT Vol.28(No.6) 879-885, 2024

---

Existing aviation safety management has been difficult to conduct comprehensive analysis because individual stakeholders collect and analyze aviation safety data on their own. To implement the ecosystem of aviation safety data, the establishment of related infrastructure is also important, but above all, it is prerequisite to settle foundations and processes for protecting aviation safety data. In this regard, this study aims to designate aviation safety data security levels based on legal grounds and establish measures and procedures to sharing them. As a result of the study, we found that aviation safety data can be classified into three grades in the current legal system, and it was confirmed that a method of separating access rights according to each security level is necessary. This study can provide implications in that it has laid a detailed foundation for protecting aviation safety data.

안주현 Joohyun Ahn , 채성희 Seong Hee Che , 김현덕 Hyeon Deok Kim

DOI: JANT Vol.28(No.6) 886-891, 2024

---

Recent aviation safety oversight has been shifting towards a risk-based approach, which adjusts inspection intervals according to the risk level of entities being monitored, focusing resources on vulnerable areas to maintain safety within acceptable levels and make efficient use of oversight resources. This study identifies and analyzes key elements used in safety assessments by international organizations and domestic and foreign airlines. It defines essential elements needed for the future improvement of the risk-based safety oversight system in Korea and suggests directions for further development.

윤성진 Sung-jin Yoon , 이찬희 Chan Hee Lee , 배휘종 Hui-jong Bae , 김복기 Bok-ki Kim , 이남식 Nam-sik Lee

DOI: JANT Vol.28(No.6) 892-899, 2024

---

Telemetry systems transmit real-time measurements from unmanned aerial vehicles (UAVs) to remote receivers. Given the sensitive nature of this data, encryption is essential to prevent unauthorized access. While AES-256 block cipher is commonly used, vulnerabilities can arise in key generation and management. To address this, integrating a Quantum Random Number Generator (QRNG) can enhance security by providing stable randomness and Non-determinism characteristic for cryptographic keys and Ciphertext. This approach balances security improvements with system performance and has been validated within standard telemetry unit frameworks.

김영은 Young-eun Kim , 유필중 Pil-joong Yoo , 김영재 Young-jae Kim , 남태현 Tae-hyeon Nam , 고상훈 Sang-hoon Koh

DOI: JANT Vol.28(No.6) 900-906, 2024

---

This paper presents an FPGA-based pipeline and double-RAM architecture to prevent data loss in missile assembly test set. This paper presents the design of an FPGA-based pipeline and double-RAM architecture to address data loss issues in missile assembly test set. To handle data efficiently and prevent loss during communication between the missile assembly test set, the proposed structure utilizes double-RAM, enabling simultaneous read and write operations to minimize latency. The system's reliability was validated through simulations, where a 20% increase in data volume was introduced, and continuous data was received at an 800 Hz rate. The results demonstrate significant improvements in data processing performance, confirming the system's ability to handle increased data loads without loss.

박종혁 Jong Hyeok Park

DOI: JANT Vol.28(No.6) 907-912, 2024

---

This paper discusses the development of a transmitter that provides an RF output exceeding 10 W with PCM/FM modulation in the S-band. The transmission frequency can be adjusted within 2,200 MHz to 2,400 MHz in 1.5 MHz increments, and a variable amplifier allows the modulated output to be controlled between 1 W and 10 W to meet user requirements. An EMI filter is applied at the power input stage to minimize electromagnetic interference, and an LPF is placed at the RF output stage to suppress unwanted signals and harmonics to below -60 dBc. A drop-in type isolator is installed to protect the transmitter from reflected power caused by shorted RF ports, though it slightly increases the size. Additionally, efficient thermal dissipation through the housing bottom enhances stability for heat-generating components like the FPGA and HPA. The developed transmitter is a critical component for data transmission systems, primarily used in guided weapons and aerospace fields, offering high performance, reliability, miniaturization, and low-power operation.

한형석 Hyung-seok Han

DOI: JANT Vol.28(No.6) 913-918, 2024

---

In this paper, we address a new stable condition for the simultaneous presence of two uncertainties in a linear discrete time-varying system with time-varying delay time. The stability conditions for the interval-varying systems, where the system matrices are given in the form of interval matrices, and the unstructured uncertainty of the system matrices of the delay state variables coexist, have been recently studied. As a result, the stability bounds for two kinds of uncertainties were derived as an analytical formula. In this study, we propose new conditions that can maintain the advantages of containing the stability conditions of a linear discrete system with a variety of stability conditions and stability guarantees, and improve their performance in terms of stability conditions and stability ranges. The proposed new stability bounds are presented for each of the two types of uncertainties and compared with previous results through numerical examples, which verifies the effectiveness and excellence.

박유광 Yu-kwang Park , 길현철 Hyun-cheol Gil

DOI: JANT Vol.28(No.6) 919-924, 2024

---

The equipment that constitutes a telemetry system must be synchronized to the absolute time generated by the GPS receiver in order to trigger events at precise times. The more synchronized the system, the better the consistency of the data can be ensured, making time synchronization increasingly important. In this paper, we propose a method for generating an NCO-based precise synchronized clock using the IRIG-B standard time code generated by a GPS receiver in a telemetry system. To validate the proposed method, we designed and implemented it using VHDL logic and conducted simulations with Modelsim. Additionally, we verified the generation of synchronized clocks by receiving IRIG-B time codes from a GPS receiver using FPGA. The proposed method will be beneficial in terms of cost, as it generates clocks synchronized to the GPS absolute time even when using low-cost oscillators with low precision in each device of the telemetry system.

강영흥 Young-heung Kang

DOI: JANT Vol.28(No.6) 925-931, 2024

---

Uumanned aerial vehicle(UAV) cellular wireless networks are emerging as an alternative to solve the problems of natural disasters and sudden traffic demands, However, issues of capacity, coverage, and energy efficiency are considered of paramount importance. To address these issues, a new and promising power domain non-orthogonal multiple access(NOMA) technology with receiver successive interference cancellation(SIC) in intra- and inter-cell interference environment for UAV wireless networks can improve spectral and energy efficiency by utilizing the additional degree of freedom of UAV mobility compared to orthogonal multiple access(OMA). In this paper, the gain improvement effects through NOMA user throughput and outage probability performance have been discussed in UAV cellular downlink and uplink considering the interference envitronment.

노재성 Jae-sung Roh

DOI: JANT Vol.28(No.6) 932-938, 2024

---

The Internet of Things is a network technology in which various objects are connected based on a network to process information. In order to connect objects to the Internet, the characteristics and environment of the communication channel must be considered. Recently proposed LPWA (low power wide area) network technology has the advantage of supporting low data transmission rate and wide communication range while operating at very low power, so it can be applied to various applications of the Internet of Things. In this paper, the performance of the LoRa system, one of the representative LPWA network technologies, was analyzed. In particular, the error rate performance of the LPWA LoRa system according to the main parameters in the Rician fading channel was evaluated. As a result of analyzing the performance of the LPWA LoRa system, it was found that the error rate performance of the LoRa system applying the CSS(chirp spread spectrum) modulation method improved as the signal-to-noise power ratio of the received signal, the strength parameter of the Rician fading, and the spreading factor increased in the Rician fading channel including AWGN(additive white gaussian noise) and Rayleigh fading.

장시영 Siyeong Jang , 윤창석 Changseok Yoon , 정재필 Jae-pil Chung , 남성식 Sung Sik Nam

DOI: JANT Vol.28(No.6) 939-945, 2024

---

This paper explores the implementation of a DCO-OFDM-based underwater wireless optical communication (UWOC) video transmission testbed using a software-defined radio (SDR) platform. Motivated by the increasing need for high-bandwidth underwater communication for applications like marine resource exploration and environmental monitoring, UWOC offers a promising alternative to traditional acoustic and RF-based systems, which suffer from low bandwidth and severe signal attenuation, respectively. The study utilizes DCO-OFDM modulation to overcome signal distortion and achieve efficient data transmission in underwater environments. Through experimental testing, the system demonstrated real-time transmission and playback of Full HD video at 30 FPS with minimal delay. The results underline the potential of DCO-OFDM for enabling high-bandwidth underwater communication.

정우진 Woo-jin Jung , 한조원 Jo-won Han , 최원혁 Won-hyuck Choi

DOI: JANT Vol.28(No.6) 946-952, 2024

---

Smart crosswalks are representative technologies that reduce the risk of pedestrian accidents by utilizing IoT sensors and real-time signal control technology. With the recent development of deep learning object detecting algorithms, the environment to build a real-time pedestrian detection system has improved. Yolo algorithm shows high recognition rate and high recognition speed in real-time object detection through many performance improvements. Among them, Yolov8 algorithm has been upgraded a lot in performance compared to the existing model by adopting C2f module, PAN module and anchor-free method. Therefore, in this paper, a real pedestrian data set is constructed to analyze the performance of each version of Yolo algorithm and select the algorithm version suitable for the pedestrian detection system through actual training. After constructing an on-device based pedestrian detection system by transplanting the trained algorithm into Jetson Nano board, it checks whether pedestrians are detected normally through actual simulation.

정우진 Woo-jin Jung , 김예찬 Ye-chan Kim , 최원혁 Won-hyuck Choi

DOI: JANT Vol.28(No.6) 953-958, 2024

---

An autonomous vehicle is a vehicle that can operate on its own without the manipulation of a driver or passenger, and the standard of SAE (Society of Automatic Engineers) is used internationally for the level of autonomous driving technology. Currently, the system and development of level 4 autonomous vehicles are underway, centering on foreign it companies such as Google, Nvidia, and Tesla. For self-driving cars at level-3 or higher, the autonomous driving system must play the role of the driver, so it is necessary to advance the judgment system to implement high-level autonomous driving. In order to construct a deep learning-based autonomous driving system, sufficient scenarios and datasets are required. This study proposes an image captioning algorithm that creates a scenario by judging the situation of the passenger in the vehicle through the CNN-LSTM model, and analyzes whether appropriate data are generated for learning the monitoring system through simulation.

한조원 Jo-won Han , 정우진 Woo-jin Jung , 최원혁 Won-hyuk Choi

DOI: JANT Vol.28(No.6) 959-965, 2024

---

The purpose of this study is to propose a driver-based pedestrian safety algorithm that helps drivers easily recognize pedestrian signal status, crosswalks, and sidewalk areas, thereby enhancing pedestrian safety. The study uses a monocular camera and one of the object detection algorithms, image segmentation. To detect pedestrians, crosswalks, and vehicles from the driver's perspective, it is necessary to have the ability to accurately detect objects and boundaries through segmentation. Prior to the experiments, the structures of the Mask R-CNN and YOLOv8 models, which provide object detection and segmentation, were compared. Mask R-CNN adds a Segmentation Mask Branch structure after RoIAlign, enabling pixel-level mask prediction for each object. The YOLOv8-seg model utilizes the principles of the YOLACT algorithm, adding a Segmentation Branch structure to perform instance segmentation. The paper uses the YOLOv8-seg model and analyzes the performance and training of various models on the COCO-seg dataset. After conducting experiments, the real-time performance was verified through live streaming from within the vehicle.

이혜민 Hye-min Lee , 최현범 Hyeon-beom Choi , 이영우 Youngwoo Lee , 서정욱 Jeongwook Seo

DOI: JANT Vol.28(No.6) 966-972, 2024

---

Expression intuitively can convey emotion and detect users' emotional changes in real time within the metaverse. However, rapid changes in expressions can complicate accurate emotion identification due to mixed feelings. Redundant collection of the same emotion may lead to unnecessary data transmission, consuming excessive computing resources. This paper proposes an AIoT emotional mapping system using statistical filtering in the metaverse. By filtering out undesired emotions, the filtered emotions are mapped to metaverse blocks via an IoT server based on the oneM2M standard. An experiment with a 9-second happiness video showed that filtering reduced the data to 11 instances of anger and 55 instances of happiness, effectively optimizing resource usage.

정우진 Woo-jin Jung , 한조원 Jo-won Han , 안재홍 Jae-hong Ahn , 최원혁 Won-hyuck Choi

DOI: JANT Vol.28(No.6) 973-978, 2024

---

Urban air mobility (UAM) is attracting attention as a next-generation transportation method as a new solution due to the rapidly increasing population concentration and traffic congestion in urban areas. However, the introduction of such UAM requires a new traffic management system different from the existing traffic management system (ATM), and as one of the types of surveillance systems, this paper proposes a secondary surveillance system of a flying object in a monocular camera-based UAM corridor using a camera sensor as a technology to assist the function of this surveillance system. Therefore, the real-time object detection method uses YOLOv8 artificial intelligence computational technology, which is superior for real-time detection using higher FPS performance, advanced backbone network, and improved pyramid structure compared to previous models. As a result of conducting experiments based on algorithms learned in the actual external environment, it was confirmed that objects were detected normally.

박병주 Byeong-ju Park , 김요셉 Yo-seap Kim , 문일영 Il-young Moon

DOI: JANT Vol.28(No.6) 979-985, 2024

---

This study introduces a method to enhance object recognition performance in autonomous driving systems by combining reinforcement learning with CycleGAN-based data augmentation. To address challenges posed by adverse weather conditions such as rain, fog, and snow, we generated diverse datasets using CycleGAN and trained YOLOv8, a state-of-the-art object detection model. This approach achieved higher recognition accuracy compared to models trained on standard datasets. Additionally, reinforcement learning was utilized to dynamically optimize critical hyperparameters, particularly the learning rate, during the training process. Experimental results demonstrate that this method improves not only the training efficiency but also the robustness of the object detection model under varying environmental conditions. The findings highlight the potential of integrating data augmentation and reinforcement learning to overcome environmental challenges in autonomous driving, contributing to the research of more reliable and adaptable autonomous systems.

김상윤 Sangyun Kim , 안현석 Hyun-seok Ahn , 윤창석 Changseok Yoon , 정재필 Jae-pil Chung , 남성식 Sung Sik Nam

DOI: JANT Vol.28(No.6) 986-994, 2024

---

To ensure safety management, various types of environmental sensors are installed in the interior and critical equipment of aircraft and ships, and these sensors are used to detect abnormal situations. Among the sensor data collected in real time, data generated during abnormal situations must be utilized as essential information for situation detection, storage, and analysis. However, outlier data can reduce the accuracy of abnormal situation data and should therefore be carefully evaluated. This paper proposes an algorithm that combines the Interquartile Range (IQR) statistical analysis method with inter-sensor correlations to determine whether real-time sensor data are generated from an actual abnormal situation or are outliers caused by sensor faults or errors. The proposed algorithm was implemented in a real-time sensor data collection and processing system and tested in a fire test environment. Experimental results demonstrated that the algorithm effectively detects abnormal situation data that were missed by conventional IQR-based outlier detection methods.

Subeen Moon , Hyukjun Oh

DOI: JANT Vol.28(No.6) 995-998, 2024

---

In the realm of national defense and electronic warfare, real-time monitoring and information sharing are crucial for promptly detecting and responding to enemy threats. This paper discusses the development of a real-time intelligent radar system utilizing FMCW radar and CFAR algorithms, along with the IMM filter for object tracking. The proposed approach adapts threshold values based on clutter characteristics, significantly improving object detection accuracy. Furthermore, the system integrates high-speed signal processing and remote control capabilities, maximizing radar system efficiency in diverse battlefield environments.

Eugene Bang

DOI: JANT Vol.28(No.6) 999-1002, 2024

---

This paper explores the impact of the antenna group delay variation (AGDV) induced error on the advanced receiver autonomous integrity monitoring (ARAIM). We modeled the AGDV error as a random process sigma term in the protection level computation by overbounding AGDV observables analyzed within the European Dual Frequency Multipath Model for Aviation (DUFMAN) project. It was found from 99.5% availability simulations that the worst impact of AGDV error contribution resulted in about 7% availability loss for the recent Galileo SARPs update, which might not be marginal.