이화용 Hwa-yong Lee , 이나윤 Na-yoon Lee , 장효석 Hyo-seok Chang

DOI:10.12673/jant.2026.30.1.1 JANT Vol.30(No.1) 1-6, 2026

---

A non-instrument runway is currently operated only under visual flight rules (VFR) due to differences in definitions between the International Civil Aviation Organization (ICAO) and the domestic administrative regulation, Airport Safety Operation Standards. This study addresses high cancellation rates at aerodromes with non-instrument runways by analyzing overseas cases and applying the findings to Ulleung Airport, scheduled to open as a non-instrument runway, thereby proposing measures to reduce cancellations. In particular, with the operation of the Korea Augmentation Satellite System (KASS), approaches similar to precision approaches have become feasible even at non-instrument runways, enabling research on instrument approach procedures utilizing satellite based augmentation system (SBAS). For CAT B aircraft, flight cancellation and delay rates could decrease to approximately 12% when applying the procedure minima, reflecting an improvement of approximately 26.9% compared to VFR operations. The proposed measures are expected to alleviate high cancellation rates, ensure the safety of aircraft operations, and contribute to the revitalization of small airports.

장효석 Hyo-seok Chang , 이승준 Seung-jun Lee , 김도현 Do-hyun Kim

DOI:10.12673/jant.2026.30.1.7 JANT Vol.30(No.1) 7-12, 2026

---

Fuel efficiency critically determines airline profitability and environmental sustainability. In 2024, fuel costs comprised ~32% of operating expenses, highlighting the importance of precise "Cost of Weight" (COW) estimation―the extra fuel penalty from additional weight. While IATA offers standardized guidelines, deterministic models inadequately capture nonlinear interactions in real flight operations. This study applies random forest regression to enhance fuel consumption prediction accuracy by modeling complex factors including aircraft weight, flight distance, altitude, and speed. Results show weight elasticities of 0.669-0.76, where a 1% weight increase raises fuel burn by 0.669-0.76%. Widebody long-haul aircraft exhibit highest sensitivity, while short-haul types show lower elasticities. These findings demonstrate that flight characteristics and operational contexts significantly shape fuel efficiency. Data-driven machine learning approaches thus provide superior tools for optimizing fuel planning and payload management in diverse airline operations.

정세민 Se-min Chung , 이정완 Jeong-wan Lee , 장효석 Hyo-seok Chang

DOI:10.12673/jant.2026.30.1.13 JANT Vol.30(No.1) 13-18, 2026

---

Air traffic control (ATC) communications provide critical real-time information on aircraft status and operational intent. Early detection of abnormal situations can reduce operational risks; however, excessive alert generation may cause alert fatigue and degrade response quality. Therefore, alert designs should consider both importance and reliability. This study proposes a machine learning-based framework for early abnormal situation detection and a confidence-based staged alerting scheme for airline operations control centers(OCC). A total of 165 ATC utterances were collected from LiveATC archives using risk-related keywords and manually transcribed. Text features were extracted using term frequency-inverse document frequency (TF-IDF), and four classifiers were evaluated. The random forest model achieved the best overall performance and was selected as the final classifier. A prototype integrating whisper-based speech-to-text and the classifier was implemented, and a two-level alerting policy (advisory and alert) based on prediction confidence was designed to support practical OCC operations.

김준환 Jun-hwan Kim , 이현철 Hyun Cheol Lee

DOI:10.12673/jant.2026.30.1.19 JANT Vol.30(No.1) 19-27, 2026

---

This study applied the Kano model to identify differences in perceptions of the in-flight servicescape between passengers of full-service carriers (FSCs) and low-cost carriers (LCCs). The results indicate that the two groups showed generally similar evaluation patterns in the ambient and signs/symbols/artifacts dimensions. In the ambient dimension, most attributes were perceived as must-be qualities, except for low noise, which was evaluated as an attractive quality. Similarly, in the signs/symbols/artifacts dimension, in-flight decorations were perceived as indifferent qualities, whereas exit signs and onboard signage were recognized as must-be qualities. However, different patterns emerged in the space/function dimension. While FSC passengers perceived most attributes in this dimension as must-be qualities, LCC passengers tended to evaluate them as attractive qualities. Based on these findings, this study proposes customer satisfaction strategies tailored to each business model and suggests that individual airlines can utilize the results to design differentiated services that reflect their operational characteristics and customer expectations.

김유두 Yu-doo Kim

DOI:10.12673/jant.2026.30.1.28 JANT Vol.30(No.1) 28-34, 2026

---

This study proposes the design of AI-based automated management functions within a digital twin platform to maximize the efficiency of underground facility construction and maintenance. Underground construction often suffers from schedule delays due to reliance on human experience, given the diverse material specifications and complex task dependencies. This paper establishes a real-time management system by integrating bill of quantities (BoQ), work breakdown structure (WBS), and RFID (radio frequency identification) based material flow data. The key designed functions include an inventory shortage notification system based on thresholds, automatic detection of input errors relative to the schedule, and a comprehensive scoring system for maintenance aging. These functions enable proactive risk management and provide an objective, data-driven decision-making environment for facility management.

여준호 Junho Yeo , 이종익 Jong-ig Lee

DOI:10.12673/jant.2026.30.1.35 JANT Vol.30(No.1) 35-40, 2026

---

In this paper, a filtering microstrip patch antenna using multiple slot pairs at 5.8 GHz band was studied. The proposed antenna achieves the filtering characteristic and bandwidth enhancement by incorporating two pairs of slots onto the lower radiating edge of the patch in an inset-fed rectangular microstrip patch antenna. To compare the performance, a conventional inset-fed rectangular microstrip patch antenna was designed using an IS680-320 substrate with dimensions of 40 mm × 40 mm × 0.762 mm. Simulation results show that the frequency band of the conventional inset-fed rectangular microstrip patch antenna with a voltage standing wave ratio less than 2 was 5.747-5.855 GHz(1.86%), whereas that of the proposed filtering microstrip patch antenna using multiple slot pairs was 5.674-5.927 GHz(4.36%). The measured frequency band of the fabricated filtering microstrip patch antenna using multiple slot pairs was 5.685-5.915(3.97%). Although the measured bandwidth is slightly reduced compared to the simulation, it still satisfies the requirements for the 5.8 GHz ISM band.

김요셉 Yo-sep Kim , 박병주 Byeong-ju Park , 유연휘 Yeon-hwi You , 한주현 Ju-hyun Han , 문일영 Il-young Moon

DOI:10.12673/jant.2026.30.1.41 JANT Vol.30(No.1) 41-47, 2026

---

OpenAI’s CLIP model achieves strong cross-modal retrieval performance through large-scale image-text pretraining; however, its high-dimensional embeddings impose substantial memory and computational overhead, limiting scalability in resource-constrained environments. We propose a post-processing-based Joint PCA method that reduces CLIP embedding dimensionality without additional training. By jointly learning principal components from combined image and text embeddings, the method preserves inter-modal alignment during dimensionality reduction. Experiments on the Flickr30k dataset show that CLIP embeddings exhibit low intrinsic dimensionality, with the top 200 components explaining over 90% of the total variance. A 128-dimensional representation reduces memory and computational costs by 75% while maintaining retrieval performance, demonstrating that statistical post-processing can improve retrieval efficiency without retraining.

권순녀 Soon-nyu Kweon , 이성렬 Seong-real Lee

DOI:10.12673/jant.2026.30.1.48 JANT Vol.30(No.1) 48-53, 2026

---

A novel dispersion-map structure for a midway optical phase conjugation-based MSSI system is proposed and numerically evaluated for 960 Gb/s WDM transmission. The proposed scheme adopts a triangular-wave-shaped accumulated dispersion profile in both former and latter half sections, where the maximum accumulated dispersion is gradually varied over successive repetition periods. Distortion compensation performance is assessed for various maximum dispersion combinations and repetition patterns. The results show that the proposed dispersion maps provide improved compensation performance and wider tolerances for launch power and net residual dispersion compared with conventional schemes. It is also observed that the compensation characteristics depend on the repetition direction of the dispersion profile, and that larger maximum accumulated dispersion enhances the overall compensation effectiveness.

한호준 Ho-jun Han , 문일영 Il-young Moon

DOI:10.12673/jant.2026.30.1.54 JANT Vol.30(No.1) 54-60, 2026

---

Defect detection models used in industrial settings struggle to improve their performance due to the scarcity of training data. To address this issue, this paper proposes a novel data augmentation method that combines a ControlNet-based generative model, valid region extraction, and Poisson blending. First, ControlNet is used to generate various defect patterns. Next, a valid region extraction strategy is applied to select optimal locations by analyzing the statistical characteristics of the background where the defects will be located. Finally, Poisson blending is used to create a synthetic image that enhances realism. Experimental results using the KolektorSDD dataset demonstrate that the proposed technique significantly improves the mean IoU of the ResNet34 model from 0.5667 to 0.8919. The proposed data augmentation technique successfully addresses data imbalance and is expected to contribute to improving model performance due to data sparsity.

Tae-yong Park , Koangyong Hyun , Seongwon Oh

DOI:10.12673/jant.2026.30.1.61 JANT Vol.30(No.1) 61-64, 2026

---

As maritime expansion in the Exclusive Economic Zone accelerates, evaluating the impact of offshore wind farms on navigation systems is essential for maritime safety. This study investigates the signal propagation and interference effects of offshore wind farm on the AIS(automatic identification system) using 3D ray-tracing simulations near onshore environment. To address the complex multipath interference arising from turbine structures and sea-surface reflections, a Rayleigh fading model was specifically adopted to ensure a conservative, worst-case performance analysis. The results indicate that the presence of wind farms reduces received power by approximately 3.2 to 15.4 dB. Despite this degradation, the maximum packet error rate was found to be 3.11%, remaining well below the 20% international regulatory limit. Thus, while wind farms cause signal quality loss, they do not significantly impede AIS reliability, although further research for large-scale developments is necessary.