Doppler observations contain information that reflects the line-of-sight relative velocity between a satellite and a receiver. Owing to this characteristic, they exhibit stable behavior even over short time intervals and are advantageous for precisely estimating velocity. Recently, there has been increasing interest in position and velocity estimation techniques that utilize the large Doppler shifts provided by Low Earth Orbit (LEO) satellites. The feasibility of using Doppler measurements from LEO satellites, however, has mainly been investigated through simulation studies while few experimental evaluations under real-world conditions have been reported. In this study, Doppler observations from GPS, Galileo, and BeiDou satellites were collected using a commercial Global Navigation Satellite System (GNSS) receiver. The accuracy of Doppler-based positioning was assessed using least-squares estimation in both static and kinematic conditions. For the static scenario, observation data from the GAMG reference station were used. The resulting 3D Root Mean Square Error (RMSE) ranged from approximately 50 to 100 m, depending on the number of visible satellites and the observation conditions. For the kinematic scenario, experiments were conducted in an urban area and Real-Time Kinematic (RTK) solutions were used as the ground truth. The results show that the 3D RMSE of Doppler-only positioning increased to approximately 70 to 400 m due to signal blockage and multipath effects commonly present in urban environments. In contrast, velocity estimation maintained a stable 3D RMSE of approximately 0.1 m/s, thus demonstrating far greater robustness and consistency compared to position estimation. Based on the reliable velocity estimates obtainable from Doppler observations, an Extended Kalman Filter (EKF) model integrating Doppler and pseudorange observations was developed and experimentally evaluated in kinematic conditions. The proposed integrated model reduced the 3D RMSE by approximately 99% compared to Doppler-only positioning and by about 58% compared to pseudorange-only positioning. These findings experimentally demonstrate Doppler-based GNSS techniques can enhance positioning performance.

The Quasi-Zenith Satellite System (QZSS) Civil Navigation (CNAV) message provides two almanac types, Midi and Reduced, to aid satellite signal acquisition. An Interface Control Document (ICD) defines operational parameters for both, including the reference time (t_oa) update interval and message transmission intervals. It also specifies three-dimensional position accuracy for the Midi almanac to ≤10 km (1σ). However, the position accuracy of the parameter-simplified Reduced almanac remains unspecified in the ICD. Therefore, verification using real broadcast data is necessary to confirm whether the operational characteristics defined in the ICD for both almanacs, such as the t_oa interval - specifically for the Reduced almanac messages - are consistently maintained. This study addresses these points through an experimental analysis using received QZSS CNAV L2C signal data. The primary objectives were to evaluate the operational specifications using QZSS CNAV data collected from experiments, in comparison with the ICD, and to quantify the position accuracy of the Reduced almanac, which is not specified in the ICD. To this end, we first analyzed the actual update interval of the t_oa shared by both almanacs and the reception intervals of Message Types 12 and 31 (containing the Reduced almanac). Subsequently, we evaluated the position accuracy derived from the Reduced almanac through comparison with precise orbits and by identifying error sources. The analysis showed that the update interval for the shared t_oavaried between 0.25 and 1.75 days, deviating from the one-day period defined in the ICD. The reception intervals for Message Types 12 and 31, which are typically 288 seconds, occasionally exceeded the 1200 second maximum defined in the ICD. The accuracy assessment confirmed the Midi almanac meets its specification, whereas the Reduced almanac exhibited position errors of several thousand kilometers, stemming from its use of fixed constants for dynamic orbital parameters. These findings provide an empirical evaluation of the QZSS almanacs' actual operational characteristics and the Reduced almanac's accuracy limitations. This information can serve as a reference for future almanac parameter design for a geosynchronized orbit used for regional navigation systems such as the Korean Positioning System (KPS).

Global Navigation Satellite System (GNSS) spoofing attacks usually rely on elevated signal power. As a result, the Carrier-to-Noise Density Ratio (C/N0) is a useful indicator for detection. In stationary GNSS receivers such as reference stations, receiver-velocity information cannot be used to verify signal authenticity, and detection thus relies primarily on signal quality indicators such as the received C/N0. Conventional C/N0 detection methods using a C/N0 value have limitations because C/N0 can vary due to environmental factors even under nominal conditions. Furthermore, spoofing-induced variations may not be clearly distinguishable or may be masked by interference such as jamming, leading to false alarms. To address these limitations, this study proposes a C/N0-based spoofing detection method that considers the ionospheric scintillation characteristics of the Korean Peninsula. Scintillation-induced C/N0 attenuation is modeled for the nominal (H0) state using a formulation of Tracking Error Variance at Output of Phase-Locked Loop (PLL) for L1 Global Positioning System (GPS), while spoofing conditions (H1) are represented using power advantages of {+3, +6, +9 dB} and a 0.25-chip code delay. A time-series detection statistic is constructed by estimating C/N0 every 1 ms and applying a sliding-window χ² metric with an adaptive threshold to mitigate C/N0 fluctuations caused by ionospheric scintillation. Simulation results show that the proposed method achieves an Area Under the ROC Curve (AUC) greater than 0.9 for all Pseudo-Random Noise (PRN) codes, thus confirming clear separability between H0 and H1. These findings demonstrate that the proposed approach provides an effective means of detecting C/N0-based spoofing for stationary GNSS reference stations in mid-latitude ionospheric scintillation environments such as the Korean Peninsula.

This paper proposes a robust optimization framework for indoor Wi-Fi access point (AP) localization that combines multi-start initialization with explicit handling of missing directional measurements. Our previous study employed an environment-adaptive path-loss model with directional bias compensation and showed that unknown APs can be localized with reasonable accuracy even under limited measurement conditions. However, the suggested approach suffered from performance instability due to sensitivity to initial values and the presence of unreliable or missing heading information. To address these issues, the method proposed in this study (i) filters out invalid directional measurements when estimating the path-loss exponent and directional biases, (ii) performs AP-wise localization using two complementary initial positions derived from geometric and signal-strength heuristics, and (iii) jointly refines the locations and transmit powers of all unknown APs while keeping the global path-loss exponent and directional biases fixed. Experiments conducted on Wi-Fi RSSI data collected at 48 measurement points in a real indoor environment show that the proposed algorithm reduces the mean localization error by approximately 38.5%, from 3.23 m to 1.99 m, compared with the previous work. In particular, for AP_Y, which exhibited the largest localization error with the previous work, the error is reduced to less than half across a wide range of data-usage ratios. These results indicate that the proposed framework improves the stability and practical applicability of model-based indoor Wi-Fi localization with unknown AP positions.

While perception in autonomous driving enables the detection of surrounding objects, such as pedestrians, vehicles, and road infrastructure, inaccuracies in perception algorithms may lead to critical situations, such as critical safety risks. Extensive validation across diverse object classes, repetitive experiments, and rigorous edge-case testing is thus required. Although driving simulators provide scalable and cost-effective testing environments, they are limited in accurately reproducing real vehicle dynamics and are therefore mainly used in the early development stages. This study proposes an augmentedreality LiDAR point cloud system for sim-to-real validation that integrates the advantages of real-world and simulation-based testing. Virtual LiDAR point clouds generated from simulator environments are fused with real-world LiDAR data, enabling the injection of virtual objects into real driving scenarios while preserving real vehicle dynamics and geometry. The implementation includes analysis and integration of point cloud density in both domains for realistic fusion, occlusion modeling of virtual LiDAR point clouds as observed in real environments, and a parallel processing algorithm for real-time data handling. The proposed approach can safely and accurately evaluate perception algorithms by allowing the injection of virtual objects into real-world scenarios, thereby ensuring that vehicle dynamics and geometry are faithfully represented. Furthermore, this method addresses the data-hungry nature of recent AI-based autonomous driving models by leveraging the scalability of simulator-based data augmentation.

Autonomous driving technologies have been studied across diverse platforms and environments, and localization remains an important task in this field. Although Global Navigation Satellite System (GNSS)-based navigation is commonly used, its performance can degrade due to its dependency on radio signals. This limitation motivates the need for radio-free, sensor-based navigation methods. In this work, we present a Light Detection and Ranging (LiDAR)-based place recognition method designed for Unmanned Aerial Vehicles (UAVs) rather than ground vehicles that can be used as loop-closure detection and relocalization. The proposed method includes a preprocessing step, a feature extraction and descriptor computation approach, and a voting-based candidate matching method. Finally, the proposed method is validated using real-world UAV flight datasets to demonstrate its effectiveness.

The Korean Positioning System (KPS), a regional satellite navigation system composed of seven to eight satellites, is susceptible to a lack of visible satellites in certain environments due to its small constellation size. Furthermore, securing a sufficient number of ground monitoring stations, which are essential for Precise Orbit Determination (POD) of KPS satellites, over a wide area on the Korean Peninsula is a significant challenge. This paper explores augmenting the KPS with a Low Earth Orbit (LEO) satellite constellation to address these limitations. For users, such an augmentation increases the number of visible satellites and thereby reduces the Dilution of Precision (DOP). For the system, LEO satellites can function as space-based mobile monitoring stations, effectively expanding the tracking network and improving satellite POD performance. While a larger LEO constellation yields greater performance benefits, it also incurs higher costs. Therefore, this study addresses the problem of designing a LEO constellation that minimizes the number of satellites while maximizing performance improvements. We formulate this as a multi-objective optimization problem with conflicting objectives, seeking Pareto optimal solutions. A genetic algorithm-based multi-objective optimization approach, which is well-suited for constellation design, is utilized to find the Pareto front. The results indicate that an efficient Pareto optimal solution for minimizing user DOP can be achieved with a constellation of approximately 24 LEO satellites. To enhance satellite POD performance, the most efficient design consists of six LEO satellites. Crucially, we designed an effective six to eight satellite LEO constellation that improves POD even when ground stations are confined to the domestic region and demonstrate the feasibility of domestic-only KPS ground infrastructure. The presented methodology can provide valuable guidelines for designing optimal LEO constellations to augment and enhance the future KPS.

Growing interest in space exploration in recent years has led to active research on lunar missions as the first step toward deep space exploration. Lunar exploration requires a variety of enabling technologies, among which establishing a reliable Position, Navigation and Timing (PNT) system on the Moon has gained increasing attention. Lunar missions traditionally have relied on one-to-one tracking from Earth-based ground stations. However, with the rapidly increasing number of lunar exploration missions, deep space communication networks such as the Deep Space Network (DSN) have become saturated. As an alternative, the use of Earth-based Global Navigation Satellite System (GNSS) signals, which are capable of supporting an unlimited number of users, has been proposed for lunar navigation, and extensive research efforts are currently in progress. In this study, the availability of Earth GNSS signals in a Elliptical Lunar Frozen Orbit (ELFO) around the Moon was analyzed through simulation and virtual measurements were generated based on the analysis results. The simulation results show that signal availability in the lunar ELFO environment is limited, with most observable signals originating from the side lobes of GNSS satellites. Simple Least Squares-based positioning approaches consequently exhibit performance limitations. These results indicate the necessity of developing more advanced filtering-based navigation methods suitable for the lunar environment.

Recent positioning technologies utilizing low Earth orbit (LEO) satellites are being actively studied as a promising complement to the limitations of Global Navigation Satellite Systems (GNSS). However, since Positioning, Navigation, and Timing (PNT) data for LEO satellites are not publicly available, existing approaches typically require specialized receivers or high-fidelity simulators. Their applications thus may be constrained by costs or research environment limitations. This study proposes a Doppler observation modeling approach based solely on publicly available Two-Line Element (TLE) data and MATLAB builtin functions, without requiring dedicated receivers or high-fidelity simulators. The proposed method is validated through least-squares-based Doppler positioning. A performance evaluation was conducted using 10 minutes of simulated Doppler data from Starlink LEO satellites, with the Gamaksan Observatory as the reference point. The proposed method achieved a 3D Root Mean Square Error (RMSE) of 4.3 m when more than 400 LEO satellites were utilized. As the number of available satellites decreased, the positioning error increased to 5.5 m with approximately 100 satellites and further to 9.8 m when only approximately 20 satellites were used. In comparison, GPS satellite-based Doppler positioning resulted in a 3D RMSE of 146.3 m. This large difference in RMSE demonstrates the superior performance of LEO-based Doppler positioning.

The Satellite-Based Augmentation System (SBAS) is a system that provides real-time correction and integrity information for satellite clock, orbit, and ionospheric delay errors using geostationary satellites to enhance the accuracy, integrity, continuity, and availability of Global Navigation Satellite Systems (GNSS). Satellite navigation systems including the Wide Area Augmentation System (WAAS) in the United States, the European Geostationary Navigation Overlay Service (EGNOS) in Europe, and the MTSAT Satellite-Based Augmentation System (MSAS) in Japan are currently in operation. In Korea, the Korea Augmentation Satellite System (KASS) has been developed to provide correction services for Global Positioning System (GPS) satellites. To ensure the reliability of SBAS services, it is essential to quantitatively monitor and validate the correction performance of individual error components such as satellite clock and orbit errors. In this study, the satellite clock and orbit correction performance provided by KASS was quantitatively analyzed using the International GNSS Service (IGS) final precise ephemeris and clock products as reference values. The analysis results showed that the Root Mean Square (RMS) User Range Error (URE) decreased from 0.45 m for broadcast navigation messages to 0.32 m when KASS corrections were applied. MSAS corrections, analyzed for comparison, exhibited an RMS URE of 0.25 m. These results confirm that KASS correction information effectively improves satellite clock and orbit accuracy for GPS satellites.

Global Navigation Satellite System (GNSS)-synchronized surface buoy networks using Time Difference of Arrival (TDOA) provide an effective solution for underwater localization in GNSS-denied environments. However, most previous studies have focused on planar or depth-invariant target motion. Practical underwater operations often involve targets moving at varying depths and irregular measurement updates due to acoustic channel conditions. This paper analyzes the localization characteristics of a GNSS-synchronized surface buoy TDOA system for underwater targets at varying depths. A three-dimensional simulation framework is developed to investigate the effect of target depth variation on measurement geometry and horizontal position error characteristics. In addition, the impact of irregular measurement updates on Extended Kalman Filter (EKF)-based tracking performance is examined. To mitigate depth-induced uncertainty, a depth aiding scheme using pressure sensor measurements is incorporated into the estimation framework. Simulation results show that target motion at varying depths leads to non-uniform horizontal error behavior even under regular update conditions, and that irregular updates can further amplify tracking instability. The application of depth aiding effectively suppresses horizontal error growth by constraining depth uncertainty and thereby improves overall estimation robustness. The results provide practical insights for the design and operation of GNSS-synchronized surface buoy TDOA localization systems in three-dimensional underwater scenarios.

The National Geographic Information Institute (NGII) is preparing to transition from dual-frequency to triple-frequency GNSS data services by upgrading equipment at 99 Continuously Operating Reference Stations (CORS) in Korea. In response to this transition, this paper proposes an enhanced cycle slip detection algorithm based on a pre-processing algorithm (KPrep1) from previous research to support triple-frequency observations and analyzed its performance. For performance validation, five days of triple-frequency data collected from NGII-operated stations were utilized, and the results were compared with those from RINGO, software developed by the Geospatial Information Authority of Japan (GSI). The comparative analysis showed that the difference in the number of detected cycle slips ranged from a minimum of one to a maximum of six, with an average difference of approximately two. Furthermore, a very high correlation coefficient of 0.98 was observed between the two programs.

The wide-area differential regional navigation satellite system (WAD-RNSS) integrates wide-area differential navigation technology with a regional navigation satellite system (RNSS) to provide navigation and correction signals as a unified service. The WAD-RNSS ground segment, which consists of reference stations, a master control station, and ground antennas, is in charge of generating information for navigation and correction signals. To ensure system stability, it is necessary to implement a security architecture to protect against threats. This paper presents a method to apply a security architecture to the WAD-RNSS ground segment.