Soohyun Kim

Computing services can be extended to infrastructure-limited areas through unmanned aerial vehicle (UAV)-assisted mobile edge computing (MEC). However, service provision continues to be challenging when spatially clustered ground users (GUs) are initially unknown and their intermittent workloads expire within finite validity windows. Therefore, conventional trajectory controllers assuming known GU locations and requests are unsuitable under sensing-limited conditions, where out-of-range workloads remain hidden between visits. Partially observable control has been adopted to address these challenges; however, existing frameworks are insufficient, as the discovery of unknown service regions and the repeated revisits required to re-observe hidden demand are not jointly considered. In this paper, a deadline-paced patrol with goal-conditioned multi-agent reinforcement learning (DPP-GCMARL) framework is proposed. In the proposed framework, a rule-based deadline-paced patrol (DPP) layer converts sensing-derived cell memory into deconflicted target cells using staleness normalized by the collection deadline. The assigned targets are then mapped to continuous movement commands by a learned goal-conditioned movement (GCM) layer. Simulation results show that DPP-GCMARL improves the end-to-end completed-workload ratio compared with multi-agent proximal policy optimization (MAPPO). It also increases exploration-coupled fairness and the worst-cluster collection ratio while reducing the mean travel per UAV. These results substantiate the proposed framework’s potential for deadline-constrained MEC service in infrastructure-limited environments.

In this paper, a supervised deep reinforcement learning (SDRL)-based positioning scheme is proposed for indoor wireless localization. The proposed scheme formulates the positioning problem as a Markov decision process and introduces a target-aware reward design based on the artificial potential field (APF) to alleviate the sparse reward problem commonly encountered in search-based reinforcement learning. In the proposed scheme, supervision is provided at the reward level by incorporating the target position into the reward design, rather than at the action level via expert demonstrations. A multi-scale action set with 49 candidates is further adopted to provide a favorable trade-off between estimation accuracy and search efficiency. An anchor-based environment construction strategy is developed by selecting the four strongest reference points (RPs) and transforming their coordinates with respect to the strongest RP. Simulation results show that the proposed scheme achieves a mean absolute error (MAE) below 0.8 m and success rates above 99.1% within 1 m and 99.2% within 2 m under the default Bluetooth Low Energy setting, while the convex-valid rate of the anchor-based environment exceeds 99.5%. Compared with existing methods, the proposed scheme reduces the MAE by approximately 92.3%. Ablation studies confirm that multi-scale actions reduce the average search steps by approximately 69.5% compared with a single-scale baseline. The proposed scheme also retains stable performance across BLE, Wi-Fi, and Zigbee infrastructures when trained under a representative path-loss setting without retraining and maintains sub-meter accuracy under mild shadow fading. These results confirm that the proposed scheme can improve positioning accuracy and search efficiency for indoor wireless localization.

In this letter, a novel diffusion model (DM)-based channel estimation method is proposed to mitigate abrupt performance degradation in millimeter wave (mmWave) multiple-input multiple-output (MIMO) systems under severe pilot scarcity. To address this issue, the proposed method jointly exploits first- and second-order posterior moments using the moment matching principle and Tweedie’s formula. This method improves estimation accuracy and robustness under high uncertainties, while maintaining a comparable computational burden. From the simulation results, it is confirmed that the proposed method successfully mitigates the abrupt performance degradation. The proposed method also substantially outperforms state-of-the-art DM-based methods in estimation accuracy under severe pilot scarcity, while achieving comparable computational complexity.

The recent development of 3D generative AI encompassing generation and editing technologies has been increasingly investigated to advance immersive applications. To enrich visual aesthetics, 3D stylization techniques focus on transferring artistic effects from reference style images to 3D scenes. However, existing 3D stylization techniques primarily focus on global style transfer, which can result in unwanted modifications to background regions and a lack of localized control. To address these limitations, we propose LocalGaussStyle, a novel approach for localized style transfer on scenes represented by 3D Gaussian splatting. The proposed pipeline consists of two phases: object localization and localized stylization. First, 2D instance segmentation masks are projected into a 3D scene to precisely localize target objects. Next, a boundary-aware optimization is designed to perform style transfer and mitigate style leakage caused by the spatial overlap of Gaussians. In addition, geometry-decoupled adaptive densification (GDAD) is employed to enhance the geometric resolution of Gaussians within the target object, thereby improving the representation capacity. The LocalGaussStyle facilitates high-fidelity style transfer that preserves the geometry and appearance of the non-target regions. In terms of style fidelity and background preservation, the effectiveness and efficiency of the proposed method are demonstrated through extensive experiments conducted on various scenes and reference style images.

With the recent advancement of virtual try-on (VTO) technology, it is being applied to various fields. Although advancements in VTO technology have enabled not only 2D but also 3D visualization, applying style transfer to hyper-local regions remains challenging due to the complex surface curvature and ambiguous boundaries of 3D objects. To address these challenges, we propose an immersive 3D style transfer framework based on 3D Gaussian splatting. A segmentation model is employed to accurately segment target regions, and a large-scale specialized dataset is constructed to capture the morphological diversity of human hands. Furthermore, neural style transfer is integrated with the 3D representation to enable precise style application to hyper-local regions. The proposed framework achieves a mean intersection of union (mIoU) of 0.806 in segmentation and high-fidelity stylization with learned perceptual image patch similarity (LPIPS) and reference-based LPIPS (Ref-LPIPS) scores of 0.1472 and 0.0196, respectively. These results indicate that the proposed framework can provide the quality requirements and immersive VTO experience.

An integration of deep learning and classical filtering techniques has been developed for achieving an accurate and robust positioning system. Despite significant technological advances, positioning systems often face fundamental challenges such as signal obstruction, multipath interference, and error drift. Although 3D map matching techniques using LiDAR sensors can mitigate these challenges, their practical implementation is limited by high costs, computational burdens, and dependency on pre-modeled environments. In response, recent research has increasingly focused on enhancing positioning systems through deep learning–based sensor fusion and adaptive filtering methods, emphasizing improvements in accuracy and operational robustness. In this paper, a seamless hierarchical context-aware intelligent positioning system (SH-CAIPS) is proposed by integrating sensor fusion, two-stage outlier mitigation gates, and deep reinforcement learning (DRL)-based noise scaling with an innovation-based adaptive Kalman filter (IAKF). As a result, the proposed positioning system can enhance robustness by adaptively adjusting measurement noise and dynamically updating the position based on the confidence score between model predictions and sensor measurements. From simulation results, it is confirmed that the positioning performances can be significantly improved based on the proposed SH-CAIPS compared with conventional positioning systems.

A wide range of research has been conducted on path planning and collision avoidance to enhance the operational efficiency of unmanned aerial vehicles (UAVs). The existing works have mainly assumed an environment with static obstacles and global positioning system (GPS) signals. However, practical environments have often been involved with dynamic obstacles, dense areas with numerous obstacles in confined spaces, and blocked GPS signals. In order to consider these issues for practical implementation, a deep reinforcement learning (DRL)-based method is proposed for path planning and collision avoidance in GPS-denied and dense environments. In the proposed method, robust path planning and collision avoidance can be conducted by using the received signal strength (RSS) value with the extended Kalman filter (EKF). Additionally, the attitude of the UAV is adopted as part of the action space to enable the generation of smooth trajectories. Performance was evaluated under single- and multi-target scenarios with numerous dynamic obstacles. Simulation results demonstrated that the proposed method can generate smoother trajectories and shorter path lengths while consistently maintaining a lower collision rate compared to conventional methods.

Reliable connectivity for Internet of Things (IoT) in remote regions remains a challenging issue due to the rapidly varying Doppler shifts inherent in low Earth orbit (LEO) satellite systems. Therefore, conventional transmission schemes struggle under the dynamic LEO conditions, which can lead to degraded communication quality and a restriction of the efficacy of conventional transmission techniques. Semantic communication (SC) has been adopted to address these challenges by transmitting task-relevant information rather than raw data; however, the current SC frameworks are insufficient for LEO environments, as they rely on fixed signal-to-noise ratio (SNR) boundaries for modulation scheme selection. In this article, we propose an adaptive SC framework tailored for LEO-based IoT systems that integrates a deep reinforcement learning (DRL) agent with a vector-quantized variational autoencoder (VQ-VAE). With the proposed framework, modulation schemes are dynamically selected by the DRL agent based on real-time Doppler and delay spread, while residual frequency offsets at the receiver are mitigated by a post-equalization. Simulation results show that the proposed method can achieve comparable reconstruction quality with substantially lower symbol overhead, memory usage, and inference time compared to conventional SC baselines. The potential of the proposed framework to facilitate efficient and robust SC in LEO satellite-enabled IoT networks is substantiated by these results.

The orthogonal frequency-division multiplexing (OFDM) transmission technique is well known to be efficient for data transmission but is susceptible to performance degradation due to factors such as high-order modulation schemes, multipath fading, and noise. In this paper, an approach for reconstructing images received by the OFDM transmission technique is proposed based on generative AI. The approach exploits a conditional diffusion model (CDM) that incorporates conditional factors reflecting the degree of distortion in the received images by the OFDM technique. Additionally, it employs a method to learn the variance in the reverse process during training, considering the level of distortion associated with various modulation schemes. Through this adaptability, the proposed model is experimentally demonstrated to optimize image reconstruction performance under various modulation schemes in low-SNR environments. The proposed conditional diffusion model can enhance the PSNR of OFDM-based received images by up to 8 dB in low-SNR conditions with various modulation schemes.