Selected work

Representative Papers

Affine-Aligned Atlas for Canonical Gaussian Construction in Video Representation

Oct 01, 2026

This study addresses the challenge in dynamic 3D Gaussian Splatting where global video motion causes misalignment between the canonical space and individual frames, thereby imposing excessive burden on deformation models. To mitigate this, we propose constructing an affine-aligned atlas space that explicitly absorbs global motion via per-frame affine transformations prior to canonical Gaussian construction, effectively decoupling global motion from local deformations. Notably, this strategy modifies only the canonical construction phase, enabling seamless integration into existing frameworks with negligible parameter overhead. Our approach substantially improves reconstruction quality for sequences involving significant camera motion, offering an efficient and generalizable optimization paradigm for dynamic scene modeling.

0 citationsRead paper

Measurement Campaigns, Datasets, and Curve Fitting Officially Used by 3GPP in the Release 19 for Channel Modeling in TR 38.901 for 7-24 GHz

Mar 26, 2026

This work addresses the absence of a 3GPP-adopted standardized channel model and corresponding measurement data for the 7–24 GHz frequency band. To bridge this gap, the study systematically conducts millimeter-wave channel measurements and constructs a large-scale, wideband measurement dataset spanning the entire 7–24 GHz range. Channel parameters are extracted and modeled in strict compliance with the 3GPP TR 38.901 framework, followed by rigorous curve fitting. The resulting dataset and modeling methodology constitute the foundational channel model officially adopted by 3GPP Release 19, marking the first public release of such standardized data and procedures for this frequency range. This contribution provides authoritative support for the design of 5G-Advanced and 6G high-frequency communication systems and advances the standardization and unification of millimeter-wave channel modeling.

0 citationsRead paper

Contour Information Aware 2D Gaussian Splatting for Image Representation

Dec 29, 2025

Existing 2D Gaussian Splatting (2DGS) suffers from blurred edges and boundary distortions under low Gaussian counts (<1k) due to insufficient contour awareness. This work proposes semantic-guided contour-aware 2DGS: it is the first to incorporate segmentation priors into the 2DGS framework, constraining Gaussian distributions within semantic regions and suppressing cross-boundary mixing. We design region-constrained rasterization and a warm-up progressive training strategy to enhance edge convergence stability. Evaluated on benchmarks including DAVIS, our method significantly improves edge PSNR and SSIM, preserves sharp contours even with extremely sparse Gaussians, achieves real-time rendering (>60 FPS), and maintains memory overhead comparable to vanilla 2DGS. Key contributions include: (i) contour-aware geometric modeling, (ii) semantic-driven Gaussian allocation, and (iii) an efficient boundary optimization paradigm.

0 citationsRead paper

Transformer-Based Rate Prediction for Multi-Band Cellular Handsets

Sep 29, 2025

Mobile multi-band cellular terminals face rapid channel variations and sparse measurements due to mobility, hand occlusion, and hardware constraints, severely hindering accurate cross-band throughput prediction. Method: This paper proposes a Transformer-based asynchronous multi-antenna–multi-band joint modeling framework. It is the first to leverage asynchronously sampled multi-array throughput time series for fine-grained throughput prediction in FR1/FR3 microcell scenarios. The model integrates ray-tracing simulation data to enhance physical interpretability and generalization. Contribution/Results: Evaluated in dense urban environments, the approach significantly improves prediction accuracy—reducing mean absolute error by 28.6% over baseline methods—and enhances band-selection reliability. It enables intelligent, low-overhead band scheduling for resource-constrained terminals without requiring synchronous multi-antenna measurements or real-time channel state feedback.

0 citationsRead paper

Single and Multi-Frequency Path Loss Models for Indoor Hotspot Scenario Based on Measurements Conducted at 6.75, 16.95, 28, 73 and 142 GHz

Sep 08, 2025

High-frequency channel modeling for indoor hotspot (InH) scenarios lacks comprehensive empirical validation across ultra-wide frequency bands. Method: Leveraging multi-band measurement data spanning 6.75–142 GHz—including 7–24 GHz, 0.5–100 GHz, and 0.5–150 GHz—this study systematically develops and validates single- and multi-frequency large-scale path loss models under both line-of-sight (LOS) and non-LOS (NLOS) conditions. It introduces high-frequency extensions of the CIFX and ABGX models and employs 1-GHz-bandwidth time-domain sliding correlation probing via the NYU WIRELESS platform for high-accuracy channel characterization. Contribution/Results: For the first time, CI, FI, and ABG model parameters are uniformly calibrated across 0.5–150 GHz. The proposed 7–24 GHz and 0.5–100 GHz models have been adopted into the 3GPP Release-19 study item, directly supporting the evolution of 3GPP TR 38.901 and providing critical empirical foundations for ITU/3GPP InH standardization.

0 citationsRead paper

Recent publications

Latest Papers

Affine-Aligned Atlas for Canonical Gaussian Construction in Video Representation

Oct 01, 2026

This study addresses the challenge in dynamic 3D Gaussian Splatting where global video motion causes misalignment between the canonical space and individual frames, thereby imposing excessive burden on deformation models. To mitigate this, we propose constructing an affine-aligned atlas space that explicitly absorbs global motion via per-frame affine transformations prior to canonical Gaussian construction, effectively decoupling global motion from local deformations. Notably, this strategy modifies only the canonical construction phase, enabling seamless integration into existing frameworks with negligible parameter overhead. Our approach substantially improves reconstruction quality for sequences involving significant camera motion, offering an efficient and generalizable optimization paradigm for dynamic scene modeling.

0 citationsRead paper

Measurement Campaigns, Datasets, and Curve Fitting Officially Used by 3GPP in the Release 19 for Channel Modeling in TR 38.901 for 7-24 GHz

Mar 26, 2026

This work addresses the absence of a 3GPP-adopted standardized channel model and corresponding measurement data for the 7–24 GHz frequency band. To bridge this gap, the study systematically conducts millimeter-wave channel measurements and constructs a large-scale, wideband measurement dataset spanning the entire 7–24 GHz range. Channel parameters are extracted and modeled in strict compliance with the 3GPP TR 38.901 framework, followed by rigorous curve fitting. The resulting dataset and modeling methodology constitute the foundational channel model officially adopted by 3GPP Release 19, marking the first public release of such standardized data and procedures for this frequency range. This contribution provides authoritative support for the design of 5G-Advanced and 6G high-frequency communication systems and advances the standardization and unification of millimeter-wave channel modeling.

0 citationsRead paper

Contour Information Aware 2D Gaussian Splatting for Image Representation

Dec 29, 2025

Existing 2D Gaussian Splatting (2DGS) suffers from blurred edges and boundary distortions under low Gaussian counts (<1k) due to insufficient contour awareness. This work proposes semantic-guided contour-aware 2DGS: it is the first to incorporate segmentation priors into the 2DGS framework, constraining Gaussian distributions within semantic regions and suppressing cross-boundary mixing. We design region-constrained rasterization and a warm-up progressive training strategy to enhance edge convergence stability. Evaluated on benchmarks including DAVIS, our method significantly improves edge PSNR and SSIM, preserves sharp contours even with extremely sparse Gaussians, achieves real-time rendering (>60 FPS), and maintains memory overhead comparable to vanilla 2DGS. Key contributions include: (i) contour-aware geometric modeling, (ii) semantic-driven Gaussian allocation, and (iii) an efficient boundary optimization paradigm.

0 citationsRead paper

Transformer-Based Rate Prediction for Multi-Band Cellular Handsets

Sep 29, 2025

Mobile multi-band cellular terminals face rapid channel variations and sparse measurements due to mobility, hand occlusion, and hardware constraints, severely hindering accurate cross-band throughput prediction. Method: This paper proposes a Transformer-based asynchronous multi-antenna–multi-band joint modeling framework. It is the first to leverage asynchronously sampled multi-array throughput time series for fine-grained throughput prediction in FR1/FR3 microcell scenarios. The model integrates ray-tracing simulation data to enhance physical interpretability and generalization. Contribution/Results: Evaluated in dense urban environments, the approach significantly improves prediction accuracy—reducing mean absolute error by 28.6% over baseline methods—and enhances band-selection reliability. It enables intelligent, low-overhead band scheduling for resource-constrained terminals without requiring synchronous multi-antenna measurements or real-time channel state feedback.

0 citationsRead paper

Single and Multi-Frequency Path Loss Models for Indoor Hotspot Scenario Based on Measurements Conducted at 6.75, 16.95, 28, 73 and 142 GHz

Sep 08, 2025

High-frequency channel modeling for indoor hotspot (InH) scenarios lacks comprehensive empirical validation across ultra-wide frequency bands. Method: Leveraging multi-band measurement data spanning 6.75–142 GHz—including 7–24 GHz, 0.5–100 GHz, and 0.5–150 GHz—this study systematically develops and validates single- and multi-frequency large-scale path loss models under both line-of-sight (LOS) and non-LOS (NLOS) conditions. It introduces high-frequency extensions of the CIFX and ABGX models and employs 1-GHz-bandwidth time-domain sliding correlation probing via the NYU WIRELESS platform for high-accuracy channel characterization. Contribution/Results: For the first time, CI, FI, and ABG model parameters are uniformly calibrated across 0.5–150 GHz. The proposed 7–24 GHz and 0.5–100 GHz models have been adopted into the 3GPP Release-19 study item, directly supporting the evolution of 3GPP TR 38.901 and providing critical empirical foundations for ITU/3GPP InH standardization.

0 citationsRead paper