Xiaobin Hong
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Xiaobin Hong has a total of 207 co-authors. Here are the top co-authors by number of shared publications:
Jian Wu
59 shared publications- Jian Wu
School of Electronic Engineering, Beijing University of Posts and Telecommunications, Beijing 100876, China
35 shared publications Jifang Qiu
State Key Laboratory of Information Photonics and Optical Communications, Beijing University of Posts and Telecommunicationshttps://ror.org/04w9fbh59, Beijing 100876, China
26 shared publications- Zhisheng Yang
State Key Laboratory of Information Photonics & Optical Communications Beijing University of Posts and Telecommunications Beijing China
26 shared publications Jintong Lin
Beijing University of Posts and Telecommunications, State Key Laboratory of Information Photonics and Optical Communications, P. O. Box 55, Beijing 100876, China
25 shared publicationsHongxiang Guo
State Key Laboratory of Information Photonics and Optical Communications, Beijing University of Posts and Telecommunicationshttps://ror.org/04w9fbh59, Beijing 100876, China
23 shared publications
Latest Publications
- 23 January 2026
- Published by Optica Publishing Group in Optics Express
- Vol. 34 (2) , 3444-3463
Geometry-aware multimodal fusion for large-scale 3D scene understanding
Multimodal 3D semantic segmentation plays a crucial role in engineering applications such as autonomous driving, robotics, and 3D image acquisition and display systems. However, the complex geometry of real scenes, varying object scales, and the inherent sparsity and non-uniform sampling of LiDAR-based acquisition still present significant challenges. To address these issues, we propose an adaptive geometry fusion method for robust and efficient multimodal 3D understanding. The method adopts a dual-path 3D feature extractor: one path captures spatial structural relationships through position-based encoding, while the other integrates a geometry-aware adaptive aggregation module that models local structures using learnable kernel positions and hybrid distance–appearance weighting. Complementary image features from passive optical imaging are fused with LiDAR features to enhance semantic discrimination in regions with sparse or ambiguous 3D measurements. The overall framework remains lightweight, containing only 5.2M parameters, and achieves real-time inference at 25 ms per sample, enabling deployment in resource-constrained sensing and visualization platforms. Experiments on Semantic3D, SemanticKITTI (71.2% mIoU, surpassing the lightweight baseline RandLA-Net by 15.3%), and nuScenes demonstrate competitive performance and strong generalization. Furthermore, evaluations on real-world colored point clouds acquired from a LiDAR–camera system validate the effectiveness of the proposed method in outdoor environments, providing semantically structured 3D representations suitable for downstream 3D display and visualization applications.
- 23 January 2026
- Published by Optica Publishing Group in Optics Express
- Vol. 34 (2) , 3444-3463
Geometry-aware multimodal fusion for large-scale 3D scene understanding
Multimodal 3D semantic segmentation plays a crucial role in engineering applications such as autonomous driving, robotics, and 3D image acquisition and display systems. However, the complex geometry of real scenes, varying object scales, and the inherent sparsity and non-uniform sampling of LiDAR-based acquisition still present significant challenges. To address these issues, we propose an adaptive geometry fusion method for robust and efficient multimodal 3D understanding. The method adopts a dual-path 3D feature extractor: one path captures spatial structural relationships through position-based encoding, while the other integrates a geometry-aware adaptive aggregation module that models local structures using learnable kernel positions and hybrid distance–appearance weighting. Complementary image features from passive optical imaging are fused with LiDAR features to enhance semantic discrimination in regions with sparse or ambiguous 3D measurements. The overall framework remains lightweight, containing only 5.2M parameters, and achieves real-time inference at 25 ms per sample, enabling deployment in resource-constrained sensing and visualization platforms. Experiments on Semantic3D, SemanticKITTI (71.2% mIoU, surpassing the lightweight baseline RandLA-Net by 15.3%), and nuScenes demonstrate competitive performance and strong generalization. Furthermore, evaluations on real-world colored point clouds acquired from a LiDAR–camera system validate the effectiveness of the proposed method in outdoor environments, providing semantically structured 3D representations suitable for downstream 3D display and visualization applications.
- Open Access
Unsupervised Optical-Sensor Extrinsic Calibration via Dual-Transformer Alignment
Accurate extrinsic calibration between optical sensors, such as camera and LiDAR, is crucial for multimodal perception. Traditional methods based on specific calibration targets exhibit poor robustness in complex optical environments such as glare, reflections, or low light, and they rely on cumbersome manual operations. To address this, we propose a fully unsupervised, end-to-end calibration framework. Our approach adopts a dual-Transformer architecture: a Vision Transformer extracts semantic features from the image stream, while a Point Transformer captures the geometric structure of the 3D LiDAR point cloud. These cross-modal representations are aligned and fused through a neural network, and a regression algorithm is used to obtain the 6-DoF extrinsic transformation matrix. A multi-constraint loss function is designed to enhance structural consistency between modalities, thereby improving calibration stability and accuracy. On the KITTI benchmark, our method achieves a mean rotation error of 0.21° and a translation error of 3.31 cm; on a self-collected dataset, it attains an average reprojection error of 1.52 pixels. These results demonstrate a generalizable and robust solution for optical-sensor extrinsic calibration, enabling precise and self-sufficient perception in real-world applications.
Open AccessUnsupervised Optical-Sensor Extrinsic Calibration via Dual-Transformer Alignment
Accurate extrinsic calibration between optical sensors, such as camera and LiDAR, is crucial for multimodal perception. Traditional methods based on specific calibration targets exhibit poor robustness in complex optical environments such as glare, reflections, or low light, and they rely on cumbersome manual operations. To address this, we propose a fully unsupervised, end-to-end calibration framework. Our approach adopts a dual-Transformer architecture: a Vision Transformer extracts semantic features from the image stream, while a Point Transformer captures the geometric structure of the 3D LiDAR point cloud. These cross-modal representations are aligned and fused through a neural network, and a regression algorithm is used to obtain the 6-DoF extrinsic transformation matrix. A multi-constraint loss function is designed to enhance structural consistency between modalities, thereby improving calibration stability and accuracy. On the KITTI benchmark, our method achieves a mean rotation error of 0.21° and a translation error of 3.31 cm; on a self-collected dataset, it attains an average reprojection error of 1.52 pixels. These results demonstrate a generalizable and robust solution for optical-sensor extrinsic calibration, enabling precise and self-sufficient perception in real-world applications.
- 17 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 43 (23) , 10752-10760
Mitigation of Stimulated Raman Effects in Coded Raman Optical Time-Domain Reflectometry
Distributed temperature sensors based on coded Raman optical time-domain reflectometry (ROTDR) obtain distributed temperature profiles along a sensing fiber by decoding backscattered anti-Stokes and Stokes signals obtained with coded pulse sequences. However, using long sequences or pulses with high peak power can induce backward stimulated Raman scattering (SRS), disrupting the linear superposition of spontaneous Raman scattering responses required for reliable decoding, and causing significant temperature estimation errors. To address this, this paper proposes a compensation method that estimates SRS-induced gain and depletion relative to a low-power coded Stokes and anti-Stokes references acquired under negligible SRS conditions. This correction restores the linearity of Raman traces with minimal added complexity, enabling a correct decoding process that minimizes errors in the retrieved temperature profile. Experimental results over a 24 km fiber using 130-bit genetic-optimized codes and 1 W peak power pulses demonstrate that the method effectively suppresses SRS-induced distortions without compromising the spatial resolution or the SNR improvement provided by the coding gain. Compared to direct decoding, the method reduces temperature errors by up to 18 °C (under specific tested conditions) and improves the retrieval temperature accuracy, demonstrating scalability to longer code lengths and higher launch powers.
- 17 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 43 (23) , 10752-10760
Mitigation of Stimulated Raman Effects in Coded Raman Optical Time-Domain Reflectometry
Distributed temperature sensors based on coded Raman optical time-domain reflectometry (ROTDR) obtain distributed temperature profiles along a sensing fiber by decoding backscattered anti-Stokes and Stokes signals obtained with coded pulse sequences. However, using long sequences or pulses with high peak power can induce backward stimulated Raman scattering (SRS), disrupting the linear superposition of spontaneous Raman scattering responses required for reliable decoding, and causing significant temperature estimation errors. To address this, this paper proposes a compensation method that estimates SRS-induced gain and depletion relative to a low-power coded Stokes and anti-Stokes references acquired under negligible SRS conditions. This correction restores the linearity of Raman traces with minimal added complexity, enabling a correct decoding process that minimizes errors in the retrieved temperature profile. Experimental results over a 24 km fiber using 130-bit genetic-optimized codes and 1 W peak power pulses demonstrate that the method effectively suppresses SRS-induced distortions without compromising the spatial resolution or the SNR improvement provided by the coding gain. Compared to direct decoding, the method reduces temperature errors by up to 18 °C (under specific tested conditions) and improves the retrieval temperature accuracy, demonstrating scalability to longer code lengths and higher launch powers.
- 28 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 43 (5) , 2418-2427
Modeling the Demodulated Phase Precision of Standard Heterodyne-Detection Phase-Sensitive Optical Time-Domain Reflectometry
An analytical model linking the demodulated phase precision and system parameters for standard heterodyne-detection phase-sensitive optical time-domain reflectometry ( $\Phi$ -OTDR) is proposed and experimentally validated. This model innovatively leverages the independence of additive noise and phase noise, circumventing the traditional challenge of clarifying the intricate relationship between phase-noise-involved signal-to-noise ratio (SNR) and the phase precision. The developed model serves as a theoretical tool for understanding how phase precision is determined by system parameters, allowing for theoretical anticipations of phase precision at any fiber position under given measurement conditions. Additionally, the model enables the optimization of system parameters for high performance and low cost, with no need to build up an experimental system.
- 28 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 43 (5) , 2418-2427
Modeling the Demodulated Phase Precision of Standard Heterodyne-Detection Phase-Sensitive Optical Time-Domain Reflectometry
An analytical model linking the demodulated phase precision and system parameters for standard heterodyne-detection phase-sensitive optical time-domain reflectometry ( $\Phi$ -OTDR) is proposed and experimentally validated. This model innovatively leverages the independence of additive noise and phase noise, circumventing the traditional challenge of clarifying the intricate relationship between phase-noise-involved signal-to-noise ratio (SNR) and the phase precision. The developed model serves as a theoretical tool for understanding how phase precision is determined by system parameters, allowing for theoretical anticipations of phase precision at any fiber position under given measurement conditions. Additionally, the model enables the optimization of system parameters for high performance and low cost, with no need to build up an experimental system.
- 21 October 2024
- Published by Optica Publishing Group in Optics Express
- Vol. 32 (23) , 40244-40258
Omnidirectional path loss model for non-line-of-sight ultraviolet communications
To analyze the channel characteristics in omnidirectional reception scenarios for non-line-of-sight ultraviolet (UV) communication, we derive expressions for the reception direction distribution based on the incident photons at the receiver and propose an omnidirectional reception path loss (PL) channel model based on the Monte Carlo (MC) method. Furthermore, we validate the proposed omnidirectional model by comparing it with the existing MC numerical model that traverses all reception directions. Results indicate that the average computation time of the proposed omnidirectional PL model is less than 0.03% of the traversal model while maintaining comparable accuracy. Additionally, we present the variations in off-axis and inclination angles corresponding to the receiver’s direction at different coordinate positions. Therefore, this paper provides valuable guidance for rapidly determining the omnidirectional energy field distribution and optimizing receiver orientation in UV communication systems.
- 21 October 2024
- Published by Optica Publishing Group in Optics Express
- Vol. 32 (23) , 40244-40258
Omnidirectional path loss model for non-line-of-sight ultraviolet communications
To analyze the channel characteristics in omnidirectional reception scenarios for non-line-of-sight ultraviolet (UV) communication, we derive expressions for the reception direction distribution based on the incident photons at the receiver and propose an omnidirectional reception path loss (PL) channel model based on the Monte Carlo (MC) method. Furthermore, we validate the proposed omnidirectional model by comparing it with the existing MC numerical model that traverses all reception directions. Results indicate that the average computation time of the proposed omnidirectional PL model is less than 0.03% of the traversal model while maintaining comparable accuracy. Additionally, we present the variations in off-axis and inclination angles corresponding to the receiver’s direction at different coordinate positions. Therefore, this paper provides valuable guidance for rapidly determining the omnidirectional energy field distribution and optimizing receiver orientation in UV communication systems.
- Open Access
- 9 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 42 (16) , 5786-5796
Analytical Signal-to-Noise Ratio Model on Frequency-Scanned Brillouin Optical Time-Domain Reflectometry
An analytical signal-to-noise ratio (SNR) model of standard frequency-scanned Brillouin optical time-domain reflectometry (FS-BOTDR) is established, by comprehensively taking into account nearly all non-negligible signal and noise terms with substantial impact on the sensing performance. The accuracy and generality of the proposed model are experimentally validated, supported by the fact that under various experimental conditions all experimental results are in good quantitative agreement with theoretical calculations. Serving as a theoretical tool, the established model makes it possible to accurately anticipate and comprehensively interpret the SNR behavior of the FS-BOTDR under any given measurement condition, without the need to construct an experimental system or carry out experiments. It also enables theoretically quantifying the SNR difference between FS-BOTDR and other distributed optical fiber sensors that already have mature SNR models. Furthermore, the proposed model provides guidelines on optimizing system parameters towards the highest possible SNR with minimized energy consumption.
- 9 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 42 (16) , 5786-5796
Open AccessAnalytical Signal-to-Noise Ratio Model on Frequency-Scanned Brillouin Optical Time-Domain Reflectometry
An analytical signal-to-noise ratio (SNR) model of standard frequency-scanned Brillouin optical time-domain reflectometry (FS-BOTDR) is established, by comprehensively taking into account nearly all non-negligible signal and noise terms with substantial impact on the sensing performance. The accuracy and generality of the proposed model are experimentally validated, supported by the fact that under various experimental conditions all experimental results are in good quantitative agreement with theoretical calculations. Serving as a theoretical tool, the established model makes it possible to accurately anticipate and comprehensively interpret the SNR behavior of the FS-BOTDR under any given measurement condition, without the need to construct an experimental system or carry out experiments. It also enables theoretically quantifying the SNR difference between FS-BOTDR and other distributed optical fiber sensors that already have mature SNR models. Furthermore, the proposed model provides guidelines on optimizing system parameters towards the highest possible SNR with minimized energy consumption.
- 30 April 2024
- Published by Optica Publishing Group in Optics Letters
- Vol. 49 (9) , 2469-2472
Real-time FPGA prototyping of Doppler frequency shift compensation using DSP-assisted automatic frequency control
This Letter presents a real-time coherent receiver using digital signal processing (DSP)-assisted automatic frequency control (AFC) to compensate for the Doppler frequency shift (DFS). DFS compensation range of ±8 GHz and the frequency shifting rate of 33 MHz/s are demonstrated in an FPGA-based 2.5 Gbaud QPSK coherent optical system. The experimental results indicate that the scheme achieves a sensitivity of −47 dBm at a bit error rate (BER) of 2E-4. The power penalty induced by the DFS compensation is less than 1 dB.
- 30 April 2024
- Published by Optica Publishing Group in Optics Letters
- Vol. 49 (9) , 2469-2472
Real-time FPGA prototyping of Doppler frequency shift compensation using DSP-assisted automatic frequency control
This Letter presents a real-time coherent receiver using digital signal processing (DSP)-assisted automatic frequency control (AFC) to compensate for the Doppler frequency shift (DFS). DFS compensation range of ±8 GHz and the frequency shifting rate of 33 MHz/s are demonstrated in an FPGA-based 2.5 Gbaud QPSK coherent optical system. The experimental results indicate that the scheme achieves a sensitivity of −47 dBm at a bit error rate (BER) of 2E-4. The power penalty induced by the DFS compensation is less than 1 dB.
- 15 November 2023
- Published by Optica Publishing Group in Optics Express
- Vol. 31 (24) , 40705-40716
Real-time low-complexity diversity combining algorithm for free space coherent optical communication systems over atmospheric turbulence channel
A novel diversity combining scheme, in conjunction with the complex-valued decision-directed least mean square (CV-DD-LMS) algorithm, is evaluated, and a real-time experimental validation is presented. This proposed scheme employs the CV-DD-LMS algorithm to concurrently perform beam combination and carrier phase recovery (CPR), thereby effectively reducing the overall complexity of digital signal processing. Furthermore, in the numerical simulation, under a low signal-to-noise ratio (SNR), a scheme utilizing the CV-DD-LMS algorithm effectively avoids cycle slips (CS) and outperforms schemes employing independent CPR modules. We experimentally validate this novel scheme by implementing it on an FPGA in a real-time 2.5Gb/s QPSK diversity-receiving system with three inputs. The back-to-back sensitivity is assessed using static received optical power, while the dynamic performance is evaluated by employing variable optical attenuators (VOAs) to simulate a power fluctuation at a frequency of 100kHz. The result proves that the implementation of the CV-DD-LMS algorithm yields stable performance while effectively reducing computational complexity.
- 15 November 2023
- Published by Optica Publishing Group in Optics Express
- Vol. 31 (24) , 40705-40716
Real-time low-complexity diversity combining algorithm for free space coherent optical communication systems over atmospheric turbulence channel
A novel diversity combining scheme, in conjunction with the complex-valued decision-directed least mean square (CV-DD-LMS) algorithm, is evaluated, and a real-time experimental validation is presented. This proposed scheme employs the CV-DD-LMS algorithm to concurrently perform beam combination and carrier phase recovery (CPR), thereby effectively reducing the overall complexity of digital signal processing. Furthermore, in the numerical simulation, under a low signal-to-noise ratio (SNR), a scheme utilizing the CV-DD-LMS algorithm effectively avoids cycle slips (CS) and outperforms schemes employing independent CPR modules. We experimentally validate this novel scheme by implementing it on an FPGA in a real-time 2.5Gb/s QPSK diversity-receiving system with three inputs. The back-to-back sensitivity is assessed using static received optical power, while the dynamic performance is evaluated by employing variable optical attenuators (VOAs) to simulate a power fluctuation at a frequency of 100kHz. The result proves that the implementation of the CV-DD-LMS algorithm yields stable performance while effectively reducing computational complexity.
- 4 November 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 01-05
Digital Back Propagation Algorithm for Nonlinear Compensation in 54.5GBaud 4000km 16QAM Raman Amplification Transmission System
In this paper, a digital back propagation (DBP) nonlinearity compensation algorithm is proposed and experimentally investigated over backward-pumped Raman amplified 4000-km fiber links employing 54.5-GBaud 16-ary quadrature amplitude modulation (16-QAM). By utilizing DBP for Raman amplification, significant improvements are achieved. The Q-factor shows an enhancement of 0.83 dB, and the optimal launch power is improved by 3 dBm. These results highlight the effectiveness of DBP in mitigating nonlinear impairments and optimizing the performance of Raman amplification systems.
- 4 November 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 01-05
Digital Back Propagation Algorithm for Nonlinear Compensation in 54.5GBaud 4000km 16QAM Raman Amplification Transmission System
In this paper, a digital back propagation (DBP) nonlinearity compensation algorithm is proposed and experimentally investigated over backward-pumped Raman amplified 4000-km fiber links employing 54.5-GBaud 16-ary quadrature amplitude modulation (16-QAM). By utilizing DBP for Raman amplification, significant improvements are achieved. The Q-factor shows an enhancement of 0.83 dB, and the optimal launch power is improved by 3 dBm. These results highlight the effectiveness of DBP in mitigating nonlinear impairments and optimizing the performance of Raman amplification systems.
- 4 November 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-4
An Image-free Location Method Using Photonic Integrated Interferometric System
Object location technology has wide application prospects in the military, sports, security, transportation and other fields. This paper proposes an image-free target location method by combining photonic integrated circuits and the Fourier projection slicing theorem.
- 4 November 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-4
An Image-free Location Method Using Photonic Integrated Interferometric System
Object location technology has wide application prospects in the military, sports, security, transportation and other fields. This paper proposes an image-free target location method by combining photonic integrated circuits and the Fourier projection slicing theorem.