Xinyang Li
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Xinyang Li has a total of 20 co-authors. Here are the top co-authors by number of shared publications:
- Songjie Yang
National Key Laboratory of Science and Technology on Communications, University of Electronic Science and Technology of China, Chengdu, China
6 shared publications - Xiang Ling
National Key Laboratory of Wireless Communications, University of Electronic Science and Technology of China, Chengdu, China
5 shared publications - Boyu Ning
National Key Laboratory of Wireless Communications, University of Electronic Science and Technology of China (UESTC), Chengdu, China
2 shared publications Hua Chen
Faculty of Electrical Engineering and Computer Science, Ningbo University, Ningbo, China
2 shared publicationsJianhui Song
National Key Laboratory of Wireless Communications, University of Electronic Science and Technology of China, Chengdu, China
2 shared publications- Zhongpei Zhang
University of Electronic Science and Technology of China
2 shared publications
Latest Publications
- 17 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Wireless Communications
- Vol. 25 (15361276) , 15199-15215
Indirect and Direct Multiuser Hybrid Beamforming for Far-Field and Near-Field Communications: A Deep Learning Approach
Hybrid beamforming for extremely large-scale multiple-input multiple-output (XL-MIMO) systems is challenging in the near field because the channel depends jointly on angle and distance, and the multiuser interference (MUI) is strong. Existing deep learning methods typically follow either a decoupled design that optimizes analog beamforming without explicitly accounting for MUI, or an end-to-end (E2E) joint analog–digital optimization that can be unstable under nonconvex constant-modulus (CM), pronounced analog–digital coupling, and gradient pattern of sum-rate loss. To address both issues, we develop a complex-valued E2E framework based on a variant minimum mean square error (variant-MMSE) criterion, where the digital precoder is eliminated in closed form via Karush–Kuhn–Tucker (KKT) conditions so that analog learning is trained with a stable objective. The network employs a grouped complex-convolution sensing front-end for uplink (UL) measurements, a shared complex multi-layer perceptron (MLP) for per-user feature extraction, and a merged constant-modulus head to output the analog precoder. In the indirect mode, the network designs hybrid beamformers from estimated channel state information (CSI). In the direct mode where explicit CSI is unavailable, the network learns the sensing operator and the analog mapping from short pilots, after which additional pilots estimate the equivalent channel and enable a KKT closed-form digital precoder. Simulations show that the indirect mode approaches the performance of iterative variant-MMSE optimization with a complexity reduction proportional to the antenna number. In the direct mode, the proposed method improves spectral efficiency over sparse-recovery pipelines and recent deep learning baselines under the same pilot budget.
- 17 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Wireless Communications
- Vol. 25 (15361276) , 15199-15215
Indirect and Direct Multiuser Hybrid Beamforming for Far-Field and Near-Field Communications: A Deep Learning Approach
Hybrid beamforming for extremely large-scale multiple-input multiple-output (XL-MIMO) systems is challenging in the near field because the channel depends jointly on angle and distance, and the multiuser interference (MUI) is strong. Existing deep learning methods typically follow either a decoupled design that optimizes analog beamforming without explicitly accounting for MUI, or an end-to-end (E2E) joint analog–digital optimization that can be unstable under nonconvex constant-modulus (CM), pronounced analog–digital coupling, and gradient pattern of sum-rate loss. To address both issues, we develop a complex-valued E2E framework based on a variant minimum mean square error (variant-MMSE) criterion, where the digital precoder is eliminated in closed form via Karush–Kuhn–Tucker (KKT) conditions so that analog learning is trained with a stable objective. The network employs a grouped complex-convolution sensing front-end for uplink (UL) measurements, a shared complex multi-layer perceptron (MLP) for per-user feature extraction, and a merged constant-modulus head to output the analog precoder. In the indirect mode, the network designs hybrid beamformers from estimated channel state information (CSI). In the direct mode where explicit CSI is unavailable, the network learns the sensing operator and the analog mapping from short pilots, after which additional pilots estimate the equivalent channel and enable a KKT closed-form digital precoder. Simulations show that the indirect mode approaches the performance of iterative variant-MMSE optimization with a complexity reduction proportional to the antenna number. In the direct mode, the proposed method improves spectral efficiency over sparse-recovery pipelines and recent deep learning baselines under the same pilot budget.
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Wireless Communications Letters
- Vol. 15, 5229-5233
Distribution-Aware Generative Codebook Learning for Low-Overhead Near-Field Beam Selection
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Wireless Communications Letters
- Vol. 15, 5229-5233
Distribution-Aware Generative Codebook Learning for Low-Overhead Near-Field Beam Selection
- 15 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Wireless Communications Letters
- Vol. 15 (21622337) , 1020-1024
Multi-Beam Training for Movable Antenna Enhanced Wireless Communications
This letter investigates the effect of movable antennas on improving spectral efficiency in multi-subarray beam training of hybrid beamforming systems, forming movable subarrays (MSAs). During beam training, a joint search for both position and beam is conducted, but this increases the codebook dimension and search complexity. To solve this, we propose two low-complexity algorithms. First, an iterative one-sided search (IOSS) is developed to jointly optimize the configuration for each subarray, achieving rapid convergence. To further reduce training overhead, we also introduce a hierarchical angle-position search (HAPS), which offers performance comparable to IOSS with even lower complexity. Numerical results demonstrate that MSAs achieve gains of 39% over traditional fixed subarrays (FSAs) with exhaustive search. Our proposed IOSS algorithm secures a 42% gain, while HAPS achieves a comparable gain of 33%. These findings highlight the significant potential of MSAs.
- 15 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Wireless Communications Letters
- Vol. 15 (21622337) , 1020-1024
Multi-Beam Training for Movable Antenna Enhanced Wireless Communications
This letter investigates the effect of movable antennas on improving spectral efficiency in multi-subarray beam training of hybrid beamforming systems, forming movable subarrays (MSAs). During beam training, a joint search for both position and beam is conducted, but this increases the codebook dimension and search complexity. To solve this, we propose two low-complexity algorithms. First, an iterative one-sided search (IOSS) is developed to jointly optimize the configuration for each subarray, achieving rapid convergence. To further reduce training overhead, we also introduce a hierarchical angle-position search (HAPS), which offers performance comparable to IOSS with even lower complexity. Numerical results demonstrate that MSAs achieve gains of 39% over traditional fixed subarrays (FSAs) with exhaustive search. Our proposed IOSS algorithm secures a 42% gain, while HAPS achieves a comparable gain of 33%. These findings highlight the significant potential of MSAs.
- 30 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 74 (11) , 17321-17336
Joint Optimization of STARS-Assisted Airn–Ground ISAC System Using Deep Reinforcement Learning
Unmanned Aerial Vehicles (UAVs) are highly promising platforms with significant potential benefits for integrated sensing and communication (ISAC). However, optimizing the UAV trajectory within the ISAC system is challenging for traditional optimization methods due to numerous coupled variables. Additionally, dynamic user access and disconnection demand scalability beyond conventional methods. To address these challenges, this paper proposes the attention-based decoupling space (AT-DS) framework. It employs one agent with an attention-based actor (A2) module for STARS beamforming and power allocation and another agent with an original actor network for UAV flight control. Our simulation results indicate that: 1) AT-DS, compared to other baseline methods, can simultaneously maximize ISAC performance and accomplish the task of reaching the destination within limited energy, achieving optimal overall performance; and 2) AT-DS exhibits scalability, adapting to variations in the number of communication users in the system during ISAC services.
- 30 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 74 (11) , 17321-17336
Joint Optimization of STARS-Assisted Airn–Ground ISAC System Using Deep Reinforcement Learning
Unmanned Aerial Vehicles (UAVs) are highly promising platforms with significant potential benefits for integrated sensing and communication (ISAC). However, optimizing the UAV trajectory within the ISAC system is challenging for traditional optimization methods due to numerous coupled variables. Additionally, dynamic user access and disconnection demand scalability beyond conventional methods. To address these challenges, this paper proposes the attention-based decoupling space (AT-DS) framework. It employs one agent with an attention-based actor (A2) module for STARS beamforming and power allocation and another agent with an original actor network for UAV flight control. Our simulation results indicate that: 1) AT-DS, compared to other baseline methods, can simultaneously maximize ISAC performance and accomplish the task of reaching the destination within limited energy, achieving optimal overall performance; and 2) AT-DS exhibits scalability, adapting to variations in the number of communication users in the system during ISAC services.
- 18 December 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 73 (5) , 7368-7373
Hybrid NOMA Assisted Integrated Sensing and Communication via RIS
This paper investigates the optimization of reconfigurable intelligent surface (RIS) in an integrated sensing and communication (ISAC) system. To meet the demand of growing number of devices, power domain non-orthogonal multiple access (NOMA) is considered. However, traditional NOMA with a large number of devices is challenging due to large decoding delay and propagation error introduced by successive interference cancellation (SIC). Thus, OMA is integrated into NOMA to support more devices. We formulate a max-min problem to optimize the sensing beampattern with constraints on communication rate, through joint power allocation, active beamforming and RIS phase shift design. To solve the non-convex problem with a non-smooth objective function, we propose a low complexity alternating optimization (AO) algorithm, where a closed form expression for the intra-cluster power allocation (intra-CPA) is derived, and penalty and successive convex approximation (SCA) methods are used to optimize the beamforming and phase shift design. Simulation results show the effectiveness of the proposed algorithm in terms of improving minimum beampattern gain (MBPG) compared with other baselines. Furthermore, the trade-off between sensing and communication is analyzed and demonstrated in the simulation results.
- 18 December 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 73 (5) , 7368-7373
Hybrid NOMA Assisted Integrated Sensing and Communication via RIS
This paper investigates the optimization of reconfigurable intelligent surface (RIS) in an integrated sensing and communication (ISAC) system. To meet the demand of growing number of devices, power domain non-orthogonal multiple access (NOMA) is considered. However, traditional NOMA with a large number of devices is challenging due to large decoding delay and propagation error introduced by successive interference cancellation (SIC). Thus, OMA is integrated into NOMA to support more devices. We formulate a max-min problem to optimize the sensing beampattern with constraints on communication rate, through joint power allocation, active beamforming and RIS phase shift design. To solve the non-convex problem with a non-smooth objective function, we propose a low complexity alternating optimization (AO) algorithm, where a closed form expression for the intra-cluster power allocation (intra-CPA) is derived, and penalty and successive convex approximation (SCA) methods are used to optimize the beamforming and phase shift design. Simulation results show the effectiveness of the proposed algorithm in terms of improving minimum beampattern gain (MBPG) compared with other baselines. Furthermore, the trade-off between sensing and communication is analyzed and demonstrated in the simulation results.
- 12 January 2023
- Published by Wiley in Transactions on Emerging Telecommunications Technologies
- Vol. 34 (3)
Hierarchical multi‐beam training with the presence of the sub‐connected hybrid beamforming architecture
Millimeter‐wave techniques have been developed significantly due to their benefits of higher data rate compared with the lower band and more extensive coverage compared to the higher frequency. However, because of the equipment of a large‐scale antenna array, further low‐complexity channel estimation and precoding algorithms in the hybrid beamforming structured wireless system must be researched for real‐world implementation. This study serves that aim, particularly on multi‐beam training, with the output of this work can be regarded as baseband equivalent channel estimation. First, we present a novel ternary‐tree‐based fully‐activated hierarchical codebook based on the sub‐array division technique. The proposed hierarchical codebook surpasses its conventional binary‐tree‐based counterpart designed by using antenna deactivation in charge of the training overhead and synthesis performance of the hierarchical search. Following that, inspired by multi‐beam linear search, we develop two hierarchical linear search methods according to our proposed hierarchical codebook. Finally, simulation results show that they outperform linear search schemes regarding training complexity and the effective achievable rate.
- 12 January 2023
- Published by Wiley in Transactions on Emerging Telecommunications Technologies
- Vol. 34 (3)
Hierarchical multi‐beam training with the presence of the sub‐connected hybrid beamforming architecture
Millimeter‐wave techniques have been developed significantly due to their benefits of higher data rate compared with the lower band and more extensive coverage compared to the higher frequency. However, because of the equipment of a large‐scale antenna array, further low‐complexity channel estimation and precoding algorithms in the hybrid beamforming structured wireless system must be researched for real‐world implementation. This study serves that aim, particularly on multi‐beam training, with the output of this work can be regarded as baseband equivalent channel estimation. First, we present a novel ternary‐tree‐based fully‐activated hierarchical codebook based on the sub‐array division technique. The proposed hierarchical codebook surpasses its conventional binary‐tree‐based counterpart designed by using antenna deactivation in charge of the training overhead and synthesis performance of the hierarchical search. Following that, inspired by multi‐beam linear search, we develop two hierarchical linear search methods according to our proposed hierarchical codebook. Finally, simulation results show that they outperform linear search schemes regarding training complexity and the effective achievable rate.