Chengbo Xue
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Chengbo Xue has a total of 18 co-authors. Here are the top co-authors by number of shared publications:
- Shiwei Ren
School of Integrated Circuits and Electronics, Beijing Institute of Technology, Beijing 100081, China
7 shared publications - Rongkun Jiang
Beijing Institute of Technology
4 shared publications - Weijiang Wang
School of Integrated Circuits and Electronics, Beijing Institute of Technology, Beijing 100081, China
4 shared publications Weijiang Wang
School of Information and Electronics, Beijing Institute of Technology, Beijing, China
2 shared publications- Xiangnan Li
School of Integrated Circuits and Electronics, Beijing Institute of Technology, Beijing 100081, China
2 shared publications Yarong Ding
School of Information and Electronics, Beijing Institute of Technology, Beijing, China
2 shared publications
Latest Publications
- Open Access
- 7 May 2026
- Published by MDPI AG in Electronics
- Vol. 15 (10) , 1982
Hardware Accelerator Design for MUSIC-DOA Estimation with Bilateral Jacobi Optimization
Real-time Direction of Arrival (DOA) estimation demands high computational throughput and numerical precision. Consequently, dedicated hardware accelerators are essential. This paper presents an architecture to accelerate the MUSIC algorithm using an improved complex bilateral Jacobi eigenvalue decomposition (EVD). First, we design a triangular systolic array for Hermitian matrices. It employs an output-stationary dataflow to enable efficient parallel covariance computation. Second, we propose an enhanced EVD algorithm. It replaces CORDIC approximations with direct analytical rotations. This significantly improves numerical stability and accuracy. Third, we introduce hardware optimizations. These include unit reuse, integrated termination conditions, and pre-stored steering vectors. These measures reduce resource consumption while maintaining full functionality. Experiments on a Xilinx Virtex-6 platform validate the design. The architecture achieves a root mean square error (RMSE) below with 300 snapshots. Processing latency is only 76.17 µs. The design utilizes 10,775 LUTs and 73 DSP slices. This work balances accuracy, speed, and efficiency. It offers a practical solution for real-time, high-precision DOA systems.- 7 May 2026
- Published by MDPI AG in Electronics
- Vol. 15 (10) , 1982
Open AccessHardware Accelerator Design for MUSIC-DOA Estimation with Bilateral Jacobi Optimization
Real-time Direction of Arrival (DOA) estimation demands high computational throughput and numerical precision. Consequently, dedicated hardware accelerators are essential. This paper presents an architecture to accelerate the MUSIC algorithm using an improved complex bilateral Jacobi eigenvalue decomposition (EVD). First, we design a triangular systolic array for Hermitian matrices. It employs an output-stationary dataflow to enable efficient parallel covariance computation. Second, we propose an enhanced EVD algorithm. It replaces CORDIC approximations with direct analytical rotations. This significantly improves numerical stability and accuracy. Third, we introduce hardware optimizations. These include unit reuse, integrated termination conditions, and pre-stored steering vectors. These measures reduce resource consumption while maintaining full functionality. Experiments on a Xilinx Virtex-6 platform validate the design. The architecture achieves a root mean square error (RMSE) below with 300 snapshots. Processing latency is only 76.17 µs. The design utilizes 10,775 LUTs and 73 DSP slices. This work balances accuracy, speed, and efficiency. It offers a practical solution for real-time, high-precision DOA systems. - 22 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 25 (13) , 24812-24828
Scalable Hardware Architecture for High-Throughput Implementation of ESPRIT Algorithm
Estimating signal parameter via rotational invariance technique (ESPRIT) algorithm is a high-performance method for direction of arrival (DOA) estimation. Its most cumbersome aspect is performing eigenvalue decomposition twice, particularly with general square matrices. Current implementations lack flexibility in configuring the number of array elements, signal sources, and snapshots at the same time. To tackle these challenges, we propose a scalable hardware acceleration scheme for the QR decomposition algorithm on a Field Programmable Gate Array (FPGA). Our QR Decomposition (QRD) module leverages orientation displacement, parallel computing, and logic reuse to enhance speed and generality. Building upon this foundation, a reconfigurable design scheme with a parallel covariance matrix computation module for ESPRIT is proposed to increase calculation speed and throughput. Our ESPRIT implementation supports up to 26 array elements, 25 signal sources, and 2,048 snapshots. As evidenced by experiments, the root mean squared error (RMSE) of the QRD module and ESPRIT implementation achieve 0.02 and 0.04° respectively. On average, when the number of array elements, signal sources, and snapshots are set to 4, 1, 64 and 8, 1, 128 individually, our ESPRIT implementation completes DOA estimation in 22.04 μs and 60.50 μs, with throughput of 45,372 and 16,529 separately. In comparison with CPU implementations, our FPGA implementations offer up to 88.66% and 89.84% time savings for QRD and ESPRIT respectively.
- 22 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 25 (13) , 24812-24828
Scalable Hardware Architecture for High-Throughput Implementation of ESPRIT Algorithm
Estimating signal parameter via rotational invariance technique (ESPRIT) algorithm is a high-performance method for direction of arrival (DOA) estimation. Its most cumbersome aspect is performing eigenvalue decomposition twice, particularly with general square matrices. Current implementations lack flexibility in configuring the number of array elements, signal sources, and snapshots at the same time. To tackle these challenges, we propose a scalable hardware acceleration scheme for the QR decomposition algorithm on a Field Programmable Gate Array (FPGA). Our QR Decomposition (QRD) module leverages orientation displacement, parallel computing, and logic reuse to enhance speed and generality. Building upon this foundation, a reconfigurable design scheme with a parallel covariance matrix computation module for ESPRIT is proposed to increase calculation speed and throughput. Our ESPRIT implementation supports up to 26 array elements, 25 signal sources, and 2,048 snapshots. As evidenced by experiments, the root mean squared error (RMSE) of the QRD module and ESPRIT implementation achieve 0.02 and 0.04° respectively. On average, when the number of array elements, signal sources, and snapshots are set to 4, 1, 64 and 8, 1, 128 individually, our ESPRIT implementation completes DOA estimation in 22.04 μs and 60.50 μs, with throughput of 45,372 and 16,529 separately. In comparison with CPU implementations, our FPGA implementations offer up to 88.66% and 89.84% time savings for QRD and ESPRIT respectively.
- 29 February 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Very Large Scale Integration (VLSI) Systems
- Vol. 32 (6) , 1072-1085
A Hardware Acceleration of Maximum Likelihood Estimation Algorithm With Alternating Projection on FPGA
Maximum likelihood (ML) estimation algorithm is a special case of Bayesian estimation method in the field of direction of arrival (DOA) estimation. The ML algorithm exhibits robust capabilities with small estimation errors, making it applicable to challenging conditions, such as dealing with correlated signals and limited snapshots. However, a drawback of the ML algorithm is its multidimensional iterative nature, which results in huge computation and time consumption. To address these challenges, this article proposes a hardware implementation of the ML algorithm on a field-programmable gate array (FPGA). The design scheme presented in this article uses alternating projection (AP) to transform multidimensional operations into 1-D operations, effectively reducing the computational load. On the other hand, combined with hardware characteristics, pipeline design is used to save computing time. This design achieves a time reduction of approximately 60% in contrast to the design without the pipeline. In addition, when the number of sources is 1, the number of snapshots is 1000, the signal-to-noise ratio (SNR) is 10 dB, and the number of elements is 10, the root mean square error (RMSE) can be close to 0.02°. This exemplary performance serves as evidence that the hardware implementation in this article ensures accuracy.
- 29 February 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Very Large Scale Integration (VLSI) Systems
- Vol. 32 (6) , 1072-1085
A Hardware Acceleration of Maximum Likelihood Estimation Algorithm With Alternating Projection on FPGA
Maximum likelihood (ML) estimation algorithm is a special case of Bayesian estimation method in the field of direction of arrival (DOA) estimation. The ML algorithm exhibits robust capabilities with small estimation errors, making it applicable to challenging conditions, such as dealing with correlated signals and limited snapshots. However, a drawback of the ML algorithm is its multidimensional iterative nature, which results in huge computation and time consumption. To address these challenges, this article proposes a hardware implementation of the ML algorithm on a field-programmable gate array (FPGA). The design scheme presented in this article uses alternating projection (AP) to transform multidimensional operations into 1-D operations, effectively reducing the computational load. On the other hand, combined with hardware characteristics, pipeline design is used to save computing time. This design achieves a time reduction of approximately 60% in contrast to the design without the pipeline. In addition, when the number of sources is 1, the number of snapshots is 1000, the signal-to-noise ratio (SNR) is 10 dB, and the number of elements is 10, the root mean square error (RMSE) can be close to 0.02°. This exemplary performance serves as evidence that the hardware implementation in this article ensures accuracy.
- Open Access
- 25 April 2022
- Published by MDPI AG in Electronics
- Vol. 11 (9) , 1367
Resource-Efficient Hardware Implementation of Perspective Transformation Based on Central Projection
Perspective correction of images is an important preprocessing task in computer vision applications, which can resolve distortions caused by shooting angles, etc. This paper proposes a hardware implementation of perspective transformation based on central projection, which is simpler than the homography transformation method. In particular, it does not need to solve complex equations, thus no software assistance is required. The design can be flexibly configured with different degrees of parallelism to meet different speed requirements. Implemented on the Xilinx Zynq-7000 platform, 2893 Look-up Tables (LUTs) are required when the parallelism is one, and it can process a 20 Hz video with a resolution of 640 × 480 in real time. When the parallelism is eight, it can process 157 Hz video and requires 11,223 LUTs. The proposed design can well meet the actual needs.
- 25 April 2022
- Published by MDPI AG in Electronics
- Vol. 11 (9) , 1367
Open AccessResource-Efficient Hardware Implementation of Perspective Transformation Based on Central Projection
Perspective correction of images is an important preprocessing task in computer vision applications, which can resolve distortions caused by shooting angles, etc. This paper proposes a hardware implementation of perspective transformation based on central projection, which is simpler than the homography transformation method. In particular, it does not need to solve complex equations, thus no software assistance is required. The design can be flexibly configured with different degrees of parallelism to meet different speed requirements. Implemented on the Xilinx Zynq-7000 platform, 2893 Look-up Tables (LUTs) are required when the parallelism is one, and it can process a 20 Hz video with a resolution of 640 × 480 in real time. When the parallelism is eight, it can process 157 Hz video and requires 11,223 LUTs. The proposed design can well meet the actual needs.
- 31 March 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems I: Regular Papers
- Vol. 69 (7) , 2941-2954
Hardware Acceleration of MUSIC Algorithm for Sparse Arrays and Uniform Linear Arrays
Multiple Signal Classification (MUSIC) is a high-performance Direction of Arrival (DOA) estimation algorithm, which has been widely used. The algorithm needs to calculate the covariance matrix, eigenvalue decomposition and spectral peak search. In the paper, the hardware structure of the existing Jacobi algorithm for Hermitian matrices is proposed. On this basis, a novel hardware acceleration of the MUSIC algorithm for sparse arrays and uniform linear arrays is proposed, and the sparse array is a nested array. There are two designs, Design 1 supports 1~10 nested array elements or 1~32 uniform linear array elements, distinguishes 1~32 sources, configures snapshots 1~2048, and the maximum number of iterations and iteration accuracy of the complex Jacobi algorithm. Design 2 only needs $101.8~\mu $ s to complete a DOA estimation when the number of array elements is 8, the number of sources is 1, and the snapshots is 128. In more detail, the Root Mean Squared Error (RMSE) of both can reach 0.03°. The logic resources on the Zynq-7000 development board are 14,761 and 28,305 Look-Up Tables (LUTs), respectively.
- 31 March 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems I: Regular Papers
- Vol. 69 (7) , 2941-2954
Hardware Acceleration of MUSIC Algorithm for Sparse Arrays and Uniform Linear Arrays
Multiple Signal Classification (MUSIC) is a high-performance Direction of Arrival (DOA) estimation algorithm, which has been widely used. The algorithm needs to calculate the covariance matrix, eigenvalue decomposition and spectral peak search. In the paper, the hardware structure of the existing Jacobi algorithm for Hermitian matrices is proposed. On this basis, a novel hardware acceleration of the MUSIC algorithm for sparse arrays and uniform linear arrays is proposed, and the sparse array is a nested array. There are two designs, Design 1 supports 1~10 nested array elements or 1~32 uniform linear array elements, distinguishes 1~32 sources, configures snapshots 1~2048, and the maximum number of iterations and iteration accuracy of the complex Jacobi algorithm. Design 2 only needs $101.8~\mu $ s to complete a DOA estimation when the number of array elements is 8, the number of sources is 1, and the snapshots is 128. In more detail, the Root Mean Squared Error (RMSE) of both can reach 0.03°. The logic resources on the Zynq-7000 development board are 14,761 and 28,305 Look-Up Tables (LUTs), respectively.
- Open Access
- 12 August 2021
- Published by Springer Nature in EURASIP Journal on Advances in Signal Processing
- Vol. 2021 (1) , 1-13
DOA estimation based on sum–difference coarray with virtual array interpolation concept
The sum–difference coarray is the union of difference coarray and the sum coarray, which is capable to obtain a higher number of degrees of freedom (DOF) than the difference coarray. However, this method fails to use all information provided by the coprime array because of the existence of holes. In this paper, we introduce the virtual array interpolation into the sum–difference coarray domain. After interpolating the virtual array, we estimate the DOA by reconstructing the covariance matrix to resolve an atomic norm minimization problem in a gridless way. The proposed method is gridless and can effectively utilize the DOF of a larger virtual array. Numerical simulation results verify the effectiveness and the superior performance of the proposed algorithm.
- 12 August 2021
- Published by Springer Nature in EURASIP Journal on Advances in Signal Processing
- Vol. 2021 (1) , 1-13
Open AccessDOA estimation based on sum–difference coarray with virtual array interpolation concept
The sum–difference coarray is the union of difference coarray and the sum coarray, which is capable to obtain a higher number of degrees of freedom (DOF) than the difference coarray. However, this method fails to use all information provided by the coprime array because of the existence of holes. In this paper, we introduce the virtual array interpolation into the sum–difference coarray domain. After interpolating the virtual array, we estimate the DOA by reconstructing the covariance matrix to resolve an atomic norm minimization problem in a gridless way. The proposed method is gridless and can effectively utilize the DOF of a larger virtual array. Numerical simulation results verify the effectiveness and the superior performance of the proposed algorithm.
- 30 July 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cognitive Communications and Networking
- Vol. 8 (1) , 136-154
Deep Learning-Aided Signal Detection for Two-Stage Index Modulated Universal Filtered Multi-Carrier Systems
The growing interests in Internet of Underwater Things (IoUT) put forward more demands for underwater acoustic (UWA) communication technologies. However, the conventional orthogonal frequency division multiplexing (OFDM) technology underperforms in terms of the resistance to carrier frequency offset (CFO), symbol time offset (STO), and applicability to multiple application scenarios. To remedy the deficiencies, this paper introduces the emerging universal filtered multi-carrier (UFMC) technology into UWA communications and devises a two-stage index modulated UFMC (TSIM-UFMC) system for performance improvement. In addition, a deep learning-aided signal detector termed as TSIMNet is proposed for the TSIM-UFMC system over multipath UWA channels. Specifically, at the transmitter, the TSIM-UFMC system adopts two-stage index modulation to characterize the different position information of data, pilot, and inactive subcarriers in each subblock for ameliorating the bit error rate (BER) performance under channel interferences. At the receiver, the TSIMNet detector exploits configurable deep neural networks to encapsulate the index demodulation, channel estimation, and constellation demodulation for optimizing the system structure. Furthermore, TSIMNet can be deployed to detect the received signal online for the TSIM-UFMC system after trained by offline data. Numerical results reveal that the proposed TSIM-UFMC system exhibits achievable advantages in combating CFO and STO, and the TSIMNet detector actualizes a near-optimal performance on BER with preferable computational complexity.
- 30 July 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cognitive Communications and Networking
- Vol. 8 (1) , 136-154
Deep Learning-Aided Signal Detection for Two-Stage Index Modulated Universal Filtered Multi-Carrier Systems
The growing interests in Internet of Underwater Things (IoUT) put forward more demands for underwater acoustic (UWA) communication technologies. However, the conventional orthogonal frequency division multiplexing (OFDM) technology underperforms in terms of the resistance to carrier frequency offset (CFO), symbol time offset (STO), and applicability to multiple application scenarios. To remedy the deficiencies, this paper introduces the emerging universal filtered multi-carrier (UFMC) technology into UWA communications and devises a two-stage index modulated UFMC (TSIM-UFMC) system for performance improvement. In addition, a deep learning-aided signal detector termed as TSIMNet is proposed for the TSIM-UFMC system over multipath UWA channels. Specifically, at the transmitter, the TSIM-UFMC system adopts two-stage index modulation to characterize the different position information of data, pilot, and inactive subcarriers in each subblock for ameliorating the bit error rate (BER) performance under channel interferences. At the receiver, the TSIMNet detector exploits configurable deep neural networks to encapsulate the index demodulation, channel estimation, and constellation demodulation for optimizing the system structure. Furthermore, TSIMNet can be deployed to detect the received signal online for the TSIM-UFMC system after trained by offline data. Numerical results reveal that the proposed TSIM-UFMC system exhibits achievable advantages in combating CFO and STO, and the TSIMNet detector actualizes a near-optimal performance on BER with preferable computational complexity.
- Open Access
- 11 May 2021
- Published by Springer Nature in Research Square
DOA Estimation Based on Sum-Difference Coarray with Virtual Array Interpolation Concept
The sum-difference coarray is the union of difference coarray and the sum coarray, which is capable to obtain a higher number of degrees of freedom (DOF) than the difference coarray. However, this method fails to use all information provided by the coprime array because of the existence of holes. In this paper, we introduce the virtual array interpolation into the sum-difference coarray domain. After interpolating the virtual array, we estimate the DOA by reconstructing the covariance matrix to resolve an atomic norm minimization problem in a gridless way. The proposed method is gridless and can effectively utilize the DOF of a larger virtual array. Numerical simulation results verify the effectiveness and the superior performance of the proposed algorithm.
- 11 May 2021
- Published by Springer Nature in Research Square
Open AccessDOA Estimation Based on Sum-Difference Coarray with Virtual Array Interpolation Concept
The sum-difference coarray is the union of difference coarray and the sum coarray, which is capable to obtain a higher number of degrees of freedom (DOF) than the difference coarray. However, this method fails to use all information provided by the coprime array because of the existence of holes. In this paper, we introduce the virtual array interpolation into the sum-difference coarray domain. After interpolating the virtual array, we estimate the DOA by reconstructing the covariance matrix to resolve an atomic norm minimization problem in a gridless way. The proposed method is gridless and can effectively utilize the DOF of a larger virtual array. Numerical simulation results verify the effectiveness and the superior performance of the proposed algorithm.