Qi Wu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Qi Wu has a total of 61 co-authors. Here are the top co-authors by number of shared publications:
- Haiming Wang
Southeast University
41 shared publications - Wei Hong
Southeast University
34 shared publications Chen Yu
School of Information Science and Engineering and the State Key Laboratory of Millimeter Waves, Southeast University, Nanjing, China
21 shared publications- Weiqi Chen
The Chinese University of Hong Kong
11 shared publications - Jiexi Yin
Karlsruhe Institute of Technology
10 shared publications - Guangyi Lu
Southeast University
9 shared publications
Latest Publications
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
- Vol. 45 (11) , 5108-5121
Fully Automated Schematic and Layout Design of Millimeter-Wave CMOS Power Amplifiers Using Prior Knowledge-Guided Decomposition and Machine Learning-Assisted Optimization
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
- Vol. 45 (11) , 5108-5121
Fully Automated Schematic and Layout Design of Millimeter-Wave CMOS Power Amplifiers Using Prior Knowledge-Guided Decomposition and Machine Learning-Assisted Optimization
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
- Vol. 45 (11) , 5094-5107
FARFLAG: Fully Automatic Radio Frequency Layout Generation and Its Application to Design of CMOS Millimeter-Wave Low-Noise Amplifiers
The design of millimeter-wave (mmWave) integrated circuits (ICs) is traditionally a slow process that requires extensive iterations and computationally intensive full-wave electromagnetic (EM) simulations. This paper presents a fully automatic radio-frequency layout generation (FARFLAG) method for the end-to-end synthesis of mmWave ICs, from device generation to final layout. The proposed method utilizes a mixture of machine-learning-assisted optimization, lookup tables, and fixed parameters for device generation. We introduce a hybrid placement and routing algorithm tailored for mmWave ICs, which is enhanced with a two-restart mechanism to improve placement robustness and exploration. The synthesis is driven by an improved Bayesian optimization approach with constrained multi-acquisition functions to navigate the design space efficiently. We demonstrate the effectiveness of the proposed framework by synthesizing two mmWave low-noise amplifiers. By integrating high-fidelity EM simulations, the FARFLAG method significantly reduces the search space compared to conventional methods, enabling the rapid synthesis of complex mmWave ICs with superior performance.
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
- Vol. 45 (11) , 5094-5107
FARFLAG: Fully Automatic Radio Frequency Layout Generation and Its Application to Design of CMOS Millimeter-Wave Low-Noise Amplifiers
The design of millimeter-wave (mmWave) integrated circuits (ICs) is traditionally a slow process that requires extensive iterations and computationally intensive full-wave electromagnetic (EM) simulations. This paper presents a fully automatic radio-frequency layout generation (FARFLAG) method for the end-to-end synthesis of mmWave ICs, from device generation to final layout. The proposed method utilizes a mixture of machine-learning-assisted optimization, lookup tables, and fixed parameters for device generation. We introduce a hybrid placement and routing algorithm tailored for mmWave ICs, which is enhanced with a two-restart mechanism to improve placement robustness and exploration. The synthesis is driven by an improved Bayesian optimization approach with constrained multi-acquisition functions to navigate the design space efficiently. We demonstrate the effectiveness of the proposed framework by synthesizing two mmWave low-noise amplifiers. By integrating high-fidelity EM simulations, the FARFLAG method significantly reduces the search space compared to conventional methods, enabling the rapid synthesis of complex mmWave ICs with superior performance.
- 12 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Antennas and Propagation
- Vol. PP (0018926X) , 1
Machine Learning-Assisted Circuit-Network-Field Co-Optimization for Metal-Bezel Smartphone Antenna Design
To meet advanced performance requirements, modern smartphone antennas frequently adopt circuit-network-field (CNF) co-design, which introduces challenges such as strong variable coupling, high sensitivity, and vast search spaces. Traditional machine learning-assisted optimization (MLAO) methods, typically focused on geometric structures, exhibit limited capability in addressing these complexities. This paper proposes a machine learning-assisted circuit-network-field co-optimization (MLA-CNFCO) method incorporating microwave network analysis. First, logarithmic sampling is applied to circuit variables to reshape the sample space, alleviating issues related to nonlinear impedance and high sensitivity. A proportion-modeling approach is then introduced to enhance the prediction accuracy of radiation pattern surrogate models. To accelerate the process, a microwave network analytical solution (MNAS) is employed to calculate reflection coefficients directly, enabling the rapid identification of superior initial solutions. These coefficients are integrated as constraints, significantly narrowing the search space. Finally, the co-design problem is effectively decoupled by embedding microwave network-based circuit optimization within potential-based geometry optimization. The effectiveness of the MLA-CNFCO framework is validated through the design, fabrication, and measurement of a high-performance smartphone antenna.
- 12 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Antennas and Propagation
- Vol. PP (0018926X) , 1
Machine Learning-Assisted Circuit-Network-Field Co-Optimization for Metal-Bezel Smartphone Antenna Design
To meet advanced performance requirements, modern smartphone antennas frequently adopt circuit-network-field (CNF) co-design, which introduces challenges such as strong variable coupling, high sensitivity, and vast search spaces. Traditional machine learning-assisted optimization (MLAO) methods, typically focused on geometric structures, exhibit limited capability in addressing these complexities. This paper proposes a machine learning-assisted circuit-network-field co-optimization (MLA-CNFCO) method incorporating microwave network analysis. First, logarithmic sampling is applied to circuit variables to reshape the sample space, alleviating issues related to nonlinear impedance and high sensitivity. A proportion-modeling approach is then introduced to enhance the prediction accuracy of radiation pattern surrogate models. To accelerate the process, a microwave network analytical solution (MNAS) is employed to calculate reflection coefficients directly, enabling the rapid identification of superior initial solutions. These coefficients are integrated as constraints, significantly narrowing the search space. Finally, the co-design problem is effectively decoupled by embedding microwave network-based circuit optimization within potential-based geometry optimization. The effectiveness of the MLA-CNFCO framework is validated through the design, fabrication, and measurement of a high-performance smartphone antenna.
- 12 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems I: Regular Papers
- Vol. PP (15498328) , 1-14
Full-Chain Automated ESD Power Clamp Design: From Specification to Silicon Validation With Diverse Performance Priorities
Electrostatic discharge (ESD) power clamps are critical for ensuring the reliability of modern integrated circuits (ICs). This paper presents an intelligent and automated design methodology that features application-driven topology selection tailored to diverse performance priorities and full-chain automation, from design specifications to silicon validation. The proposed approach begins with selecting representative circuit topologies, each tailored to specific application scenarios with distinct performance priorities. These topologies form a set of curated candidates defined by formalized design rules and established ESD protection principles. Subsequently, an optimization algorithm refines the dimensions of the device according to given specifications, while a professional floorplan and routing ensure high-current handling layout implementation. Compared with conventional manual design, the proposed methodology achieves superior design flexibility, comprehensive automation, and optimized performance, delivering balanced trade-offs among performance, power, and area (PPA). Post-layout simulations and silicon measurements in a 40-nm CMOS process validate the methodology, demonstrating the effectiveness of the intelligent algorithm in ESD protection design.
- 12 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems I: Regular Papers
- Vol. PP (15498328) , 1-14
Full-Chain Automated ESD Power Clamp Design: From Specification to Silicon Validation With Diverse Performance Priorities
Electrostatic discharge (ESD) power clamps are critical for ensuring the reliability of modern integrated circuits (ICs). This paper presents an intelligent and automated design methodology that features application-driven topology selection tailored to diverse performance priorities and full-chain automation, from design specifications to silicon validation. The proposed approach begins with selecting representative circuit topologies, each tailored to specific application scenarios with distinct performance priorities. These topologies form a set of curated candidates defined by formalized design rules and established ESD protection principles. Subsequently, an optimization algorithm refines the dimensions of the device according to given specifications, while a professional floorplan and routing ensure high-current handling layout implementation. Compared with conventional manual design, the proposed methodology achieves superior design flexibility, comprehensive automation, and optimized performance, delivering balanced trade-offs among performance, power, and area (PPA). Post-layout simulations and silicon measurements in a 40-nm CMOS process validate the methodology, demonstrating the effectiveness of the intelligent algorithm in ESD protection design.
- 4 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Antennas and Wireless Propagation Letters
- Vol. PP (15361225) , 1-5
Machine Learning-Assisted Synthesis of Heterogeneous Arrays via Active Geometric Design
Mutual coupling often leads to discrepancies between intended and synthesized array patterns due to the distortion of active element patterns (AEPs). In array synthesis, existing machine learning methods are predominantly confined to predicting AEPs for fixed structures, imposing a passive constraint that limits optimization flexibility. A holistic framework is introduced for the active geometric design of heterogeneous arrays, in which the geometric parameters of individual elements are treated as degrees of freedom alongside traditional ones. To manage the complexity associated with mutual coupling, a variational autoencoder-based surrogate model is developed to efficiently capture the mapping between diverse element structures and their radiation patterns in a latent space. Numerical comparisons indicate a 9.8× speedup in prediction over traditional methods. Furthermore, under rigorous uniform amplitude and 1-bit phase constraints, sidelobe suppression improvements of over 8 dB are achieved at scanning angles compared to homogeneous baselines, validating the efficacy of incorporating element-level geometric design into the global optimization process.
- 4 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Antennas and Wireless Propagation Letters
- Vol. PP (15361225) , 1-5
Machine Learning-Assisted Synthesis of Heterogeneous Arrays via Active Geometric Design
Mutual coupling often leads to discrepancies between intended and synthesized array patterns due to the distortion of active element patterns (AEPs). In array synthesis, existing machine learning methods are predominantly confined to predicting AEPs for fixed structures, imposing a passive constraint that limits optimization flexibility. A holistic framework is introduced for the active geometric design of heterogeneous arrays, in which the geometric parameters of individual elements are treated as degrees of freedom alongside traditional ones. To manage the complexity associated with mutual coupling, a variational autoencoder-based surrogate model is developed to efficiently capture the mapping between diverse element structures and their radiation patterns in a latent space. Numerical comparisons indicate a 9.8× speedup in prediction over traditional methods. Furthermore, under rigorous uniform amplitude and 1-bit phase constraints, sidelobe suppression improvements of over 8 dB are achieved at scanning angles compared to homogeneous baselines, validating the efficacy of incorporating element-level geometric design into the global optimization process.
- 10 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Antennas and Wireless Propagation Letters
- Vol. 25 (4) , 1742-1746
Sequential Hybrid Synthesis for Thinned Prephased Electronically Steered Phased Array Using Low-Resolution Phase Shifters and RF Switches
To reduce the complexity of large-scale electronically steered phased arrays (ESPAs) that use high-resolution attenuators and phase shifters, this letter proposes a sequential hybrid method. The method can synthesize a thinned prephased ESPA (TPESPA) that uses low-resolution phase shifters and 1-bit RF switches, while maintaining low sidelobe levels (SLLs) and high gains. The approach sequentially synthesizes the high-dimensional 0-1 amplitude vector, the prephase vector, and the one-dimensional reference phase. It thereby overcomes challenges such as slow computation, the unstable results of existing 0-1 amplitude algorithms, and the large storage demands of existing phasing methods. Simulations and measurements show that the synthesized $32\times 32$-unit TPESPA controlled by 2-bit phase shifters and 1-bit RF switches demonstrates SLLs between $-20.0$ dB and $-18.0$ dB, with an average increase in active gain of 1.1dB for a sparsity of 26.6% compared to a Taylor array using 6-bit phase-amplitude control.
- 10 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Antennas and Wireless Propagation Letters
- Vol. 25 (4) , 1742-1746
Sequential Hybrid Synthesis for Thinned Prephased Electronically Steered Phased Array Using Low-Resolution Phase Shifters and RF Switches
To reduce the complexity of large-scale electronically steered phased arrays (ESPAs) that use high-resolution attenuators and phase shifters, this letter proposes a sequential hybrid method. The method can synthesize a thinned prephased ESPA (TPESPA) that uses low-resolution phase shifters and 1-bit RF switches, while maintaining low sidelobe levels (SLLs) and high gains. The approach sequentially synthesizes the high-dimensional 0-1 amplitude vector, the prephase vector, and the one-dimensional reference phase. It thereby overcomes challenges such as slow computation, the unstable results of existing 0-1 amplitude algorithms, and the large storage demands of existing phasing methods. Simulations and measurements show that the synthesized $32\times 32$-unit TPESPA controlled by 2-bit phase shifters and 1-bit RF switches demonstrates SLLs between $-20.0$ dB and $-18.0$ dB, with an average increase in active gain of 1.1dB for a sparsity of 26.6% compared to a Taylor array using 6-bit phase-amplitude control.
- Open Access
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Electromagnetics, Antennas and Propagation
- Vol. 2, 191-212
Graphic Antenna: Reconciling Performance, Freedom, and Efficiency Using Hybrid Knowledge-Guided and Data-Driven Technology
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Electromagnetics, Antennas and Propagation
- Vol. 2, 191-212
Open AccessGraphic Antenna: Reconciling Performance, Freedom, and Efficiency Using Hybrid Knowledge-Guided and Data-Driven Technology