Jiahao Lu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Jiahao Lu has a total of 46 co-authors. Here are the top co-authors by number of shared publications:
- Dongsheng Liu
School of Integrated Circuits, Huazhong University of Science and Technology, Wuhan, China
12 shared publications - Aobo Li
School of Integrated Circuits, Huazhong University of Science and Technology, Wuhan, China
6 shared publications - Ang Hu
School of Integrated Circuits, Huazhong University of Science and Technology, Wuhan, China
4 shared publications - Jiaming Zhang
School of Integrated Circuits, Huazhong University of Science and Technology, Wuhan, China
4 shared publications - Tianze Huang
School of Integrated Circuits, Huazhong University of Science and Technology, Wuhan, China
4 shared publications - Cong Zhang
School of Optical and Electronic Information, Huazhong University of Science and Technology, Wuhan, China
3 shared publications
Latest Publications
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems II: Express Briefs
- Vol. 73 (11) , 1119-1123
A 20-Gb/s Digital CDR With Complementary Phase Interpolator Control for Low-Jitter Transfer
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems II: Express Briefs
- Vol. 73 (11) , 1119-1123
A 20-Gb/s Digital CDR With Complementary Phase Interpolator Control for Low-Jitter Transfer
- 23 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems II: Express Briefs
- Vol. 72 (12) , 2017-2021
A 1–8 b Reconfigurable 64 Mb STT-MRAM Near-Memory Computing Macro for Energy-Efficient AI-Edge Devices
This brief presents an nonvolatile near-memory-compute (nvNMC) macro for edge artificial intelligence (AI) de-vices. The proposed macro is interconnected in groups of 4 sub-macros per line, enabling sub-macro level task scheduling through a customized micro-instruction set. The sub-macro ar-chitecture employs dual-clock domains and hierarchical pipelin-ing to enhance the overall chip’s data throughput, computility, and energy efficiency. The multiply-and-accumulate (MAC) unit achieves 1-8b reconfigurable input/weight precision via bit-serial computing and incorporates a data bypass path to boost perfor-mance and efficiency during 1b computations. The proposed macro integrates a 64Mb spin-transfer torque magnetic random access memory (STT-MRAM) in a 51.29mm2 die area using 28nm CMOS technology. It presents that a peak computility of 30.81TOPS and energy efficiency of 316 TOPS/W at 1b precision.
- 23 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems II: Express Briefs
- Vol. 72 (12) , 2017-2021
A 1–8 b Reconfigurable 64 Mb STT-MRAM Near-Memory Computing Macro for Energy-Efficient AI-Edge Devices
This brief presents an nonvolatile near-memory-compute (nvNMC) macro for edge artificial intelligence (AI) de-vices. The proposed macro is interconnected in groups of 4 sub-macros per line, enabling sub-macro level task scheduling through a customized micro-instruction set. The sub-macro ar-chitecture employs dual-clock domains and hierarchical pipelin-ing to enhance the overall chip’s data throughput, computility, and energy efficiency. The multiply-and-accumulate (MAC) unit achieves 1-8b reconfigurable input/weight precision via bit-serial computing and incorporates a data bypass path to boost perfor-mance and efficiency during 1b computations. The proposed macro integrates a 64Mb spin-transfer torque magnetic random access memory (STT-MRAM) in a 51.29mm2 die area using 28nm CMOS technology. It presents that a peak computility of 30.81TOPS and energy efficiency of 316 TOPS/W at 1b precision.
- 9 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems I: Regular Papers
- Vol. 73 (4) , 2595-2607
A Lightweight and Efficient Post Quantum Crypto-Processor for ML-DSA
Powerful quantum computers have raised people’s serious concerns on traditional cryptography, leading to great efforts on searching post-quantum alternatives. The screening winner, module lattice-based digital signature algorithm (ML-DSA), is recommended due to good performance and efficiency. However, its hardware implementation exhibits heavy resource burden on logic and memory. This paper presents a lightweight and efficient ML-DSA processor towards resource-constrained applications. The proposed Toeplitz multiplier reduces memory demand with performing specific computations in regular domain representation. Lightweight SHA-3 and samplers alleviate complex connections of large registers through introducing align structure. Diverse modular operations are optimized with shift-and-addition method for decreasing DSP consumption. Moreover, we design a crossover memory scheduling with processing rearrangement to further lower storage requirement. It also enables efficient parallel executions without memory conflict. The ML-DSA processor consumes 15370/7843/7.5/2 LUTs/FFs/BRAMs/DSPs on Artix-7 device and 0.08 mm2 with 27.5 kB SRAM on TSMC 28nm technology. It achieves 51% resource reduction and $4.12\times $ area-time product (ATP) improvement over similar state-of-the-art lightweight implementation.
- 9 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems I: Regular Papers
- Vol. 73 (4) , 2595-2607
A Lightweight and Efficient Post Quantum Crypto-Processor for ML-DSA
Powerful quantum computers have raised people’s serious concerns on traditional cryptography, leading to great efforts on searching post-quantum alternatives. The screening winner, module lattice-based digital signature algorithm (ML-DSA), is recommended due to good performance and efficiency. However, its hardware implementation exhibits heavy resource burden on logic and memory. This paper presents a lightweight and efficient ML-DSA processor towards resource-constrained applications. The proposed Toeplitz multiplier reduces memory demand with performing specific computations in regular domain representation. Lightweight SHA-3 and samplers alleviate complex connections of large registers through introducing align structure. Diverse modular operations are optimized with shift-and-addition method for decreasing DSP consumption. Moreover, we design a crossover memory scheduling with processing rearrangement to further lower storage requirement. It also enables efficient parallel executions without memory conflict. The ML-DSA processor consumes 15370/7843/7.5/2 LUTs/FFs/BRAMs/DSPs on Artix-7 device and 0.08 mm2 with 27.5 kB SRAM on TSMC 28nm technology. It achieves 51% resource reduction and $4.12\times $ area-time product (ATP) improvement over similar state-of-the-art lightweight implementation.
- 26 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems I: Regular Papers
- Vol. 72 (5) , 2252-2262
An Efficient and Reconfigurable Post-Quantum Crypto-Processor for SPHINCS+
SPHINCS+ is the sole hash-based digital signature scheme among the selected post-quantum cryptography (PQC) in 2022. This algorithm possesses the ability to resist attacks from both classical and quantum computers. Due to the extensive computations and different data widths for various parameters, its hardware implementation faces the weakness of long operation time, large area requirement, and low flexibility. This paper presents an efficient and reconfigurable SPHINCS+ processor. The proposed on-the-fly WOTS+ public key generation scheme with unified chain address generator accelerated the most time-consuming operations. This optimization achieves efficient resource utilization. A security switch mechanism resolves the bit misalignment among different data widths with resource reduction. Finally, we introduce a grouped subtree and segmented signature streaming scheme. They reduce the memory to 16k bytes. The processor consumes 29410 LUTs, 14090 FFs, 4 BRAMs on Artix-7 FPGA and achieves 1.04×/2.41× ATPs (area-time-product) optimizations in Sign/ Verify with the advantage of supporting all security levels of SPHINCS+.
- 26 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems I: Regular Papers
- Vol. 72 (5) , 2252-2262
An Efficient and Reconfigurable Post-Quantum Crypto-Processor for SPHINCS+
SPHINCS+ is the sole hash-based digital signature scheme among the selected post-quantum cryptography (PQC) in 2022. This algorithm possesses the ability to resist attacks from both classical and quantum computers. Due to the extensive computations and different data widths for various parameters, its hardware implementation faces the weakness of long operation time, large area requirement, and low flexibility. This paper presents an efficient and reconfigurable SPHINCS+ processor. The proposed on-the-fly WOTS+ public key generation scheme with unified chain address generator accelerated the most time-consuming operations. This optimization achieves efficient resource utilization. A security switch mechanism resolves the bit misalignment among different data widths with resource reduction. Finally, we introduce a grouped subtree and segmented signature streaming scheme. They reduce the memory to 16k bytes. The processor consumes 29410 LUTs, 14090 FFs, 4 BRAMs on Artix-7 FPGA and achieves 1.04×/2.41× ATPs (area-time-product) optimizations in Sign/ Verify with the advantage of supporting all security levels of SPHINCS+.
- 7 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Solid-State Circuits
- Vol. 59 (9) , 2774-2787
A Low-Phase-Noise Wide-Tuning-Range Mode-Switching Oscillator Using Multi-Magnetic-Coupling and Active-Source-Degenerating Techniques
In this article, a mode-switching oscillator using multi-magnetic-coupling and active-source-degenerating techniques is proposed to achieve wide frequency tuning and low phase noise. The multi-magnetic coupled resonator can realize two switchable resonant frequencies and perform active source degeneration. Two main coils are developed to switch between the two resonant modes, thereby doubling the frequency tuning range (TR) without degrading the phase noise performance. The effective noise power from the active device is suppressed by an additional coupling between the primary and secondary coils via active source degeneration. Moreover, the strong coupling between the primary and secondary coils can boost the equivalent quality factor of the inductance part of the tank at both switchable modes. Thus, both the reduction of phase noise and the extension of the TR are achieved. Fabricated in a 40-nm CMOS process, the proposed oscillator exhibits frequency tuning from 2.88 to 5.87 GHz while consuming 20.5 to 27.9 mW power dissipation. The phase noise at 2.88 GHz oscillation frequency is -135.8 dBc/Hz at 1-MHz offset, corresponding to a figure-of-merit (FoM) of 191.5 dBc/Hz and FoMT of 208.2 dBc/Hz. The 1/ $\mathbf {f}^{3}$ phase noise corner is 68–234 kHz over the 68.3% TR.
- 7 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Solid-State Circuits
- Vol. 59 (9) , 2774-2787
A Low-Phase-Noise Wide-Tuning-Range Mode-Switching Oscillator Using Multi-Magnetic-Coupling and Active-Source-Degenerating Techniques
In this article, a mode-switching oscillator using multi-magnetic-coupling and active-source-degenerating techniques is proposed to achieve wide frequency tuning and low phase noise. The multi-magnetic coupled resonator can realize two switchable resonant frequencies and perform active source degeneration. Two main coils are developed to switch between the two resonant modes, thereby doubling the frequency tuning range (TR) without degrading the phase noise performance. The effective noise power from the active device is suppressed by an additional coupling between the primary and secondary coils via active source degeneration. Moreover, the strong coupling between the primary and secondary coils can boost the equivalent quality factor of the inductance part of the tank at both switchable modes. Thus, both the reduction of phase noise and the extension of the TR are achieved. Fabricated in a 40-nm CMOS process, the proposed oscillator exhibits frequency tuning from 2.88 to 5.87 GHz while consuming 20.5 to 27.9 mW power dissipation. The phase noise at 2.88 GHz oscillation frequency is -135.8 dBc/Hz at 1-MHz offset, corresponding to a figure-of-merit (FoM) of 191.5 dBc/Hz and FoMT of 208.2 dBc/Hz. The 1/ $\mathbf {f}^{3}$ phase noise corner is 68–234 kHz over the 68.3% TR.
- 28 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems II: Express Briefs
- Vol. 71 (9) , 4286-4290
Super-K: A Superscalar CRYSTALS-KYBER Processor Based on Efficient Arithmetic Array
Security systems based on traditional cryptography mechanisms are at risk of being cracked by quantum computers in the future. CRYSTALS-KYBER (Kyber) as the NIST finalized lattice-based post-quantum cryptography (PQC) algorithm, will be widely used in public-key encryption scenarios. Although quantum-difficulty assumptions based on lattice math problems make Kyber quantum-resistant, how to effectively implement it into systems with different security needs remains a challenge. This work presents a dual-issue superscalar Kyber processor (Super-K) that supports customized RISC-V instruction-set architecture (ISA) and implements the key encapsulation mechanism (KEM) flexibly and efficiently. A reconfigurable polynomial arithmetic unit (PAU) is designed, which optimizes the compress/decompress process, and accelerates polynomial operations efficiently by optimal parallelism. The pipelining scheduling technique is used in Super-K to improve instruction level parallelism and reduce time consumption. Super-K is implemented on UltraScale+ FPGA platform and evaluated under SMIC 40nm technology, which achieves the fastest computational speed with the lowest power consumption and 1.4×/8.2× improvement in FPGA/ASIC AT product efficiency.
- 28 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems II: Express Briefs
- Vol. 71 (9) , 4286-4290
Super-K: A Superscalar CRYSTALS-KYBER Processor Based on Efficient Arithmetic Array
Security systems based on traditional cryptography mechanisms are at risk of being cracked by quantum computers in the future. CRYSTALS-KYBER (Kyber) as the NIST finalized lattice-based post-quantum cryptography (PQC) algorithm, will be widely used in public-key encryption scenarios. Although quantum-difficulty assumptions based on lattice math problems make Kyber quantum-resistant, how to effectively implement it into systems with different security needs remains a challenge. This work presents a dual-issue superscalar Kyber processor (Super-K) that supports customized RISC-V instruction-set architecture (ISA) and implements the key encapsulation mechanism (KEM) flexibly and efficiently. A reconfigurable polynomial arithmetic unit (PAU) is designed, which optimizes the compress/decompress process, and accelerates polynomial operations efficiently by optimal parallelism. The pipelining scheduling technique is used in Super-K to improve instruction level parallelism and reduce time consumption. Super-K is implemented on UltraScale+ FPGA platform and evaluated under SMIC 40nm technology, which achieves the fastest computational speed with the lowest power consumption and 1.4×/8.2× improvement in FPGA/ASIC AT product efficiency.
- 18 December 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 44, 1-8
Hybrid Hardware-Software Architecture for Quantum Secure IoT Embedded Systems
The Internet of Things (IoT) communication prevents information from being decrypted or tampered with by unauthorized third parties. Post-quantum cryptography ensures that the information exchange between IoT devices can resist potential quantum computer attacks, thereby enhancing the quantum security of the IoT. This paper presents a quantum-secure hardware-software architecture for embedded systems. An FPGA-based post-quantum cryptographic Kyber accelerator and an STM32 microcontroller with AES encryption are integrated into the system. Customized hardware structures such as hash and modular arithmetic unit provide faster calculation speed. The command exchange protocol bridges embedded software and FPGA hardware. The server handles key generation and decryption, while the client manages encryption. AES-256 encryption is utilized to ensure the confidentiality and integrity of the data and keys, with the Kyber algorithm encrypting the session keys. Compared to previous works, this implementation demonstrates better resource utilization as well as enhanced performance in key generation, encryption, and decryption. Hardware implementation operates at 100MHz on the Artix-7 FPGA platform, consuming 20,782 LUTs, 5,077 FFs, 7.5 BRAMs, and 8 DSPs. The architecture also achieves up to 1.5× speed improvements in encryption and similar enhancements in decryption across various security levels.
- 18 December 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 44, 1-8
Hybrid Hardware-Software Architecture for Quantum Secure IoT Embedded Systems
The Internet of Things (IoT) communication prevents information from being decrypted or tampered with by unauthorized third parties. Post-quantum cryptography ensures that the information exchange between IoT devices can resist potential quantum computer attacks, thereby enhancing the quantum security of the IoT. This paper presents a quantum-secure hardware-software architecture for embedded systems. An FPGA-based post-quantum cryptographic Kyber accelerator and an STM32 microcontroller with AES encryption are integrated into the system. Customized hardware structures such as hash and modular arithmetic unit provide faster calculation speed. The command exchange protocol bridges embedded software and FPGA hardware. The server handles key generation and decryption, while the client manages encryption. AES-256 encryption is utilized to ensure the confidentiality and integrity of the data and keys, with the Kyber algorithm encrypting the session keys. Compared to previous works, this implementation demonstrates better resource utilization as well as enhanced performance in key generation, encryption, and decryption. Hardware implementation operates at 100MHz on the Artix-7 FPGA platform, consuming 20,782 LUTs, 5,077 FFs, 7.5 BRAMs, and 8 DSPs. The architecture also achieves up to 1.5× speed improvements in encryption and similar enhancements in decryption across various security levels.
- 11 August 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems II: Express Briefs
- Vol. 71 (1) , 435-439
A High Speed Post-Quantum Crypto-Processor for Crystals-Dilithium
Crystals-Dilithium is a post-quantum cryptography digital signature scheme with quantum attack resistance. In this brief, a high speed processor towards Dilithium is represented. An architecture based on multiple instruction parallel execution is proposed to reduce operation cycles. The pseudo random number extension generator is designed with byte absorption precision to adapt to practical application scenarios. In addition, a three-stage pipelined modular multiplier is optimized based on Karatsuba algorithm, increasing the maximum operating frequency of the system. On the Zynq UltraScale+ FPGA, our processor is implemented consuming 32035 LUTs and 9707 FFs at 300MHz. Compared with the related works, it has the highest operating frequency and the shortest execution time.
- 11 August 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems II: Express Briefs
- Vol. 71 (1) , 435-439
A High Speed Post-Quantum Crypto-Processor for Crystals-Dilithium
Crystals-Dilithium is a post-quantum cryptography digital signature scheme with quantum attack resistance. In this brief, a high speed processor towards Dilithium is represented. An architecture based on multiple instruction parallel execution is proposed to reduce operation cycles. The pseudo random number extension generator is designed with byte absorption precision to adapt to practical application scenarios. In addition, a three-stage pipelined modular multiplier is optimized based on Karatsuba algorithm, increasing the maximum operating frequency of the system. On the Zynq UltraScale+ FPGA, our processor is implemented consuming 32035 LUTs and 9707 FFs at 300MHz. Compared with the related works, it has the highest operating frequency and the shortest execution time.
- 25 August 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Informatics
- Vol. 19 (2) , 1874-1883
A Flexible and High-Performance Lattice-Based Post-Quantum Crypto Secure Coprocessor
Progress of quantum computing technology seriously threaten the industrial information security based on traditional public-key cryptosystem. Thus, the cryptosystem with anti-quantum attack characteristics is gradually becoming a significant research in the security field. In this article, a flexible and high-performance secure coprocessor is designed for security in industrial processes, which can execute the post-quantum cryptographic algorithm Saber efficiently. Custom instruction set and arithmetic accelerators are proposed to effectively optimize the flexibility of system architecture, and improve the performance of calculation. The hardware implementation results show that the maximum operating frequency of the coprocessor can reach 345 MHz. Compared with related state-of-the-art works, it achieves the highest operating frequency on the same Xilinx UltraScale+ FPGA platform, performing the encryption and decryption operations within 13.5 and 15.4 μ s, respectively. Meanwhile, this article achieves 1.7/3.1/5.9× area-time product improvements in look-up table flip-flop block memory storage with good flexibility.
- 25 August 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Informatics
- Vol. 19 (2) , 1874-1883
A Flexible and High-Performance Lattice-Based Post-Quantum Crypto Secure Coprocessor
Progress of quantum computing technology seriously threaten the industrial information security based on traditional public-key cryptosystem. Thus, the cryptosystem with anti-quantum attack characteristics is gradually becoming a significant research in the security field. In this article, a flexible and high-performance secure coprocessor is designed for security in industrial processes, which can execute the post-quantum cryptographic algorithm Saber efficiently. Custom instruction set and arithmetic accelerators are proposed to effectively optimize the flexibility of system architecture, and improve the performance of calculation. The hardware implementation results show that the maximum operating frequency of the coprocessor can reach 345 MHz. Compared with related state-of-the-art works, it achieves the highest operating frequency on the same Xilinx UltraScale+ FPGA platform, performing the encryption and decryption operations within 13.5 and 15.4 μ s, respectively. Meanwhile, this article achieves 1.7/3.1/5.9× area-time product improvements in look-up table flip-flop block memory storage with good flexibility.
- 16 August 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems I: Regular Papers
- Vol. 69 (11) , 4572-4582
An Efficient Unstructured Sparse Convolutional Neural Network Accelerator for Wearable ECG Classification Device
Convolution neural network (CNN) with pruning techniques has shown remarkable prospects in electrocardiogram (ECG) classification. However, efficiently deploying the existing pruned neural network to wearable devices for ECG classification is a great challenge due to the limited hardware resource and randomly distributed sparse weights. To address this issue, an efficient unstructured sparse CNN accelerator is proposed in this paper. A tile-first dataflow with compressed data storage format is presented to skip zero weight multiplications and increase the computing efficiency during inference of small-scale model with large sparsity. The two-level weight index matching structure in the dataflow exploits shifting operation to select valid data pairs and maintain the fully-pipelined calculation process. A configurable processing element (PE) array with 32-bit instruction control is proposed to increase the flexibility of the accelerator. Verified in FPGA and post-synthesis simulations in SMIC 40nm process, the proposed sparse CNN accelerator consumes $3.93~\mu $ J/classification at 2MHz clock frequency and it achieves an averaged ECG classification accuracy of 98.99%. A computing efficiency of 118.75% is realized which is improved by 48% compared to the dense baseline. In brief, the proposed efficient CNN accelerator is especially suitable for wearable ECG classification device.
- 16 August 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems I: Regular Papers
- Vol. 69 (11) , 4572-4582
An Efficient Unstructured Sparse Convolutional Neural Network Accelerator for Wearable ECG Classification Device
Convolution neural network (CNN) with pruning techniques has shown remarkable prospects in electrocardiogram (ECG) classification. However, efficiently deploying the existing pruned neural network to wearable devices for ECG classification is a great challenge due to the limited hardware resource and randomly distributed sparse weights. To address this issue, an efficient unstructured sparse CNN accelerator is proposed in this paper. A tile-first dataflow with compressed data storage format is presented to skip zero weight multiplications and increase the computing efficiency during inference of small-scale model with large sparsity. The two-level weight index matching structure in the dataflow exploits shifting operation to select valid data pairs and maintain the fully-pipelined calculation process. A configurable processing element (PE) array with 32-bit instruction control is proposed to increase the flexibility of the accelerator. Verified in FPGA and post-synthesis simulations in SMIC 40nm process, the proposed sparse CNN accelerator consumes $3.93~\mu $ J/classification at 2MHz clock frequency and it achieves an averaged ECG classification accuracy of 98.99%. A computing efficiency of 118.75% is realized which is improved by 48% compared to the dense baseline. In brief, the proposed efficient CNN accelerator is especially suitable for wearable ECG classification device.