Li Zhang

Dr. Li Zhang


Research Focus & Metrics

Academic Career Length

17Years (2010-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
28 January 2022 in IEEE Transactions on Magnetics
Verified
Journal Article
Closed access
LZLi Zhang
Li Zhang
YMYitao Ma
Yitao Ma
TETetsuo Endoh
Tetsuo Endoh
Error-correcting codes (ECCs) are important and widely implemented in memories from high-speed SRAM cache to high-volume 3D-NAND. However, from the viewpoint, ECC technology of Spin-transfer torque magnetic random-access memory (STT-MRAM) is not established yet, as the operation speed of STT-MRAM is higher than DRAM/3D-NAND, and its volume is larger than SRAM. Moreover, it is difficult for existing ECCs to guarantee low latency and low hardware complexity while achieving high error correction capabilities. In this work, the time efficiency of both encoding and decoding of Bose-Chaudhuri-Hocquenghem (BCH) codes for STT-MRAM is optimized. A divisor-distance-based (DDB) polynomial division method is proposed to accelerate the polynomial division of BCH encoding and decoding. The DDB division method leverages the characteristics of the divisor polynomial to achieve parallel-processing for multi-bit on the word-line in the same block in a simple manner. And DDB polynomial division method can execute in multiple approaches with different hardware architectures, where the DDB divider with a parallel multiplier (PM) has the lowest hardware complexity. To demonstrate the superiority of the proposed algorithm: The time efficiency of the proposed BCH codes is successfully verified in software, proposed algorithm increases the time efficiency of encoding by more than 10 times, and the decoding increased by about 10%. The hardware implementation of proposed DDB divider is presented for improving the hardware complexity, the PM-type DDB divider architecture eliminates the large memory block of the traditional lookup-table type parallel divider.

17 December 2010 in IEEE Transactions on Communications
Verified
Journal Article
Closed access
LZLi Zhang
Li Zhang
QHQin Huang
Qin Huang
SLShu Lin
Shu Lin
Codes constructed based on finite geometries form a large class of cyclic codes with large minimum distances which can be decoded with simple majority-logic decoding in one or multiple steps. In 2001, Kou, Lin and Fossorier showed that the one-step majority-logic decodable finite geometry codes form a class of cyclic LDPC codes whose Tanner graphs are free of cycles of length 4. These cyclic finite geometry LDPC codes perform very well over the AWGN channel using iterative decoding based on belief propagation (IDBP) and have very low error-floors. However, the standard IDBP is not effective for decoding other cyclic finite geometry codes because their Tanner graphs contain too many short cycles of length 4 which severely degrade the decoding performance. This paper investigates iterative decoding of two-step majority-logic decodable finite geometry codes. Three effective algorithms for decoding these codes are proposed. These algorithms are devised based on the orthogonal structure of the parity-check matrices of the codes to avoid or reduce the degrading effect of the short cycles of length 4. These decoding algorithms provide significant coding gains over the standard IDBP using either the sum-product or the min-sum algorithms.

26 February 2024 in IEEE Transactions on Industrial Informatics
Verified
Journal Article
Closed access
LZLi Zhang
Li Zhang
TLTao Li
Tao Li
TETetsuo Endoh
Tetsuo Endoh
This article proposes a novel error correction code (ECC) module designed for future object recognition systems that are expected to utilize spin-transfer torque MRAM (STT-MRAM) for weight data storage. The module focuses on achieving high throughput and a small area while ensuring the integrity of the data. A pipelined high-parallel Bose–Chaudhuri–Hocquenghem (BCH) module is proposed to address the high throughput requirement of the ECC module in STT-MRAM. A 64-parallel linear feedback shift register for the encoder and parallel factor reduced decoder are proposed to optimize the small area requirement, however, maintain the read delay within the nanosecond level. Code length and correction capability of BCH codes are optimized, which are feasible to object recognition systems with high efficiency of memory redundancy and computational complexity. The ECC's correction performance is verified using error maps obtained from practical measurements of real-working STT-MRAM chips fabricated using 37 nm MTJ/40 nm CMOS technologies. The implementation results demonstrate that the proposed ECC module achieved the highest throughput of 64 ${bit} \cdot {{f}}_{{clk}}$ and a small area of 18.27K $\mu {\mathrm{m}}^2$ with a read latency of 5.76 ns. Practical experiments utilized error maps of STT-MRAM and MobileNet weight dataset based on CNN to verify the proposed ECC. Experiment results demonstrated the proposed ECC module achieved no accuracy degradation when the bit error rate of STT-MRAM is under 2×10 -3 for neural networks.

8 January 2026 in Electronics
Verified
Journal Article
Open access
PDF file
XWXinyu Wu
Xinyu Wu
LZLi Zhang
Li Zhang
HYHAITAO YOU
HAITAO YOU
SZShizeng Zhang
Shizeng Zhang
DNDimitar Nikolov Nikolov
Dimitar Nikolov Nikolov
QCQiang Cui
Qiang Cui
The imperative for EMC-optimized gate drivers in Gallium Nitride (GaN)-based automotive DC-DC converters stems from the stringent CISPR 25 standards and GaN’s intrinsic high-speed switching characteristics, which paradoxically exacerbate electromagnetic interference (EMI). This review distinguishes itself by proposing a novel frequency-domain classification framework (Zone I: 50 MHz for switching noise and ringing), which systematically organizes and assesses gate driving techniques against the triad of fundamental GaN EMC challenges: pronounced capacitance nonlinearity, low threshold voltage, and extreme parasitic sensitivity. Unlike prior surveys that primarily catalog techniques, the analysis elevates the gate driver from a simple switch interface to the central “electromagnetic actuator” of the power stage, explicitly elucidating its pivotal role in mediating the critical trade-offs among switching speed, loss, and EMC performance. A comprehensive evaluation and comparison of advanced techniques—from spread-spectrum modulation for Zone I to adaptive current shaping and resonant topologies for Zone II—are provided, alongside an analysis of their design trade-offs. Furthermore, this review presents a first-of-its-kind, phased implementation roadmap towards holistic EMC compliance, integrating intelligent hybrid control, heterogeneous integration, and system-level co-design. This review bridges the gap between device physics and system engineering, offering structured design methodologies and a clear future direction for achieving electromagnetic integrity in next-generation automotive power electronics.

1 October 2019
Verified
Conference Paper
Closed access
LZLi Zhang
Li Zhang
NLNan Liu
Nan Liu
PZPan Zhiwen
Pan Zhiwen
XYXiaohu You
Xiaohu You
The Noisy Gradient Descent Bit Flipping (NGDBF) algorithm surpasses the previous Gradient Descent Bit Flipping (GDBF) and other Bit Flipping (BF) algorithms for decoding Low-Density Parity-Check (LDPC) codes. However, the NGDBF decoder sometimes cannot escape from the local maxima. To further enhance the performance of the NGDBF algorithm, an improved NGDBF algorithm called Tabu-List Noisy Gradient Descent Bit Flipping (TNGDBF) is proposed in this paper. A tabu-list is employed to help the NGDBF decoder escape from the local maxima. It is shown in the simulation results that the TNGDBF algorithm gains about 0.2 dB performance compared with NGDBF at the cost of a small amount of complexity increase.

4 October 2023 in IEEE Transactions on Magnetics
Verified
Journal Article
Closed access
TLTao Li
Tao Li
LZLi Zhang
Li Zhang
YMYitao Ma
Yitao Ma
TETetsuo Endoh
Tetsuo Endoh
The continuous emergence of new non-volatile devices and the rapid evolution of deep neural networks (DNNs) have brought unprecedented opportunities and challenges to the diverse applications of brain-inspired processors in edge computing. Since the research on DNNs and non-volatile devices develops on two separate and independent paths, there is a lack of investigation to establish the cross-integration of their advantages in academia and industry. This study aims to bridge the gap between these fields by associating, for the first time, the fault-tolerance of DNNs with the switching properties of non-volatile devices, thereby enabling the low-power consumption of devices without the loss of accuracy of DNNs. This paper theoretically clarifies and derives the relationship between the DNNs’ fault tolerance and the switching current of the non-volatile devices and experimentally demonstrates the DNNs’ fault tolerance by introducing random errors to their low 8-bit parameters. The experimental results demonstrate that randomly flipping 20% binaries from the low 8-bit DNNs’ parameters can reduce power consumption by 5.63%, and 100% binaries can save 34.84% energy. Incorporating the error maps of six practical spin-transfer torque magnetic random-access memory (STT-MRAM) chips (with a maximum error rate of 0.00868) into the low 8-bit of DNNs’ parameters has no impact on its precision (with a maximum loss of 0.00067) under switching current corresponding to 100% switching probability.