Huizheng Wang

Huizheng Wang

Auto-generated profile, based on publicly available information

Huizheng Wang

School of Integrated Circuits, Tsinghua University, Beijing, China

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

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

Representative Publications

View All Publications
10 March 2021 in IEEE Transactions on Signal Processing
Verified
Journal Article
Closed access
HWHuizheng Wang
Huizheng Wang
YJYahui Ji
Yahui Ji
YSYifei Shen
Yifei Shen
WSWenqing Song
Wenqing Song
MLMuhao Li
Muhao Li
CZChuan Zhang
Chuan Zhang
Massive multiple-input multiple-output (M-MIMO) brings better robustness and spectral efficiency but higher computational challenges compared to small-scale MIMO. One of the key challenges is the large-scale matrix inversion, as widely employed in channel estimation and detection. Traditionally, to address the issue, several low-complexity matrix inversion methods have been proposed, including the tri-diagonal matrix approximation (TMA) and the Neumann-series approximation (NSA). Although the previous methods effectively alleviate the computational cost, they all fail to exploit the typical properties of channel matrices, leading to unsatisfactory error-rate performance in some non-ideal scenarios. To solve the issue, in this paper, a two-level and block diagonal based improved Neumann series approximation (TL-BD-INSA) algorithm is proposed, which is suitable for both ideal uncorrelated channels and the correlated channels with multiple-antenna user equipment (MAUE) system. First, a two-level block diagonal iteration based on matrix partition is employed, which exhibits performance comparable to the exact method while having a lower computational load. An improved normalization factor is then introduced to accelerate convergence. Numerical results show that, for 128×32 MIMO with MAUE non-ideal channel, the proposed algorithm performs only 0.25 dB away from the exact matrix inversion when bit error rate (BER) = 10 -3 . The implementation on Xilinx Virtex-7 FPGA and ASIC with TSMC 45 nm shows that the proposed detector can achieve 1731 bps/slices and 0.463 Gbps/mm 2 hardware efficiency, respectively, demonstrating that the proposed system can achieve a well trade-off between error performance and implementation efficiency.

20 June 2025
Verified
Conference Paper
Open access
QYQize Yang
Qize Yang
TWTaiquan Wei
Taiquan Wei
SGSihan Guan
Sihan Guan
CLChengran Li
Chengran Li
HSHaoran Shang
Haoran Shang
YHYang Hu
Yang Hu

25 January 2023 in IEEE Transactions on Circuits and Systems I: Regular Papers
Verified
Journal Article
Closed access
HWHuizheng Wang
Huizheng Wang
BCBingyang Cheng
Bingyang Cheng
XTXiaosi Tan
Xiaosi Tan
XYXiaohu You
Xiaohu You
CZChuan Zhang
Chuan Zhang
Expectation propagation (EP) achieves near-optimal performance for large-scale multiple-input multiple-output (L-MIMO) detection, however, at the expense of unaffordable matrix inversions. To tackle the issue, several low-complexity EP detectors have been proposed. However, they all fail to exploit the properties of channel matrices, thus resulting in unsatisfactory performance in non-ideal scenarios. To this end, in this paper, a block-diagonal Neumann-series-based expectation propagation approximation (BD-NS-EPA) algorithm is proposed, which is applicable for both ideal uncorrelated channels and the correlated channels with multiple-antenna user equipment system. First, a block-diagonal-based Neumann iteration is employed, which skillfully exerts the main information of the channels while reducing computational cost. An adjustable sorting message updating scheme then is introduced to reduce the update of redundant nodes during iterations. Numerical results show that, for $128\times 32$ MIMO with the non-ideal channel, the proposed algorithm exhibits 0.3 dB away from the original EP when bit error-rate (BER) $=10^{-3}$ , at the cost of mere 3% normalized complexity. The implementation results on SMIC 65-nm CMOS technology suggest that the proposed detector can achieve 1.252 Gbps/W and 0.275 Mbps/kGE hardware efficiency, further demonstrating that the proposed detectors can achieve a good trade-off between error-rate performance and hardware efficiency.

1 September 2024 in Integrated Circuits and Systems
Verified
Journal Article
Open access
HWHuizheng Wang
Huizheng Wang
QYQize Yang
Qize Yang
TWTaiquan Wei
Taiquan Wei
XYXingmao Yu
Xingmao Yu
CLChengran Li
Chengran Li
SWShaojun Wei
Shaojun Wei
Transformer-based large language models (LLMs) have made significant strides in the field of artificial intelligence (AI). However, training these LLMs imposes immense demands on computational power and bandwidth for hardware systems. Wafer-scale chips (WSCs) offer a promising solution, yet they struggle with limited on-chip memory and complex tensor partitioning. To fully harness the high-bandwidth, low-latency on-chip interconnect benefits of WSCs and to alleviate the on-chip memory limitations, a specialized mapping and architecture co-exploration method is essential. Despite existing efforts in memory optimization and mapping, current approaches fall short for WSC scenarios. To bridge this gap, we introduce TMAC, an architecture-mapping co-exploration framework that integrates recomputation into the design space, fully exploiting optimization opportunities overlooked by existing works. Further, TMAC takes advantage of the superior on-chip interconnect performance of WSCs by incorporating a more flexible tensor partition scheme. TMAC then introduces a novel operator-centric encoding scheme (OCES) designed to comprehensively describe the mapping space for training LLMs. Unlike previous studies that focus solely on communication volume analysis based on mapping, TMAC explores the design space by evaluating the combined impact of mapping and architecture on training performance. However, fully accounting for these untapped optimization opportunities increases the complexity of the design space. To address this, we streamline the simulation process, reducing the time needed for exploration. Compared to AccPar, Deepspeed and Megatron, TMAC delivers a $3.1\times$ , $2.9\times$ , $1.6\times$ performance gain. In terms of memory usage, TMAC requires $3.6\times$ , $3.1\times$ less memory than AccPar and Deepspeed, respectively and is comparable to Megatron's full recomputation method.

19 October 2021 in IEEE Transactions on Circuits and Systems II: Express Briefs
Verified
Journal Article
Closed access
HWHuizheng Wang
Huizheng Wang
WXWeihong Xu
Weihong Xu
ZZZaichen Zhang
Zaichen Zhang
XYXiaohu You
Xiaohu You
CZChuan Zhang
Chuan Zhang
By utilizing stochastic computing (SC), the hardware consumption of convolutional neural networks (CNNs) can be decreased significantly. However, long stream length is required to produce acceptable results, which leads to extended computation time. As a result, the inherent random fluctuation error and long latency of processing random bitstreams have made previous SC-CNN implementations inefficient compared with conventional binary designs. To address these issues, in this brief, an efficient convolution architecture based on fast FIR algorithm (FFA) is proposed by employing FFA to reduce the computational complexity. Further, the combination of two-line SC and Sobol sequences is applied to decrease the processing cycles. The functional simulation targeting LeNet-5 with MNIST dataset and RTL synthesis results show that the proposed design yields higher area efficiency than previous SC-based ones and achieves 64%, 11% higher efficiency in area and energy compared to the 5-bit fixed-point design while maintaining comparable accuracy.