高级搜索

留言板

尊敬的读者、作者、审稿人, 关于本刊的投稿、审稿、编辑和出版的任何问题, 您可以本页添加留言。我们将尽快给您答复。谢谢您的支持!

姓名
邮箱
手机号码
标题
留言内容
验证码

面向心电检测的混合多模卷积神经网络加速器设计

刘冬生 魏来 邹雪城 陆家昊 成轩 胡昂 李德建 赵旭 蒋曲明

刘冬生, 魏来, 邹雪城, 陆家昊, 成轩, 胡昂, 李德建, 赵旭, 蒋曲明. 面向心电检测的混合多模卷积神经网络加速器设计[J]. 电子与信息学报, 2023, 45(1): 33-41. doi: 10.11999/JEIT220534
引用本文: 刘冬生, 魏来, 邹雪城, 陆家昊, 成轩, 胡昂, 李德建, 赵旭, 蒋曲明. 面向心电检测的混合多模卷积神经网络加速器设计[J]. 电子与信息学报, 2023, 45(1): 33-41. doi: 10.11999/JEIT220534
LIU Dongsheng, WEI Lai, ZOU Xuecheng, LU Jiahao, CHENG Xuan, HU Ang, LI Dejian, ZHAO Xu, JIANG Quming. Design of Hybrid Multimode Convolutional Neural Network Accelerator for Electrocardiogram Detection[J]. Journal of Electronics & Information Technology, 2023, 45(1): 33-41. doi: 10.11999/JEIT220534
Citation: LIU Dongsheng, WEI Lai, ZOU Xuecheng, LU Jiahao, CHENG Xuan, HU Ang, LI Dejian, ZHAO Xu, JIANG Quming. Design of Hybrid Multimode Convolutional Neural Network Accelerator for Electrocardiogram Detection[J]. Journal of Electronics & Information Technology, 2023, 45(1): 33-41. doi: 10.11999/JEIT220534

面向心电检测的混合多模卷积神经网络加速器设计

doi: 10.11999/JEIT220534
基金项目: 国家自然科学基金(62134002),国家重点研发计划(2021YFA0715502),湖北省重点研发项目(YFYB2020000413),东莞引进创新科研团队计划(201760712600139)
详细信息
    作者简介:

    刘冬生:男,博士,教授,研究方向为能效无线传输技术及芯片设计、后量子密码算法及密码芯片设计

    魏来:男,硕士生,研究方向为CNN硬件加速

    邹雪城:男,博士,教授,研究方向为超大规模集成电路

    陆家昊:男,博士生,研究方向为AI架构与处理器

    成轩:男,硕士生,研究方向为AI算法高效硬件实现

    胡昂:男,博士,研究方向为射频SoC

    李德建:女,硕士,研究方向为载波通信、低功耗设计

    赵旭:男,博士,研究方向为载波通信、无线通信

    蒋曲明:女,硕士,研究方向为电磁场与微波技术

    通讯作者:

    魏来 wei1123212130@163.com

  • 中图分类号: TN402

Design of Hybrid Multimode Convolutional Neural Network Accelerator for Electrocardiogram Detection

Funds: The National Natural Science Foundation of China (62134002), The National Key Research and Development Program of China (2021YFA0715502), The Key Research and Development Project of Hubei Province (YFYB2020000413), The Introduced Innovative R&D Team Program of Dongguan (201760712600139)
  • 摘要: 随着医疗资源日益匮乏以及人口老龄化日趋严重,心血管疾病已对人类健康造成了极大的威胁。具有心电(ECG)检测的便携式设备能有效降低心血管疾病对患者的威胁,因此该文设计了一种面向心电检测的混合多模卷积神经网络加速器。该文首先介绍了一种用于心电信号分类的1维卷积神经网络(1D-CNN)模型,随后针对该模型设计了一种高效的卷积神经网络(CNN)加速器,该加速器采用了一种多并行展开策略和多数据流的运算模式完成了卷积循环的加速和优化,能在时间上和空间上高度复用数据,同时提高了硬件资源利用率,从而提升了硬件加速器的硬件效率。最后基于Xilinx ZC706硬件平台完成了原型验证,结果显示,所设计卷积神经网络加速器消耗的资源为2247 LUTs, 80 DSPs。在200 MHz的工作频率下,该设计的整体性能可达到28.1 GOPS,并且硬件效率达到了12.82 GOPS/kLUT。
  • 图  1  各类型ECG信号的波形特征

    图  2  卷积层运算的4级循环

    图  3  全连接层运算的两级循环

    图  4  1D-CNN模型结构图[13]

    图  5  各层计算对应的运算模式

    图  6  3D-PE阵列结构

    图  7  CNN加速器的总体硬件架构

    图  8  FPGA硬件分类结果的混淆矩阵

    表  1  软硬件平台测试集的性能指标(%)

    平台AccSenSpecPpr
    Intel Core i5-1040099.2097.9799.6098.80
    ZYNQ 7000 ZC70699.1997.9299.6098.80
    下载: 导出CSV

    表  2  CNN加速器性能比较

    文献FPGABit-widthCNN ModelClock
    (MHz)
    DSP
    Utilization
    Resource
    (kLUT)
    Throughput
    (GOPS)
    Efficiency
    (GOPS/kLUT)
    DSP Efficiency
    (GOPS/DSP)
    [7]Arria 10
    GX 1150
    16 bit FixedVGG-162003036235k1030.24.3840.339
    [17]Zynq XC7Z04516 bit FixedVGG-16150780183k137.00.7490.175
    [18]Zynq
    ZCU 102
    16 bit FixedVGG-162001352390k495.41.2700.366
    [19]Zynq XC7Z04516 bit FixedVGG-1617257659k316.25.3580.549
    本设计Zynq ZC70616 bit Fixed1D-CNN200802.2k28.112.8200.360
    下载: 导出CSV
  • [1] United Nations. World population ageing 2017: highlights[R]. ST/ESA/SER. A/397, 2017.
    [2] ZHANG Jiancan. Construction and application of “internet” mental health cloud platform based on trinity model analysis[C]. 2021 International Conference on Internet, Education and Information Technology (IEIT), Suzhou, China, 2021: 1–5.
    [3] LI Feiteng, WU Jiaquan, JIA Menghan, et al. Automated heartbeat classification exploiting convolutional neural network with channel-wise attention[J]. IEEE Access, 2019, 7: 122955–122963. doi: 10.1109/ACCESS.2019.2938617
    [4] QIAN Yanmin, BI Mengxiao, TAN Tian, et al. Very deep convolutional neural networks for noise robust speech recognition[J]. IEEE/ACM Transactions on Audio, Speech, and Language Processing, 2016, 24(12): 2263–2276. doi: 10.1109/TASLP.2016.2602884
    [5] LI Zhen, WANG Yuqing, ZHI Tian, et al. A survey of neural network accelerators[J]. Frontiers of Computer Science, 2017, 11(5): 746–761. doi: 10.1007/s11704-016-6159-1
    [6] WANG Ying, WANG Yongchen, SHI Cong, et al. An edge 3D CNN accelerator for low-power activity recognition[J]. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 2021, 40(5): 918–930. doi: 10.1109/TCAD.2020.3011042
    [7] MA Yufei, CAO Yu, VRUDHULA S, et al. Optimizing the convolution operation to accelerate deep neural networks on FPGA[J]. IEEE Transactions on Very Large Scale Integration (VLSI) Systems, 2018, 26(7): 1354–1367. doi: 10.1109/TVLSI.2018.2815603
    [8] PARK S W, PARK J, BONG K, et al. An energy-efficient and scalable deep learning/inference processor with tetra-parallel MIMD architecture for big data applications[J]. IEEE Transactions on Biomedical Circuits and Systems, 2015, 9(6): 838–848. doi: 10.1109/TBCAS.2015.2504563
    [9] TU Fengbin, WU Weiwei, WANG Yang, et al. Evolver: A deep learning processor with on-device quantization–voltage–frequency tuning[J]. IEEE Journal of Solid-State Circuits, 2021, 56(2): 658–673. doi: 10.1109/JSSC.2020.3021661
    [10] ZHU Lingsong, LIU Dongsheng, LI Xiaohuan, et al. An efficient hardware architecture for epileptic seizure detection using EEG signals based on 1D-CNN[C]. 2021 IEEE 14th International Conference on ASIC (ASICON), Kunming, China, 2021: 1–4.
    [11] LIU Xinheng, CHEN Yao, HAO Cong, et al. WinoCNN: Kernel sharing winograd systolic array for efficient convolutional neural network acceleration on FPGAs[C]. 2021 IEEE 32nd International Conference on Application-specific Systems, Architectures and Processors (ASAP), NJ, USA, 2021: 258–265.
    [12] YUAN Zhe, LIU Yongpan, YUE Jinshan, et al. STICKER: An energy-efficient multi-sparsity compatible accelerator for convolutional neural networks in 65-nm CMOS[J]. IEEE Journal of Solid-State Circuits, 2020, 55(2): 465–477. doi: 10.1109/JSSC.2019.2946771
    [13] WEI Lai, LIU Dongsheng, LU Jiahao, et al. A low-cost hardware architecture of convolutional neural network for ECG classification[C]. 2021 9th International Symposium on Next Generation Electronics (ISNE), Changsha, China, 2021: 1–4.
    [14] DU Li, DU Yuan, LI Yilei, et al. A reconfigurable streaming deep convolutional neural network accelerator for internet of things[J]. IEEE Transactions on Circuits and Systems I:Regular Papers, 2018, 65(1): 198–208. doi: 10.1109/TCSI.2017.2735490
    [15] MA Yufei, CAO Yu, VRUDHULA S, et al. An automatic RTL compiler for high-throughput FPGA implementation of diverse deep convolutional neural networks[C]. 2017 27th International Conference on Field Programmable Logic and Applications (FPL), Ghent, Belgium, 2017: 1–8.
    [16] MA Yufei, SUDA N, CAO Yu, et al. Scalable and modularized RTL compilation of convolutional neural networks onto FPGA[C]. 2016 26th International Conference on Field Programmable Logic and Applications (FPL), Lausanne, Switzerland, 2016: 1–8.
    [17] GUO Kaiyuan, SUI Lingzhi, QIU Jiantao, et al. Angel-eye: A complete design flow for mapping CNN onto embedded FPGA[J]. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 2018, 37(1): 35–47. doi: 10.1109/TCAD.2017.2705069
    [18] ZHU Chaoyang, HUANG Kejie, YANG Shuyuan, et al. An efficient hardware accelerator for structured sparse convolutional neural networks on FPGAs[J]. IEEE Transactions on Very Large Scale Integration (VLSI) Systems, 2020, 28(9): 1953–1965. doi: 10.1109/TVLSI.2020.3002779
    [19] WANG Jichen, LIN Jun, and WANG Zhongfeng. Efficient hardware architectures for deep convolutional neural network[J]. IEEE Transactions on Circuits and Systems I:Regular Papers, 2018, 65(6): 1941–1953. doi: 10.1109/TCSI.2017.2767204
  • 加载中
图(8) / 表(2)
计量
  • 文章访问数:  574
  • HTML全文浏览量:  277
  • PDF下载量:  126
  • 被引次数: 0
出版历程
  • 收稿日期:  2022-04-28
  • 修回日期:  2022-06-18
  • 网络出版日期:  2022-06-28
  • 刊出日期:  2023-01-17

目录

    /

    返回文章
    返回