復值Hopfield神經(jīng)網(wǎng)絡的信號盲檢測一步計算電路

洪慶輝; 孫辰; 肖平旦; 韋正苗; 杜四春

doi:10.11999/JEIT240224

復值Hopfield神經(jīng)網(wǎng)絡的信號盲檢測一步計算電路

doi: 10.11999/JEIT240224 cstr: 32379.14.JEIT240224

湖南大學信息科學與工程學院長沙 410082

基金項目: 國家自然科學基金(62234008, 62371186)，湖湘青年英才項目(2023RC3103)，湖南省自然科學基金(2023JJ30168, 2022JJ30160, 2021JJ40111)，國家重點研發(fā)計劃(2022YFB3903800)

詳細信息

作者簡介:
洪慶輝：男，副教授，研究方向為模擬存算一體電路設計及應用

孫辰：男，碩士生，研究方向為復數(shù)神經(jīng)網(wǎng)絡電路設計

肖平旦：男，博士生，研究方向為基于憶阻器的存內計算電路設計及其應用

韋正苗：男，博士生，研究方向為模擬電路求解矩陣方程的新方法及其應用

杜四春：男，副教授，研究方向為模擬/混合、射頻集成電路設計

通訊作者:
杜四春　jt_dsc@hnu.edu.cn

中圖分類號: TN402
計量
- 文章訪問數(shù): 278
- HTML全文瀏覽量: 72
- PDF下載量: 66
- 被引次數(shù): 0
出版歷程
- 收稿日期: 2024-03-29
- 修回日期: 2024-10-10
- 網(wǎng)絡出版日期: 2024-10-16
- 刊出日期: 2024-11-01

One-step Calculation Circuit of Blind Signal Detection using Complex-valued Hopfield Neural Network

College of Information Science and Engineering, Hunan University, Changsha 410082, China

Funds: The National Natural Science Foundation of China (62234008, 62371186), Huxiang Young Talents Project (2023RC3103), The Natural Science Foundation of Hunan Province(2023JJ30168, 2022JJ30160, 2021JJ40111), The National Key R&D Program of China (2022YFB3903800)

摘要

摘要: 信號盲檢測在大規(guī)模通信網(wǎng)絡中具有重要的意義并得到了廣泛的應用，如何快速得到信號盲檢測結果是新一代實時通信網(wǎng)絡的迫切需求。為此，該文從模擬電路的角度設計了一種能加速信號盲檢測的復值Hopfield神經(jīng)網(wǎng)絡(CHNN)電路，該電路可一步完成大規(guī)模并行計算，提高信號盲檢測速度，同時該電路可以通過調整憶阻器的電導和輸入電壓來實現(xiàn)可編程功能。Pspice仿真結果表明，該電路的計算精度可達99%以上，運行時間比Matlab軟件仿真快3個數(shù)量級，此外，該電路具有良好的魯棒性，即使在20%的噪聲干擾下，仍能保持99%以上的計算精度。
- 電路設計 /
- 憶阻器 /
- 復值Hopfield神經(jīng)網(wǎng)絡 /
- 信號盲檢測
Abstract: Blind signal detection is of great significance in large-scale communication networks and has been widely used. How to quickly obtain blind signal detection results is an urgent need for the new generation of real-time communication networks. Considering this demand, a Complex-valued Hopfield Neural Network (CHNN) circuit is designed that can accelerate blind signal detection from an analog circuit perspective, the proposed circuit can accelerate the blind signal detection by rapidly performing massively parallel calculation in one step. At the same time, the circuit can be programmable by adjusting the conductance and input voltage of the memristor. The Pspice simulation results show that the computing accuracy of the proposed circuit can exceed 99%. Compared with Matlab software simulation, the proposed circuit is three orders of magnitude faster in terms of computing time. And the accuracy can be maintained at more than 99% even under the interference of 20% noise.
- Circuit design /
- Memristor /
- Complex-valued Hopfield Neural Network(CHNN) /
- Blind signal detection

HTML全文

圖 1 信號盲檢測的處理過程

下載: 全尺寸圖片幻燈片

圖 2 K=8時的復值激活函數(shù)

下載: 全尺寸圖片幻燈片

圖 3 復值乘法電路

下載: 全尺寸圖片幻燈片

圖 4 復值激活函數(shù)電路

下載: 全尺寸圖片幻燈片

圖 5 復值激活函數(shù)電路輸出結果

下載: 全尺寸圖片幻燈片

圖 6 信號盲檢測CHNN電路

下載: 全尺寸圖片幻燈片

圖 7 憶阻器的兩種模式

下載: 全尺寸圖片幻燈片

圖 8 CHNN電路處理流程圖

下載: 全尺寸圖片幻燈片

圖 9 CHNN電路的輸出結果

下載: 全尺寸圖片幻燈片

圖 10 電路精度及BER性能比較

下載: 全尺寸圖片幻燈片

圖 11 電壓噪聲波形及噪聲干擾下電路的精度

下載: 全尺寸圖片幻燈片

圖 12 線電阻干擾下電路的精度

下載: 全尺寸圖片幻燈片

圖 13 不同隨機誤差條件下電路的平均精度

下載: 全尺寸圖片幻燈片

圖 14 憶阻器編程失敗情況下電路的平均精度

下載: 全尺寸圖片幻燈片

表 1 電路和軟件計算時間比較(ms)

輸入信號數(shù)量計算時間
Pspice Matlab

5 階 0.001 9.5
10 階 0.03 10.8
20 階 0.04 11.2
40 階 0.07 13.3
80 階 0.16 15.2

下載: 導出CSV

表 2 不同硬件的計算時間

方式其他電路^[19] FPGA^[22] DSP^[22]

計算時間 8.2$ \times $ 3.8$ \times $ 8.2$ \times $

下載: 導出CSV

參考文獻(22)

[1]	SATO Y. A method of self-recovering equalization for multilevel amplitude-modulation systems[J]. IEEE Transactions on Communications, 1975, 23(6): 679–682. doi: 10.1109/TCOM.1975.1092854.
[2]	TONG Lang, XU Guanghan, and KAILATH T. Blind identification and equalization based on second-order statistics: A time domain approach[J]. IEEE Transactions on Information Theory, 1994, 40(2): 340–349. doi: 10.1109/18.312157.
[3]	LI Jin, FENG Dazheng, and LI Bingbing. Space-time blind equalization of dispersive MIMO systems driven by QAM signals[J]. IEEE Transactions on Vehicular Technology, 2018, 67(5): 4136–4148. doi: 10.1109/TVT.2018.2790432.
[4]	HU Rong, YU Shujuan, ZHANG Yun, et al. WSN blind detection based on SOS[C]. The 11th World Congress on Intelligent Control and Automation, Shenyang, China, 2014: 5763–5766. doi: 10.1109/WCICA.2014.7053704.
[5]	LV Zhisheng, FENG Bin, and TAN Li. An improved dual-mode blind equalization algorithm for QAM signals[C]. 2022 6th International Conference on Robotics and Automation Sciences (ICRAS), Wuhan, China, 2022: 283–287. doi: 10.1109/ICRAS55217.2022.9842274.
[6]	BLOCHBERGER M, ELVANDER F, ALI R, et al. Distributed cross-relation-based frequency-domain blind system identification using online-admm[C]. 2022 International Workshop on Acoustic Signal Enhancement (IWAENC), Bamberg, Germany, 2022: 1–5. doi: 10.1109/IWAENC53105.2022.9914714.
[7]	YUAN Kang, ZHUO Junnan, GAO Wei, et al. Diffusion constant modulus algorithm for blind equalization[C]. 2022 IEEE International Conference on Signal Processing, Communications and Computing (ICSPCC), Xi’an, China, 2022: 1–6. doi: 10.1109/ICSPCC55723.2022.9984630.
[8]	XIAO Ying and DONG Yuhua. Blind equalization based on neural network under LS criterion by gradient iteration algorithm[C]. 2009 Fifth International Conference on Natural Computation, Tianjian, China, 2009: 91–94. doi: 10.1109/ICNC.2009.134.
[9]	ZHANG Yun and ZHANG Zhiyong. Blind 8PSK signals detection using discrete complex-valued Hopfield network[C]. 2010 IEEE 12th International Conference on Communication Technology, Nanjing, China, 2010: 983–986. doi: 10.1109/ICCT.2010.5688787.
[10]	WU Jinwen, JIN Haihong, YU Shujuan, et al. A novel blind detection algorithm based on multi-clustering complex neural network[C]. Proceedings of 2021 IEEE 4th Advanced Information Management, Communicates, Electronic and Automation Control Conference (IMCEC), Chongqing, China, 2021: 1017–1021. doi: 10.1109/IMCEC51613.2021.9482354.
[11]	MEI Ruru, WANG Zhugang, and HU Wanru. Robust blind equalization algorithm using convolutional neural network[J]. IEEE Signal Processing Letters, 2022, 29: 1569–1573. doi: 10.1109/LSP.2022.3189319.
[12]	XIE Xihai, FAN Chenzhao, WANG Shuai, et al. Varible step size multi-layer neural network blind equalization algorithm[C]. 2022 4th International Conference on Natural Language Processing (ICNLP), Xi’an, China, 2022: 477–481. doi: 10.1109/ICNLP55136.2022.00087.
[13]	ZHANG Hui, GU M, JIANG Xudong, et al. An optical neural chip for implementing complex-valued neural network[J]. Nature Communications, 2021, 12(1): 457. doi: 10.1038/s41467-020-20719-7.
[14]	YU Yaning and ZHANG Ziye. State estimation for complex-valued inertial neural networks with multiple time delays[J]. Mathematics, 2022, 10(10): 1725. doi: 10.3390/math10101725.
[15]	HONG Qinghui, FU Haotian, LIU Yiyang, et al. In-memory computing circuit implementation of complex-valued hopfield neural network for efficient portrait restoration[J]. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 2023, 42(10): 3338–3351. doi: 10.1109/TCAD.2023.3242858.
[16]	NISHIKAWA I, IRITANI T, and SAKAKIBARA K. Improvements of the traffic signal control by complex-valued Hopfield networks[C]. The 2006 IEEE International Joint Conference on Neural Network Proceedings, Vancouver, Canada, 2006: 459–464. doi: 10.1109/IJCNN.2006.246717.
[17]	王雷敏, 程佳俊, 胡成, 等. 基于改進的憶阻器在字符聯(lián)想記憶中的應用[J]. 電子與信息學報, 2023, 45(7): 2667–2674. doi: 10.11999/JEIT220709. WANG Leimin, CHENG Jiajun, HU Cheng, et al. Application of improved memristor in character associative memory[J]. Journal of Electronics & Information Technology, 2023, 45(7): 2667–2674. doi: 10.11999/JEIT220709.
[18]	UYKAN Z. Shadow-Cuts minimization/maximization and complex Hopfield neural networks[J]. IEEE Transactions on Neural Networks and Learning Systems, 2021, 32(3): 1096–1109. doi: 10.1109/TNNLS.2020.2980237.
[19]	HONG Qinghui, LI Ya, and WANG Xiaoping. Memristive continuous Hopfield neural network circuit for image restoration[J]. Neural Computing and Applications, 2020, 32(12): 8175–8185. doi: 10.1007/s00521-019-04305-7.
[20]	YAN Renao, HONG Qinghui, WANG Chunhua, et al. Multilayer memristive neural network circuit based on online learning for license plate detection[J]. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 2022, 41(9): 3000–3011. doi: 10.1109/TCAD.2021.3121347.
[21]	肖平旦, 洪慶輝, 杜四春, 等. 基于憶阻器的可編程矩陣行列式求解電路設計及應用[J]. 中國科學: 信息科學, 2023, 53(5): 1008–1025. doi: 10.1360/SSI-2022-0229. XIAO Pingdan, HONG Qinghui, DU Sichun, et al. Design and application of a programmable matrix determinant-solving circuit based on memristors[J]. SCIENTIA SINICA Informationis, 2023, 53(5): 1008–1025. doi: 10.1360/SSI-2022-0229.
[22]	YI Qian. FPGA implementation of neural network accelerator[C]. 2018 2nd IEEE Advanced Information Management, Communicates, Electronic and Automation Control Conference (IMCEC), Xi’an, China, 2018: 1903–1906. doi: 10.1109/IMCEC.2018.8469659.