Acta Optica Sinica, Volume. 44, Issue 19, 1906002(2024)

Modeling and FPGA Application of Optoelectronic Oscillation Chaotic System Based on Deep Learning

Zhuoyu Zhang1, Lin Jiang1、*, Boyang Chen1, Guohao Feng1, Jiacheng Feng1, and Lianshan Yan1,2
Author Affiliations
  • 1School of Information Science and Technology, Southwest Jiaotong University, Chengdu 611756, Sichuan , China
  • 2Yantai Research Institute of New Generation Information Technology, Southwest Jiaotong University, Yantai 264001, Shandong , China
  • show less
    Figures & Tables(16)
    LSTM modeling and FPGA deployment process of OEO chaotic system. (a) OEO system structure based on MZM; (b) LSTM model; (c) target FPGA platform
    Time series and scatter plots of model prediction data and real data under different s. (a)(b) s=1; (c)(d) s=2
    Model deployment overall workflow
    Chaotic AI model based on LSTM structure before and after optimization
    Hardware structure of chaotic AI model
    LSTM arithmetic pipeline of chaotic AI model
    Overall architecture diagram of FPGA system based on chaotic AI model
    FPGA end chaotic generation model resource occupancy diagram
    Scatter plot of model prediction results between FPGA and PC ends
    Phase diagrams of OEO chaotic system and FPGA end chaotic AI model. (a) OEO system; (b) FPGA end model
    Chaotic AI model output results of FPGA before and after adding disturbance. (a) Time series plot of the first prediction point; (b) MSE between model outputs before and after adding disturbance
    Comparison of output waveform of chaotic AI model between FPGA and PC. (a) FPGA; (b) PC
    Histograms under different m values. (a) m=8; (b) m=7; (c) m=6; (d) m=5
    • Table 1. MSE and r of model prediction data and real data under different s

      View table

      Table 1. MSE and r of model prediction data and real data under different s

      s124816
      MSE0.002250.003820.006290.01570.0289
      r0.9810.9680.9470.8610.723
    • Table 2. Comparison of model resources and computational latency before and after optimization

      View table

      Table 2. Comparison of model resources and computational latency before and after optimization

      ModelResource (Ratio /%)Latency /ns
      DSPBRAM_36
      Before optimization82(20.5)75(28.3)2300
      After optimization56(14.0)31.5(11.9)1280
    • Table 3. NIST SP 800-22 test results of random bit sequence generated by 5-LSB

      View table

      Table 3. NIST SP 800-22 test results of random bit sequence generated by 5-LSB

      Statistical testP-valueProportionResult
      Frequency0.0174250.991Success
      BlockFrequency0.2939520.990Success
      CumulativeSums0.4335900.991Success
      Runs0.0126500.991Success
      LongestRun0.3298500.987Success
      Rank0.6537730.984Success
      FFT0.4190210.990Success
      NonOverlappingTemplate0.9240760.992Success
      OverlappingTemplate0.6454480.981Success
      Universal0.6017660.990Success
      ApproximateEntropy0.1160650.989Success
      RandomExcursions0.8241920.994Success
      RandomExcursionsVariant0.9698580.994Success
      Serial0.8801450.993Success
      LinearComplexity0.0790510.982Success
    Tools

    Get Citation

    Copy Citation Text

    Zhuoyu Zhang, Lin Jiang, Boyang Chen, Guohao Feng, Jiacheng Feng, Lianshan Yan. Modeling and FPGA Application of Optoelectronic Oscillation Chaotic System Based on Deep Learning[J]. Acta Optica Sinica, 2024, 44(19): 1906002

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: Fiber Optics and Optical Communications

    Received: Apr. 21, 2024

    Accepted: May. 16, 2024

    Published Online: Oct. 12, 2024

    The Author Email: Jiang Lin (linjiang@swjtu.edu.cn)

    DOI:10.3788/AOS240879

    CSTR:32393.14.AOS240879

    Topics