Chinese Optics Letters, Volume. 22, Issue 4, 040601(2024)

Neural network equalization based on delta-sigma modulation

Bo Liu1、*, Jianxin Ren1, Xiangyu Wu1, Shuaidong Chen1, Yaya Mao1, and Li Zhao2
Author Affiliations
  • 1Nanjing University of Information Science & Technology, Nanjing 210044, China
  • 2Key Laboratory for Information Science of Electromagnetic Waves (MoE), Fudan University, Shanghai 200433, China
  • show less
    Figures & Tables(10)
    Noise transfer function. (a) Zero points and pole points; (b) frequency response of amplitude.
    Principle of NNEs.
    Experimental setup of PDM-8192-QAM W-band wireless delivery over 4.6 km. Photos of (a) Guanghua Building in the Handan campus; (b) transmission link; (c) Physics Building in the Jiangwan campus; (d) transmitter; (e) receiver.
    Offline DSP for transmitter and receiver.
    (a) Electrical spectrum of received 13.5-GHz IF signals; (b) electrical spectrum of baseband signals after NNE.
    Curves of BER versus power into the PD at the case of 10 Gbaud PDM-QPSK signal. (i) constellation with only DSM; (ii) constellation with DSM and NNE.
    Curves of BER versus power into the PD at the case of 1 Gbaud PDM-OFDM signal. (a) Constellation of 1024-QAM (i) and constellation of 2048-QAM (ii); (b) constellation of 4096-QAM (i) and constellation of 8192-QAM (ii).
    • Table 1. Parameters in the NNE

      View table
      View in Article

      Table 1. Parameters in the NNE

      ParameterValue
      Input size62
      Neural cells in hidden layer 140
      Neural cells in hidden layer 220
      Output size2
      Epoch20
      OptimizerAdam
      Batch size64
    • Table 2. DSM Parameters

      View table
      View in Article

      Table 2. DSM Parameters

      ParameterValue
      OSR10
      Number of quantization bits1
      Synthesizing methodChebyshev NTF
      Order6
      Max NTF gain1.55
      NTF center frequency0
    • Table 3. Important Devices Used in This Work

      View table
      View in Article

      Table 3. Important Devices Used in This Work

      DeviceModelImportant Parameters
      AWGTektronix AWG7122CSampling rate: 12 GSa/s
      ECLAgilent N7714ALinewidth: <100 kHz
      I/Q modulatorFujitsu FTM7938EZ3-dB bandwidth: 30 GHz
      LNAAT-LNA-75110Gain: 21 dB
      PAAT-PA-80135-1320GNSaturate output: 20 dBm
      EACENTELLAX OA3MHQMGain: 25 dBm
      PDFinisar XPDV412xRBandwidth: 100 GHz
      CASAC-2507-094-S2Nominal gain: 25 dBi
      OSCTektronix DSO73304DSampling rate: 100 GSa/s
    Tools

    Get Citation

    Copy Citation Text

    Bo Liu, Jianxin Ren, Xiangyu Wu, Shuaidong Chen, Yaya Mao, Li Zhao. Neural network equalization based on delta-sigma modulation[J]. Chinese Optics Letters, 2024, 22(4): 040601

    Download Citation

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

    Category: Fiber Optics and Optical Communications

    Received: Jun. 16, 2023

    Accepted: Jan. 11, 2024

    Published Online: Apr. 28, 2024

    The Author Email: Bo Liu (bo@nuist.edu.cn)

    DOI:10.3788/COL202422.040601

    Topics