Chinese Optics Letters, Volume. 22, Issue 6, 060602(2024)

Channel estimation-based time-frequency neural network for post-equalization in underwater visible light communication

Haoyu Zhang1... Li Yao1, Chaoxu Chen1, Yuan Wei1, Chao Shen1,2,3, Jianyang Shi1,2,3, Junwen Zhang1,2,3, Ziwei Li1,2,3, and Nan Chi1,23,* |Show fewer author(s)
Author Affiliations
  • 1Key Laboratory for Information Science of Electromagnetic Waves (MoE), Department of Communication Science and Engineering, School of Information Science and Technology, Fudan University, Shanghai 200433, China
  • 2Shanghai CIC of LEO Satellite Communication Technology, Fudan University, Shanghai 200433, China
  • 3Shanghai ERC of LEO Satellite Communication and Application, Fudan University, Shanghai 200433, China
  • show less
    Figures & Tables(7)
    Architecture of CBV-TFNet post-equalizer.
    Experimental setup of UVLC system.
    (a) BER performance of different PBs and passband weights used by loss function. Spectra of the original, received, and NN equalized signal of (b) using PB of 723 MHz and weight of 1.0 and (c) using PB of 797 MHz and weight of 0.9.
    (a) Error band diagram of bitrate in continuously changing epoch using different loss function and model input. STFT spectrum of (b) received signal, (c) original signal, (d) received signal after mask.
    The bitrate contour plot of different working points with the post-equalizer using (a) LMS-Volterra; (b) DNN; (c) TFDNet; (d) CBV-TFNet. The bit-power loading result in the communication test with the post-equalizer using (e) LMS-Volterra; (f) DNN; (g) TFDNet; (h) CBV-TFNet.
    Comparison of consecutive window input and output capability of DNN based and CBV-TFNet equalizers.
    • Table 1. Hyperparameters and Communication Performance of Different Methods-Based Equalizers

      View table
      View in Article

      Table 1. Hyperparameters and Communication Performance of Different Methods-Based Equalizers

      MethodWindow lengthNodes of NN structureNumber of real multiplicationsPeak bitrate (Gbps)Dynamic range (4.75 Gbps threshold)
      LMS-Volterra737205/sym4.5160
      DNN73(73, 256, 1)37,705/sym4.7740.216
      TFDNet72(144, 256, 128, 144)16,276/sym4.8550.373
      CBV-TFNet (this work)72(144, 200, 128, 144)14,383/sym4.9560.491
    Tools

    Get Citation

    Copy Citation Text

    Haoyu Zhang, Li Yao, Chaoxu Chen, Yuan Wei, Chao Shen, Jianyang Shi, Junwen Zhang, Ziwei Li, Nan Chi, "Channel estimation-based time-frequency neural network for post-equalization in underwater visible light communication," Chin. Opt. Lett. 22, 060602 (2024)

    Download Citation

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

    Category: Fiber Optics and Optical Communications

    Received: Dec. 1, 2023

    Accepted: Feb. 22, 2024

    Posted: Feb. 22, 2024

    Published Online: Jun. 18, 2024

    The Author Email: Nan Chi (nanchi@fudan.edu.cn)

    DOI:10.3788/COL202422.060602

    CSTR:32184.14.COL202422.060602

    Topics