• Acta Optica Sinica
  • Vol. 41, Issue 14, 1406003 (2021)
Ziyue Zhu1、2, Mengxin Zhao2, Yichen Zhang1, and Jian Chen1、2、*
Author Affiliations
  • 1Key Laboratory of Specialty Fiber Optics and Optical Access Networks, Shanghai University, Shanghai 200444, China
  • 2Joint International Research Laboratory of Specialty Fiber Optics and Advanced Communication, Shanghai University, Shanghai 200444, China
  • show less
    DOI: 10.3788/AOS202141.1406003 Cite this Article Set citation alerts
    Ziyue Zhu, Mengxin Zhao, Yichen Zhang, Jian Chen. MIMO Equalization Technology Based on Neural Network in High-Speed IM-DD Mode Division Multiplexing Transmission System[J]. Acta Optica Sinica, 2021, 41(14): 1406003 Copy Citation Text show less
    Structure of the 2×2 IM-DD MDM transmission system
    Fig. 1. Structure of the 2×2 IM-DD MDM transmission system
    Structure of the NN-based equalizer
    Fig. 2. Structure of the NN-based equalizer
    Algorithm flowchart of the NN-based equalizer
    Fig. 3. Algorithm flowchart of the NN-based equalizer
    Structure of the proposed DFNN equalizer
    Fig. 4. Structure of the proposed DFNN equalizer
    Experimental setup for IM-DD MDM
    Fig. 5. Experimental setup for IM-DD MDM
    BER versus number of taps. (a) BER versus L1 by setting L2=101; (b) BER versus L2 by setting L1=161,201,201
    Fig. 6. BER versus number of taps. (a) BER versus L1 by setting L2=101; (b) BER versus L2 by setting L1=161,201,201
    BER performance comparison of LP11a and LP11b for obtb transmission at the transmission rate of 2×50 Gbit/s
    Fig. 7. BER performance comparison of LP11a and LP11b for obtb transmission at the transmission rate of 2×50 Gbit/s
    BER performance comparison of LP11a and LP11b for 1000-m FMF transmission at the transmission rate of 2×50 Gbit/s
    Fig. 8. BER performance comparison of LP11a and LP11b for 1000-m FMF transmission at the transmission rate of 2×50 Gbit/s
    BER performance comparison under different bit rates. (a) obtb transmission; (b) 1000-m FMF transmission
    Fig. 9. BER performance comparison under different bit rates. (a) obtb transmission; (b) 1000-m FMF transmission
    AlgorithmComplexityMultiplication
    LMSCLMS=2S+2E+2160+60+2=222
    NNCNNtrain=Nepoch(Ntapsnhid1+nhid1nhid2+nhid2no)CNNpredict=Ntapsnhid1+nhid1nhid2+nhid2noCNNtrain=116160CNNpredict=3872
    DFNNCDFNNtrain=Nepoch(Ntapsnhid1+nhid1nhid2+nhid2no)CDFNNpredict=Ntapsnhid1+nhid1nhid2+nhid2noCDFNNtrain=116160CDFNNpredict=3872
    Table 1. Computation complexity comparison between LMS, NN, and DFNN
    Ziyue Zhu, Mengxin Zhao, Yichen Zhang, Jian Chen. MIMO Equalization Technology Based on Neural Network in High-Speed IM-DD Mode Division Multiplexing Transmission System[J]. Acta Optica Sinica, 2021, 41(14): 1406003
    Download Citation