• Acta Optica Sinica
  • Vol. 43, Issue 7, 0706004 (2023)
Yong Chen1、*, Zhiqian Wu1, Huanlin Liu2, Chenyi Hu1, Jinlan Wu1, and Chuangshi Wang1
Author Affiliations
  • 1Key Laboratory of Industrial Internet of Things & Network Control, Ministry of Education, Chongqing University of Posts and Telecommunications, Chongqing 400065, China
  • 2School of Communication and Information Engineering, Chongqing University of Posts and Telecommunications, Chongqing 400065, China
  • show less
    DOI: 10.3788/AOS221812 Cite this Article Set citation alerts
    Yong Chen, Zhiqian Wu, Huanlin Liu, Chenyi Hu, Jinlan Wu, Chuangshi Wang. Neural-Network-Based Channel Estimation Method for Visible Light Communication Systems[J]. Acta Optica Sinica, 2023, 43(7): 0706004 Copy Citation Text show less
    Structure diagram of ACO-OFDM VLC system
    Fig. 1. Structure diagram of ACO-OFDM VLC system
    Channel estimation scheme of ACO-OFDM based on DNN
    Fig. 2. Channel estimation scheme of ACO-OFDM based on DNN
    Overall structure of DNN
    Fig. 3. Overall structure of DNN
    Schematic of GC optimization process. (a) Schematic of GC optimization; (b) visualization of the effect of GC method
    Fig. 4. Schematic of GC optimization process. (a) Schematic of GC optimization; (b) visualization of the effect of GC method
    Optimization process of GC-RMSprop
    Fig. 5. Optimization process of GC-RMSprop
    MSE of different DNN models varying with the iterations
    Fig. 6. MSE of different DNN models varying with the iterations
    BER of different methods at 8 and 64 pilots. (a) 8 pilots; (b) 64 pilots
    Fig. 7. BER of different methods at 8 and 64 pilots. (a) 8 pilots; (b) 64 pilots
    MSE of different methods at 8 and 64 pilots. (a) 8 pilots; (b) 64 pilots
    Fig. 8. MSE of different methods at 8 and 64 pilots. (a) 8 pilots; (b) 64 pilots
    BER and MSE performance of GC-DNN method under different number of pilots. (a) BER; (b) MSE
    Fig. 9. BER and MSE performance of GC-DNN method under different number of pilots. (a) BER; (b) MSE
    BER of different methods with or without CP
    Fig. 10. BER of different methods with or without CP
    ParameterValue
    Refractive index1.5
    Viewing angle /(°)70
    Half power angle /(°)60
    Photoelectric conversion efficiency0.51
    Receiving area /cm21
    Single LED emission power /W0.05
    Table 1. Parameter setting of LOS Lambertian communication model
    ParameterValue
    Learning rate η0.001
    Fraction of gradient decay ρ0.9
    Computability guard ε10-7
    Maximum number of epochs500
    Table 2. Parameter setting of GC-DNN training phase
    Yong Chen, Zhiqian Wu, Huanlin Liu, Chenyi Hu, Jinlan Wu, Chuangshi Wang. Neural-Network-Based Channel Estimation Method for Visible Light Communication Systems[J]. Acta Optica Sinica, 2023, 43(7): 0706004
    Download Citation