Hidden-layers Extended Recurrent Neural Network Equalizer for Short Reach Optical Interconnects

Caoyang LIU; Lin SUN; Jiawang XIAO; Bangning MAO; Ning LIU

doi:10.3788/gzxb20225112.1206001

Acta Photonica Sinica, Volume. 51, Issue 12, 1206001(2022)

Hidden-layers Extended Recurrent Neural Network Equalizer for Short Reach Optical Interconnects

Caoyang LIU¹, Lin SUN¹, Jiawang XIAO¹, Bangning MAO², and Ning LIU^1、*

¹Jiangsu New Optical Fiber Technology and Communication Network Engineering Research Center，School of Electronic and Information Engineering，Soochow University，Suzhou Jiangsu 215006，China

²College of Optical and Electronic Technology，China Jiliang University，Hangzhou 310018，China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(11)

Fig. 1. Schematic of the single hidden layer NN equalizers

Download full size

View in Article

Fig. 2. Schematic of the multiple hidden layers RNNE

Download full size

View in Article

Fig. 3. Verification setup of the multiple hidden layers RNNE for a 112-Gbps optical PAM-4 link

Download full size

View in Article

Fig. 4. BER vs. multiplication numbers for two hidden layers RNNE

Download full size

View in Article

Fig. 5. BER vs. multiplication numbers for multiple hidden layers RNNE

Download full size

View in Article

Fig. 6. Comparison of multiplication numbers of multiple hidden layers RNNE with the same hidden neurons number

Download full size

View in Article

Fig. 7. BER vs. received optical power for single and two hidden layers RNNEs

Download full size

View in Article

Fig. 8. Eye diagrams at -8 dBm ROP

Download full size

View in Article

Table 1. Parameters of optical PAM-4 link
View table
View in Article
Table 1. Parameters of optical PAM-4 link
Parameter Value
Modulation PAM⁃4
Signal bit rate 112 Gbps
Wavelength 1 550 nm
Fiber length 20 km
Dispersion coefficient 16 ps/（nm·km）
Fiber attenuation 0.16 dB/km

Table 2. N_RNN of multiple hidden layers RNNE at different thresholds
View table
View in Article
Table 2. N_RNN of multiple hidden layers RNNE at different thresholds
BER threshold
N_RNN of
one hidden layer
N_RNNof
two hidden layers
N_RNNof
three hidden layers
10 ^-2 204 187 189
3.8 $\times$ 10 ^-3 283 239 247
10 ^-3 429 329 349

Table 3. BER of multiple hidden layers RNNE with the same hidden neurons number
View table
View in Article
Table 3. BER of multiple hidden layers RNNE with the same hidden neurons number
Number of neurons per layer
BER of
one hidden layer
BER of
two hidden layers
BER of
three hidden layers
6 1.22 $\times$ 10^-2 4.85 $\times$ 10^-3 2.80 $\times$ 10^-3
7 6.90 $\times$ 10^-3 1.80 $\times$ 10^-3 1.30 $\times$ 10^-3
8 5.00 $\times$ 10^-3 1.10 $\times$ 10^-3 7.11 $\times$ 10^-4

Tools

Get Citation

Copy Citation Text

Caoyang LIU, Lin SUN, Jiawang XIAO, Bangning MAO, Ning LIU. Hidden-layers Extended Recurrent Neural Network Equalizer for Short Reach Optical Interconnects[J]. Acta Photonica Sinica, 2022, 51(12): 1206001

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites