In this work, we demonstrate that the high-accuracy computation of the continuous nonlinear spectrum can be performed by using artificial neural networks. We propose the artificial neural network (NN) architecture that can efficiently perform the nonlinear Fourier (NF) optical signal processing. The NN consists of sequential convolution layers and fully connected output layers. This NN predicts only one component of the continuous NF spectrum, such that two identical NNs have to be used to predict the real and imaginary parts of the reflection coefficient. To train the NN, we precomputed 94035 optical signals. 9403 signals were used for validation and excluded from training. The final value of the relative error for the entire validation dataset was less than 0.3%. Our findings highlight the fundamental possibility of using the NNs to analyze and process complex optical signals when the conventional algorithms can fail to deliver an acceptable result.
KEYWORDS: Nonlinear optics, Field programmable gate arrays, Signal processing, Data processing, Telecommunications, Optical signal processing, Digital signal processing, Transmittance, Statistical analysis, Receivers
The nonlinear Fourier transform (NFT) based signal processing has attracted considerable attention as a promising tool for fibre nonlinearity mitigation in optical transmission. However, the mathematical complexity of NFT algorithms and the noticeable distinction of the latter from the “conventional” (Fourier-based) methods make it difficult to adapt this approach for practical applications. In our work, we demonstrate a hardware implementation of the fast direct NFT operation: it is used to map the optical signal onto its nonlinear Fourier spectrum, i.e. to demodulate the data. The main component of the algorithm is the matrix-multiplier unit, implemented on field-programmable gate arrays (FPGA) and used in our study for the estimation of required hardware resources. To design the best performing implementation in limited resources, we carry out the processing accuracy analysis to estimate the optimal bit width. The fast NFT algorithm that we analyse, is based on the FFT, which leads to the O(N log22 N) method’s complexity for the signal consisting of N samples. Our analysis revealed the significant demand in DSP blocks on the used board, which is caused by the complex-valued matrix operations and FFTs. Nevertheless, it seems to be possible to utilise further the parallelisation of our NFT-processing implementation for the more efficient NFT hardware realisation.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.