In long-haul optical communication systems, compensating nonlinear effects through digital signal processing (DSP) is difficult due to intractable interactions between Kerr nonlinearity, chromatic dispersion (CD) and amplified spontaneous emission (ASE) noise from inline amplifiers. Optimizing the standard digital back propagation (DBP) as a deep neural network (DNN) with interleaving linear and nonlinear operations for fiber nonlinearity compensation was shown to improve transmission performance in idealized simulation environments. Here, we extend such concepts to practical single-channel and polarization division multiplexed wavelength division multiplexed experiments. We show improved performance compared to state-of-the-art DSP algorithms and additionally, the optimized DNN-based DBP parameters exhibit a mathematical structure which guides us to further analyze the noise statistics of fiber nonlinearity compensation. This machine learning-inspired analysis reveals that ASE noise and incomplete CD compensation of the Kerr nonlinear term produce extra distortions that accumulates along the DBP stages. Therefore, the best DSP should balance between suppressing these distortions and inverting the fiber propagation effects, and such trade-off shifts across different DBP stages in a quantifiable manner. Instead of the common ‘black-box’ approach to intractable problems, our work shows how machine learning can be a complementary tool to human analytical thinking and help advance theoretical understandings in disciplines such as optics.
We propose, numerically analyze and experimentally demonstrate a low-complexity, modulation-order independent, non-data-aided (NDA), feed-forward carrier phase recovery (CPR) algorithm. The proposed algorithm enables synchronous decoding of arbitrary square-quadrature amplitude modulation (QAM) constellations and it is suitable for a realistic hardware implementation based on block-wise parallel processing. The proposed method is based on principal component analysis (PCA) and it outperforms the well-known and widely used blind phase search (BPS) algorithm at low signal-to-noise ratio (SNR) values, showing much lower cycle slip rate (CSR) both numerically and experimentally. For operation at higher SNR values, a hybrid two-stage implementation combining the proposed method and BPS is also proposed and their performance are investigated benchmarking them against the two-stage BPS (2S-BPS). The complexity of the proposed simple and hybrid methods are evaluated against 2S-BPS and computational complexity savings of 92% and 40% are expected for the simple and hybrid methods, respectively.
9 September 2021 in Journal of Lightwave Technology
Machine Learning (ML) algorithms have shown to complement standard digital signal processing (DSP) tools in mitigating fiber nonlinearity and improving long-haul transmission performance. However, dynamic transmission impairments such as polarization effects and carrier phase noise corrupt the training data and conventional cost functions for neural network (NN) training become unsuitable. Simple cascade of ML and standard adaptive DSP blocks will also result in suboptimal transmission performance or require impractical training methodologies in presence of such dynamic transmission impairments. We show how the adaptive DSP blocks can be treated as extra stateful NN layers and be combined with the main NN so that standard backpropagation-like training algorithms in ML can be applied. In this case, the adaptive filters are viewed as NN states which are updated in the forward pass of the backpropagation. We study the combined training of linear and nonlinear parameters in the digital backpropagation (DBP) algorithm for fiber nonlinearity compensation (named generalized DBP (GDBP) hereafter), residual impairments, polarization effects, frequency offsets and carrier phase noise compensation filters as a single NN in a 7 × 288 Gb/s polarization multiplexed (PM)-16QAM transmission experiment over 1125 km. We derived the complete set of backpropagation-like gradients and state update equations for the static and dynamic parameters of the combined NN. We further proposed and open-sourced a JAX-based coding framework for their easy and practical implementation. GDBP is more generalized than other DBP variants proposed in literature and for a given total number of steps, GDBP is the first experimental demonstration of optimal single-channel DBP based-fiber nonlinearity compensation algorithm. In addition, for complexity constrained situations with shortened filter taps, GDBP enables a 1 dB performance improvement over other DBP variants.
15 September 2023 in Journal of Lightwave Technology
We review recent advances in the mitigation of inter-channel four-wave-mixing (FWM) based on “XYYX” input polarization alignment, statistical fiber zero-dispersion wavelength (ZDW) distribution, real-world fiber cable installation with randomized fiber ZDW for every fiber segment of typically 2∼3km during fiber cabling to enable high-performance O-band WDM transmission for 800G-LR4 with four 800GHz-spaced 200Gb/s PAM-4 channels reaching up to 10km, 1.6T-LR8 with eight 800GHz-spaced 200Gb/s PAM-4 channels reaching up to 10km, and 400G-ER4-30km with four 400GHz-spaced 100Gb/s PAM-4 channels reaching up to 30 km. In addition, a novel unequal channel spacing scheme for co-propagating wavelength channels is shown to effectively mitigate the FWM impairment in a bidirectional 5G fronthaul transmission system with twelve 800GHz-spaced 25Gb/s NRZ channels reaching up to 20 km. These inter-channel FWM mitigation techniques are expected to aid the standardization of intensity-modulation and direct-detection (IM/DD) based high-speed O-band WDM systems in both IEEE and ITU-T, aiming to cost-effectively and energy-efficiently support future data center, 5G, and metro-access networks.
10 December 2021 in Journal of Lightwave Technology
Cylindrical vector (CV) mode as the eigenmodeof optical fibers is a strong candidate for high-capacity mode-division multiplexing (MDM) systems. Based on the Kramers–Kronig (K-K) reception, we experimentally demonstrate a 448-Gb/s MDM transmission system using four CV eigenmodes ( $\boldsymbol{|l|=2}$ ) over 3-km few-mode fiber (FMF), each mode carrying 28-GBaud 16-quadrature amplitude modulation (16-QAM) signals. In conventional MDM systems with ${\boldsymbol{M}}$ channels, blind ${\boldsymbol{N \times {N}}}$${\boldsymbol{(N\leq {M})}}$ multiple-input multiple-output (MIMO) technique is implemented to demultiplex and equalize ${\boldsymbol{N}}$ channels after transmission. However, without appropriately initializing finite impulse response (FIR) filter taps in MIMO, the singularity may show up and result in ${\boldsymbol{N}}$ MIMO outputs converged to ${\boldsymbol{P}}$${\boldsymbol{(P< {N})}}$ independent outputs with information missing. Here we employ the ${\boldsymbol{4\times 4}}$ multi-user constant modulus algorithm (MU-CMA) technique to avoid such singular problems in our MDM demonstration. To verify the reliability of MU-CMA, we test 7,408 signal sets (598 static sets and 13-hour 6,810 dynamic sets), and none of the singularities is observed. Specifically, 390 static sets out of 598 are added with an extra 3-dB mode-dependent loss (MDL), and the results of these sets prove the MDL-resistance of MU-CMA. Furthermore, the bit error rate (BER), MDL, and mode coupling for each mode of the proposed system are also investigated statically and dynamically. In the static case without extra MDL, all four CV mode channels can achieve ${\boldsymbol{7\%}}$ hard-decision forward error correction (FEC) threshold at BER of ${\boldsymbol{3.8\times 10^{-3}}}$ , under certain optical-to-signal noise ratio (OSNR). In dynamic cases, the averaged BER of four CV modes can also approach the FEC BER threshold. These experimental results verify the robustness of K-K reception and MU-CMA in CV-mode MDM systems. They also highlight the reliability and stability of CV-mode MDM systems, which show the enormous potential of the CV-mode MDM scheme employed in high-speed short-reach networks.
We experimentally compare QoT estimations for WDM systems using Machine Learning(ML) and GN-based analytical models. ML estimates the side channels with better accuracy but is temporally less stable and less generalizable to different link configurations.