A pruning method of artificial neural network based nonlinear equalizer (ANN-NLE) is proposed and validated for single-sideband 4-ary pulse amplitude modulation (SSB-PAM4) in IM/DD system. As a classifier, ANN capable to form complex boundary among different classifications, which considered as an appropriate way mitigate the impairments optical communication In this paper, first, we introduce operation principle traditional linear (LE) NLE such volterra (VE). Then make analogy LE, VE...

An intelligent optical performance monitor using multi-task learning based artificial neural network (MTL-ANN) is designed for simultaneous OSNR monitoring and modulation format identification (MFI). Signals' amplitude histograms (AHs) after constant module algorithm are selected as the input features MTL-ANN. The experimental results of 20-Gbaud NRZ-OOK, PAM4 PAM8 signals demonstrate that MTL-ANN could achieve MFI simultaneously with higher accuracy stability compared single-task ANNs...

We propose a cascaded neural network (NN) to simultaneously identify the modulation formats and monitor optical-signal-to-noise ratio (OSNR). In second-level network, it is single deep NN (DNN) rather than multiple sub-networks, which makes architecture more compact can save resource for real implementation. However, since data set constituted from all formats, universality be guaranteed but not accuracy complexity. To accelerate estimation process improve accuracy, we introduce transfer...

Supported by advanced digital signal processing algorithms and application specific integrated circuits, coherent receivers in elastic optical networks will be capable of measuring link impairments real time. Specifically, can work as soft performance monitors. Optical spectra usually contain rich information about links have been exploited to assist failure detection identification. However, acquiring needs the deployment numerous spectrum analyzers. Instead, received signals are easy...

We propose and experimentally demonstrate a method of optical signal-to-noise ratio (OSNR) monitoring modulation format identification (MFI) using binarized convolutional neural network (B-CNN) in coherent receiver. The proposed technique automatically extracts OSNR dependent features from the signals' ring constellation maps. A group schemes including nine quadrature amplitude (QAM) formats are selected as transmission signals. experimental results show that MFI accuracy can reach 100%...

We have experimentally demonstrated the signal-to-signal beating interference (SSBI) mitigation capability of Kramers-Kronig (KK) detection in a single sideband (SSB) four-ary pulse-amplitude-modulation (PAM4) direct (DD) system. A 112-Gbit/s (56-GBaud) SSB-PAM4 optical signal suitable for KK is generated by using commercial, low-cost dual-drive Mach-Zehnder modulator. The system bandwidth constraint mitigated Nyquist pulse shaping and precompensation. SSBI algorithm identified comparing...

A novel low-complexity digital signal processing solution is presented to compensate the impairments of limited bandwidth and nonlinearity. It based on equalizer, adaptive noise-whitening postfilter, maximum likelihood sequence detection (MLSD). The system performance loss induced by higher noise power at band edges after equalizer can be mitigated postfilter MLSD. Given above structure, it possible enrich providing nonlinear compensation capability. memory polynomial (MPE), instead Volterra...

We have experimentally demonstrated a single-photodiode 112-Gbit/s 16-QAM transmission over 960-km standard single mode fiber with Kramers-Kronig (KK) detection and sparse I/Q Volterra filter (VF). KK is used to mitigate the signal beating interference. A novel VF based on dual-input real-valued series proposed overcome constraints of non-linear distortions. Two conventional single-input VFs are employed as contrast prove necessity hybrid structure in when system suffers from severe...

In this paper, we comprehensively investigate the performance of error-feedback noise shaping (EFNS) in low-resolution high-speed intensity-modulation direct-detection (IM/DD) PAM4 and coherent DP-mQAM transmission systems by simulation experiments. The influence EFNS parameters digital to analog converter (DAC) on is first numerically investigated a 56-GBaud 2-km IM/DD system. combination clipping also evaluated impact overloading deeply analyzed. For transmission, for different modulation...

A loss weight adaptive multi-task learning based artificial neural network (MTL-ANN) is applied for joint optical signal-to-noise ratio (OSNR) monitoring and modulation format identification (MFI). We conduct an experiment of polarization division multiplexing (PDM) coherent system with 5 km standard single mode fiber (SSMF) transmission to verify this monitor. group schemes including nine M-QAM formats are selected as the signals. Instead circular constellation, signals' amplitude...

We investigate the overestimation trap of ANN from learning rule PRBS over unrepeated pattern. The influence input symbols and hidden neurons is studied by simulations. Besides, overestimated gain experimentally evaluated.

In this paper, the application of a clipping and digital resolution enhancer (DRE) is numerically investigated in 56-GBaud direct-detection four-ary pulse amplitude modulation (PAM4) transmission systems. The influence tap length digital-to-analog converter (DAC) samples per symbol on DRE gain first studied. After optimizing probability, sampling jitter introduced. combination can increase tolerance by ∼1.7% time for 4-bit DAC. Then, required optical signal-to-noise ratio (OSNR) at...

In this paper, we have experimentally demonstrated the feasibility of space-division-multiplexed 3 × multiple-input multiple-output (MIMO) transmission over a single OM3 multimode fiber (MMF) using commercial IEEE 802.11 n/ac access points. Throughput performance for different length links under range bending conditions was evaluated. To study throughput improvement by MIMO technology, single-input single-output (SISO), 2 and separate SMF also measured comparison. It is shown that,...

In this paper, we propose a low-complexity storage-reduced digital spectrum-based soft failure management solution including detection (SFD), identification (SFI) and magnitude estimation (FME). Five failures are considered. Random fluctuation of key link parameters is introduced in simulations to investigate the robustness proposed solution. To reduce computational complexity storage requirement for spectrum calculation, Welch’s method employed instead applying fast Fourier transform on...

The analog photonics link (APL) is widely used in microwave photonics. However, wideband and multi-carrier systems, the third inter-modulation distortion (IMD3) cross-modulation (XMD) will jointly limit spurious-free dynamic range (SFDR) of links. In this paper, we experimentally present a linearized APL, which IMD3 XMD are mitigated simultaneously by using artificial neural networks with transfer learning (TL-ANN). experiment, different networks, trained knowledge obtained from two- or...

Machine learning is regarded as an attractive solution for soft failure management in optical networks; however, the performance of trained models working on unseen data growing concern, due to scarce historical and high training costs. Hence, issues reducing improving generalization have received considerable attention. In this paper, a detection (SFD) identification scheme proposed based digital residual spectrum that leverages auto-encoder (AE) support vector machine. The acquired by...

We propose a highly generalized soft failure identification model based on digital residual spectrum and autoencoder. In case of random fluctuation, it could reach accuracy more than 97.61% for five different transmission distances.

We propose a polarization dependent loss (PDL) and chromatic dispersion (CD) insensitive, low-complexity adaptive equalizer (AEQ) for short-reach coherent optical transmission systems. The AEQ contains 1-tap butterfly finite impulse response (FIR) filter two N-tap FIR filters. It first performs demultiplexing using the filter, of which coefficients are obtained based on Stokes space. Then it mitigates inter-symbol interference (ISI) filters adjust filter’s by utilizing constant modulus...

We investigate the performance of error-feedback noise shaping (EFNS) in low-resolution IM/DD and coherent systems by simulations experiments. EFNS shows comparable to digital resolution enhancer has advantages complexity & processing delay.

We propose a novel artificial neural network nonlinearity equalizer based on multiple layers perceptron(MLP-ANN NLE) for16-quadrature amplitude modulation (16-QAM) coherent optical orthogonal frequency division multiplexing (CO-OFDM) system. The scheme is effective for compensating the four-wave mixing effect among subcarriers. Compared with situation of non-compensated nonlinearity, this proposed can increase Q factor by 1.4 dB and 0.6 at 600 km 2000 respectively transmission distance be...

As one of the most important advanced packaging technologies, heterogeneous integration is desired to help node cover integrated systems' memory wall further continue Moore's law. With continuous growing demands for high performance computing, multi-chip wafer level developed artificial intelligence. By shrinking interconnection distance and enlarging scale, System-on-Wafer (SoW) a promising architecture improving system performance. In this paper, with historically large scale over 200 dies...