Aspects of the present disclosure describe systems, methods and structures for optical fiber nonlinearity compensation using neural networks that advantageously employ machine learning (ML) algorithms for nonlinearity compensation (NLC) that advantageously provide a system-agnostic model independent of link parameters, and yet still achieve a similar or better performance at a lower complexity as compared with prior-art methods. Systems, methods, and structures according to aspects of the present disclosure include a data-driven model using the neural network (NN) to predict received signal nonlinearity without prior knowledge of the link parameters. Operationally, the NN is provided with intra-channel cross-phase modulation (IXPM) and intra-channel four-wave mixing (IFWM) triplets that advantageously provide a more direct pathway to underlying nonlinear interactions.

A method and system are herein disclosed. A coherent optical receiver receives a first optical data carrier signal at a first instant of time and a second optical data carrier signal at a second instant of time, generates at least four first data streams from the first optical data carrier signal and at least four second data streams from the second optical data carrier signal; and circuitry calculates a first aggregate power of the first data streams and a second aggregate power of the second data streams; applies an adjustable temporal low pass filter to the first aggregate power and the second aggregate power resulting in a compensation power, the adjustable temporal low pass filter adjusted to achieve a performance metric; and phase-rotates the first data streams and the second data streams proportional to the compensation power.

An apparatus includes a digital signal processor to perform perturbation-based optical nonlinearity compensation of optical data signals of a communication stream. The digital signal processor includes first digital circuits to calculate multiplicative factors for corrections to the optical data signals from products of values of the optical data signals at a reduced set of times. The reduced set is a down-sampling of the sequence of consecutive symbol times of the communication stream. The digital signal processor also includes second digital circuits to calculate the multiplicative factors for corrections to the optical data signals at the consecutive symbol times by interpolating the multiplicative factors evaluated at the reduced set of times.

A system for transmission of an optical signal, the system including an optical coupler for splitting said signal into a first copy and a second copy. The optical coupler has an input for receiving the optical signal, a first output for the first copy and a second output for the second copy. The system also includes a first optical guide connected to the first output, a second optical guide connected to the second output and a superposition module for coherently superimposing the first copy and the second copy of the signal.

Embodiments of this disclosure provide an impairment monitoring apparatus, impairment monitoring and compensating system and method. As a parameter of a impairment of a receiver end and/or a parameter of an impairment of a transmitter end is/are determined according to a preset period or a preset condition to perform compensation or calibration, complexity of calculation and power consumption of the system may be efficiently lowered. And as the parameter of the impairment of the receiver end and the parameter of the impairment of the transmitter end are not changed rapidly, determining these parameters according to the preset period or the preset condition to perform compensation or calibration will not affect an effect of the compensation or calibration, thereby ensuring a performance of the system.

The disclosed systems, structures, and methods are directed to monitoring performance of optical networks. Transmitted and received optical channel signal has both low-power gaps and an amplitude modulation pilot tone applied. The low-power gaps are applied at a gap frequency with gap power being lower than a signal power of the optical channel signal. The pilot tone is applied to the optical channel signal at a pilot tone modulation frequency, which is different from the gap frequency. Described methods include determining pilot tone modulation depth based on detected gap power in low-power gaps of the received optical channel signal. Amplifier spontaneous emission and nonlinear noise in optical link are detected separately. This permits determining and monitoring of optical signal-to-noise ratio.

Systems and methods are provided for adaptive equalization, for example for use with coherent optical reception. The equalizer has a loop for updating taps of an adaptive linear filter forming part of the equalizer. In the loop, an error calculator calculates an error, a gradient calculator calculates a gradient of the error or filtered error. One or more gradient filters are used to filter the gradient, and the filtered gradient is used to update the taps of the adaptive linear filter. A reduction in self oscillation in the equalizer is achieved by separating the frequencies with channel features that change with time and scaling them up to speed up convergence of the equalizer.

A method and apparatus of fiber nonlinear noise monitoring and an optical receiver is provided. The method includes: acquiring a training set including at least two received signal samples obtained after a transmitting signal passes different simulation transmission scenarios, calculating amplitude noise feature values after a digital signal processing, calculating statistics values of fiber nonlinear noises, marking the amplitude noise feature values; taking the amplitude noise feature values as input of a noise monitoring model, taking marks to which the amplitude noise feature values correspond as target output, and training the noise monitoring model according to samples in the training set, so as to obtain a trained noise monitoring model, and inputting a plurality of amplitude noise feature values of received signals to be monitored into the trained noise monitoring model, so as to obtain statistics values of fiber nonlinear noises to which the received signals to be monitored correspond.

An optical transmitter apparatus is disclosed. The apparatus includes a processor, a memory coupled to the processor, and one or more programs configured to be executed by the processor. The programs include instructions for nonlinearity estimation that characterizes nonlinearity in an optical communication and estimates an amount of symbol distortion caused by the nonlinearity, instructions for selecting and mapping symbols to provide, for the nonlinearity estimation, only symbols that meet predetermined nonlinearity criteria, and instructions for storing, in the memory, the amount of symbol distortion to be used for a nonlinearity pre-compensation.

A signal processing circuit includes: a processor configured to adjust phases of reception samples which is supplied at a supply interval, according to a phase adjustment amount; and a processing circuit including a finite impulse response (FIR) filter with taps and configured to process, by the FIR filter, each of the reception samples and output output symbols at an output interval different from the supply interval, the processor is configured to: derive initial values of tap coefficients for the respective taps; and derive the phase adjustment amount such that a center of centroids of the tap coefficients at respective output time points of the output symbols coincides with a center of a number of taps of the FIR filter, the tap coefficients at respective output time points of the output symbols being set according to a deviation between the supply interval and the output interval and the initial values.