Ultrahigh-order QAM signal nonlinear compensation method and system based on time characteristic memory neural network

Abstract

The invention relates to the technical field of ultrahigh-order QAM (Quadrature Amplitude Modulation) signal nonlinear compensation, in particular to an ultrahigh-order QAM signal nonlinear compensation method and system based on a time characteristic memory neural network, and the method comprises the steps: S1, obtaining a plurality of groups of orthogonal amplitude modulation signals under the condition of different optical signal-to-noise ratios, performing zero mean normalization processing on the quadrature amplitude modulation signal to obtain a pre-processed quadrature amplitude modulation signal; s2, dividing the preprocessed orthogonal amplitude modulation signal into a training set and a test set, constructing a signal nonlinear compensation model, and training the signal nonlinear compensation model by using the training set to obtain a trained signal nonlinear compensation model; and S3, based on the trained signal nonlinear compensation model, performing nonlinear compensation on a quadrature amplitude modulation signal to be compensated. According to the method, the time sequence characteristics of the signals can be extracted, the two-way time dependence relation can be captured, and the purpose of nonlinear distortion compensation is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of ultra-high order QAM signal nonlinear compensation, and in particular to an ultra-high order QAM signal nonlinear compensation method and system based on a time characteristic memory neural network. Background Art

[0002] According to the data from the authoritative "2020 Internet Report" released by Cisco, the compound annual growth rate of the number of Internet users has remained stable at about 6% from 2018 to 2023, indicating that network traffic will continue to surge in the future, thus putting forward more urgent requirements for data transmission efficiency. In this context, coherent optical communication technology has emerged in response to the challenge of traffic growth with its unique advantages.

[0003] In order to effectively improve spectral efficiency (SE), one of the strategies commonly adopted by the industry is to increase the order of signal modulation. This method has been fully verified in probability shaping (PS) experiments involving 1024-QAM and even 4096-QAM. However, the transmission practice of ultra-high-order signals is often accompanied by complex hardware deployment and high cost investment, which undoubtedly poses a significant obstacle to its widespread application. In view of this, advanced digital signal processing (DSP) algorithms are particularly important in compensating for signal damage and optimizing transmission performance.

[0004] For ultra-high-order quadrature amplitude modulation (QAM) signals, achieving effective equalization and noise suppression, especially nonlinear compensation (NLC), has become a key problem that needs to be overcome. The nonlinear characteristics caused by the Kerr effect of optical fiber set an inherent nonlinear Shannon limit for the information transmission capacity of optical fiber communication systems, especially when it interacts with the spontaneous radiation noise generated by optical amplifiers. The problem is more prominent. In addition, as the order of QAM signals increases, the points on the constellation diagram become more and more dense, resulting in a significant reduction in the Euclidean distance between points, which greatly reduces the ability of high-order signals to resist noise and linear / nonlinear damage, thereby requiring a higher optical signal-to-noise ratio (OSNR) to ensure transmission quality.

[0005] At present, mainstream nonlinear compensation technologies include digital back propagation (DBP), Volterra series-based nonlinear equalizer (VNLE), and phase conjugate double wave (PCTW). Although DBP can deal with damage by solving the back-propagation nonlinear Schrödinger equation (NLSE), its high computational complexity limits the feasibility of its commercial application. VNLE uses Volterra series transfer function for modeling, and also faces the challenges of huge consumption of computing resources and difficulty in handling high-order nonlinear effects. Although PCTW technology can effectively reduce first-order nonlinear effects through digital coherent superposition, unfortunately, it will sacrifice half of the spectrum efficiency.

[0006] In this context, neural networks have shown great potential in the field of nonlinear equalization of optical fiber communication systems with their excellent fitting and analysis capabilities. In recent years, machine learning (ML) technology has attracted much attention in solving the problem of nonlinear impairment in coherent optical communications. From the early artificial neural networks (ANN) and convolutional neural networks (CNN) used in 16-QAM to the recurrent neural networks (RNN) for 64-QAM, remarkable results have been achieved in practical applications. In particular, the successful practice of long short-term memory (LSTM) networks in mitigating impairments in 16-QAM coherent optical transmission systems further demonstrates the broad prospects of LSTM-based post-processing technology in addressing such complex challenges.

[0007] However, it is worth noting that most previous studies have focused on low-order modulated signals, and additional linear equalizers are usually required for preprocessing before the neural network structure. In the face of ultra-high-order signals, due to the sharp reduction in the Euclidean distance between constellation points, the sensitivity of the signal to linear and nonlinear damage increases sharply, which undoubtedly poses a more stringent challenge to the application of neural networks. Even a small distortion may cause the signal to decay rapidly, and the research on the application of neural networks to such high-order signals is still in its infancy, and there is still a lot of room to be explored. Summary of the invention

[0008] In order to solve the above technical problems, the present invention provides a nonlinear compensation method for ultra-high-order QAM signals based on a time feature memory neural network, comprising the following steps:

[0009] S1: Under different optical signal-to-noise ratio conditions, multiple groups of orthogonal amplitude modulation signals are acquired, and zero-mean normalization processing is performed on the orthogonal amplitude modulation signals to obtain pre-processed orthogonal amplitude modulation signals;

[0010] S2: dividing the preprocessed orthogonal amplitude modulation signal into a training set and a test set, constructing a signal nonlinear compensation model, and training the signal nonlinear compensation model using the training set to obtain a trained signal nonlinear compensation model;

[0011] S3: Based on the trained signal nonlinear compensation model, perform nonlinear compensation on the quadrature amplitude modulation signal to be compensated.

[0012] In one embodiment of the present invention, the signal nonlinear compensation model includes an input layer, a temporal convolutional network layer, a bidirectional long short-term memory neural network layer and an output layer;

[0013] The input layer processes the training set to obtain a modulated signal in the form of an embedded vector;

[0014] The temporal convolutional network layer extracts features from the modulation signal in the embedded vector form to obtain feature sequence data;

[0015] The bidirectional long short-term memory neural network layer processes the feature sequence data in a forward and reverse order respectively, captures the bidirectional time dependency therein, and obtains compensated feature sequence data;

[0016] The output layer outputs the compensated feature sequence data.

[0017] In one embodiment of the present invention, the temporal convolutional network layer is configured to include: a plurality of residual units connected in sequence, each residual unit includes a first branch and a second branch, the first branch and the second branch respectively extract features of the modulated signal in the form of the embedded vector, and perform a splicing operation on the output features of the two branches to obtain the feature sequence data.

[0018] In one embodiment of the present invention, the first branch includes a dilated convolution layer, a batch normalization layer, an activation layer, a Dropout layer, a dilated convolution layer, a batch normalization layer, an activation layer and a Dropout layer connected in sequence.

[0019] In one embodiment of the present invention, the expression of the processing process of the dilated convolutional layer on data is as follows:

[0020]

[0021] Among them, s represents the element of the input sequence x, d is the dilation factor, k is the size of the filter kernel, sd·i represents the past direction, and f is the filter set.

[0022] In one embodiment of the present invention, the activation layer uses a Tanh function.

[0023] In one embodiment of the present invention, the second branch is a residual skip branch including a 1×1 point-by-point convolutional layer.

[0024] In one embodiment of the present invention, the bidirectional long short-term memory neural network layer includes a forward long short-term memory network unit, a reverse long short-term memory network unit and a splicing operation layer;

[0025] Among them, the forward long short-term memory network unit and the reverse long short-term memory network unit process the feature sequence data in forward and reverse order respectively to obtain a forward feature sequence and a reverse feature sequence, and the splicing operation layer splices the forward feature sequence and the reverse feature sequence to obtain the compensated feature sequence data.

[0026] Based on the same inventive concept, the present invention also provides a nonlinear compensation system for ultra-high-order QAM signals based on a time-feature memory neural network, the system comprising the following modules:

[0027] A data acquisition and preprocessing module, used to acquire multiple groups of orthogonal amplitude modulation signals under different optical signal-to-noise ratios, perform zero-mean normalization processing on the orthogonal amplitude modulation signals, and obtain preprocessed orthogonal amplitude modulation signals;

[0028] A model building and training module, used for dividing the preprocessed orthogonal amplitude modulation signal into a training set and a test set, building a signal nonlinear compensation model, and training the signal nonlinear compensation model using the training set to obtain a trained signal nonlinear compensation model;

[0029] The signal nonlinear compensation module is used to perform nonlinear compensation on the orthogonal amplitude modulation signal to be compensated based on the trained signal nonlinear compensation model.

[0030] The present invention also provides a computer storage medium, which stores a computer software product. The computer software product includes several instructions for enabling a computer device to execute the instructions of the ultra-high-order QAM signal nonlinear compensation method based on time characteristic memory neural network.

[0031] The above technical solution of the present invention has the following advantages compared with the prior art:

[0032] 1. The present invention constructs a time feature memory neural network model. The method can efficiently capture and compensate for the nonlinear distortion in ultra-high-order QAM signals. This is due to the combined effect of the timing convolutional network layer and the bidirectional long short-term memory neural network layer in the model. They can respectively extract the timing features of the signal and capture the bidirectional time dependency, thereby improving the accuracy and efficiency of compensation.

[0033] 2. The temporal convolutional network layer contains multiple residual units. Each unit can extract signal features at multiple levels and angles through the combination of the first branch and the second branch. In particular, the combination of the dilated convolution layer and the 1×1 point-by-point convolution layer in the first branch not only increases the receptive field of the network, but also maintains the diversity of features, which helps to improve the generalization ability of the model.

[0034] 3. The introduction of a bidirectional long short-term memory neural network layer enables the model to process both forward and reverse feature sequence data simultaneously, thereby more comprehensively capturing the temporal dependencies in the signal. This bidirectional processing mechanism helps improve the model’s ability to understand and compensate for the dynamic characteristics of the signal. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to make the content of the present invention more clearly understood, the present invention is further described in detail below according to specific embodiments of the present invention in conjunction with the accompanying drawings, wherein:

[0036] Figure 1 It is a flow chart of a nonlinear compensation method for ultra-high-order QAM signals based on a time-feature memory neural network provided in the first embodiment of the present invention;

[0037] Figure 2 It is a schematic diagram of the structure of a signal nonlinear compensation model based on a time feature memory neural network (TFM NN);

[0038] Figure 3 It is a schematic diagram of the temporal convolutional network layer structure;

[0039] Figure 4 This is a diagram of the experimental setup for transmission of PS-PDM1024 / 4096-QAM signals over 80 km of standard single-mode optical fiber (SSMF);

[0040] Figure 5 (a) and (c) respectively represent the NGMI performance measured with / without TFMNN equalizer in a back-to-back (BTB) transmission experiment, (b) and (d) respectively represent the NGMI performance measured with traditional VE and the proposed TFMNN in an 80 km standard single-mode fiber (SSMF) transmission experiment;

[0041] Figure 6 It is a schematic diagram of the structure of the ultra-high-order QAM signal nonlinear compensation system based on the time characteristic memory neural network provided in the second embodiment of the present invention;

[0042] Description of the accompanying drawings in the specification: 100, data acquisition and preprocessing module; 200, model construction and training module; 300, signal nonlinear compensation module. DETAILED DESCRIPTION

[0043] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments so that those skilled in the art can better understand the present invention and implement it, but the embodiments are not intended to limit the present invention.

[0044] Embodiment 1

[0045] Reference Figure 1 As shown, the present invention provides a nonlinear compensation method for ultra-high-order QAM signals based on a time characteristic memory neural network, the method comprising the following steps:

[0046] S1: Under different optical signal-to-noise ratios (OSNR), multiple groups of quadrature amplitude modulation (QAM) signals are obtained. Each group of quadrature amplitude modulation signals includes the real and imaginary components of two polarization states of 10,000 symbols, and records the information of the current symbol and the surrounding environment;

[0047] Performing zero-mean normalization processing on the quadrature amplitude modulation signal to obtain a pre-processed quadrature amplitude modulation signal, wherein the pre-processed quadrature amplitude modulation signal is a pseudo-random binary sequence (PRBS);

[0048] S2: Divide the preprocessed orthogonal amplitude modulation signal into a training set and a test set, construct a temporal feature memory (TFM) neural network including an input layer, a temporal convolutional network (TCN) layer, a bidirectional long short-term memory (Bi-LSTM) neural network layer and an output layer as a signal nonlinear compensation model, use the training set to train the signal nonlinear compensation model, use a method of calculating the mean square error (MSE) every 25 epochs to monitor the learning progress of the model, and set an early stopping mechanism, that is, if the MSE does not improve significantly after 250 consecutive epochs, terminate the training in advance to prevent overfitting, and obtain the trained signal nonlinear compensation model;

[0049] S3: Based on the trained signal nonlinear compensation model, perform nonlinear compensation on the quadrature amplitude modulation signal to be compensated.

[0050] It can be seen from the above technical solution that in step S1, the present invention considers the orthogonal amplitude modulation (QAM) signal under different OSNR conditions, ensuring that the model can process signals in a variety of noise environments. Each group of QAM signals contains real and imaginary components of two polarization states, as well as information about the current symbol and its surrounding environment, which helps the model to more accurately capture the dynamic characteristics and time dependence of the signal.

[0051] In step S2, the time feature memory neural network utilizes the combined effect of the temporal convolutional network layer and the bidirectional long short-term memory neural network layer to respectively extract the temporal features of the signal and capture the bidirectional time dependency, which plays a vital role in improving the accuracy and efficiency of signal damage compensation.

[0052] In step S3, the trained model is used to perform nonlinear compensation on the QAM signal to be compensated, which can significantly reduce the nonlinear distortion in the signal and improve the transmission quality and spectrum efficiency of the signal.

[0053] like Figure 2 As shown, the signal nonlinear compensation model is a time feature memory neural network, which specifically includes an input layer, a temporal convolutional network (TCN) layer, a bidirectional long short-term memory (Bi-LSTM) neural network layer and an output layer;

[0054] The input layer processes the training set to obtain a modulated signal in the form of an embedded vector;

[0055] The temporal convolutional network layer extracts features from the modulation signal in the embedded vector form to obtain feature sequence data;

[0056] The bidirectional long short-term memory neural network layer processes the feature sequence data in a forward and reverse order respectively, captures the bidirectional time dependency therein, and obtains compensated feature sequence data;

[0057] The output layer outputs the compensated feature sequence data.

[0058] Furthermore, the temporal convolutional network layer is configured to include: a plurality of residual units connected in sequence, each residual unit includes a first branch and a second branch, the first branch and the second branch respectively extract features of the modulated signal in the form of the embedded vector, and perform a splicing operation on the output features of the two branches to obtain the feature sequence data.

[0059] Specifically, the first branch includes a dilated convolution layer, a batch normalization layer, an activation layer, a Dropout layer, a dilated convolution layer, a batch normalization layer, an activation layer, and a Dropout layer connected in sequence. Preferably, the activation layer uses a Tanh function. The second branch is a residual skip branch including a 1×1 point-by-point convolution layer.

[0060] The expression of the processing process of the dilated convolutional layer on data is as follows:

[0061]

[0062] Where s represents the element of the input sequence x, d is the dilation factor, k is the size of the filter kernel, sd·i represents the past direction, and f is the filter set. When d=1, it simplifies to a regular convolution. Larger dilation can expand the receptive field without requiring a lot of computational cost. By increasing the filter kernel size k or the dilation factor d, the receptive field of TCN will expand. For example, when the convolution kernel is 3 and the dilation factor d=o(2 i ) grows exponentially, TCN is able to cover all input values. Each layer to the next involves residual blocks, which enhances long-range dependency modeling. Therefore, TCN's dilated convolutions and residual blocks can effectively handle long-range dependencies, making it an ideal choice for tasks such as digital signal processing in coherent optical transmission.

[0063] Figure 3A dilated convolution with kernel size 3 and dilation factor d=n is shown. The receptive field of the TCN structure is determined by the network depth n, filter kernel size k, and dilation factor d. To ensure the stability of deep TCN and optimize feature extraction, residual layers are used, consisting of two causal convolutional layers and nonlinear activation layers. Batch normalization of the layers after the convolution is used to maintain stability, while dropout layers help to mitigate overfitting. To match the input and output dimensions, 1×1 convolutions are used for proper tensor alignment before element-wise addition. Weight normalization is applied to each convolutional layer in the residual block to prevent gradient explosion. To enhance generalization and reduce overfitting, L2 regularization is added to the TCN structure to constrain the L2 norm of the parameters. Each residual block consists of two convolutional layers with a stride of 1. The residual network architecture contains shortcut connections that bypass some layers and pass the raw data directly to the next layer. These additional connections do not increase the complexity of the model.

[0064] Figure 3 A dilated convolution operation with a specific configuration is intuitively shown, with a kernel size of 3 and a dilation factor d set equal to the network depth n. In the TCN (Temporal Convolutional Network) architecture, the size of the receptive field is determined by several key factors, including the network depth n, the kernel size k, and the dilation factor d. In order to enhance the stability of deep TCN and optimize its feature extraction capabilities, residual layers are designed, which consist of two causal convolutional layers and a nonlinear activation layer in between.

[0065] To maintain the stability of the model, we set up layer batch normalization after the convolution operation. At the same time, we also introduced a dropout layer to effectively reduce the risk of overfitting. To ensure that the dimensions of the input and output can be matched smoothly, we cleverly used a 1×1 convolution operation to achieve the necessary tensor alignment before adding the elements.

[0066] Inside the residual block, each convolutional layer is additionally equipped with a weight normalization mechanism, which is designed to prevent the occurrence of gradient explosion problems. In order to further improve the generalization ability of the model and reduce the risk of overfitting, the L2 regularization strategy is integrated into the TCN structure to achieve this goal by constraining the L2 norm of the model parameters. It is worth noting that each residual block is composed of two convolutional layers with a span of 1. In addition, the residual network architecture also cleverly incorporates shortcut connections (SkipConnections), which allow the original data to bypass certain layers directly and pass to the next layer. This design not only enhances the flow of information, but also does not increase the computational complexity of the model.

[0067] In summary, through this series of carefully designed components and strategies, the TCN architecture maintains high efficiency while significantly improving its stability and generalization capabilities.

[0068] By utilizing the TCN layer for feature extraction, the bidirectional long short-term memory (Bi-LSTM) neural network layer can more effectively process complex ultra-high-order QAM signals with a simple structure. The Bi-LSTM network retains the advantages of LSTM, can effectively capture time correlation, and integrates a gate mechanism to manage information flow, thereby solving the gradient vanishing problem. Its bidirectionality enhances the information integration of past and future symbols, mitigates inter-symbol interference (ISI), and provides richer temporal context.

[0069] The data flow of the Bi-LSTM layer is as follows Figure 2 As shown, the Bi-LSTM network includes a forward long short-term memory network unit, a reverse long short-term memory network unit, and a splicing operation layer;

[0070] The forward LSTM network unit and the reverse LSTM network unit process the feature sequence data in a forward and reverse order respectively to obtain a forward feature sequence h={h 0 ,h 1 ,...,h n} and the reverse signature sequence h'={h' 0 ,h' 1 ,...,h' n}, the splicing operation layer performs a splicing operation on the forward feature sequence and the reverse feature sequence to obtain the compensated feature sequence data.

[0071] In addition to performance, it is also crucial to fully evaluate the complexity of the proposed method. According to the characteristics of the TCN architecture, the complexity of the structure is calculated as follows:

[0072]

[0073] Among them, n i Represents the characteristics of the original data, n f Indicates the number of filters, n k represents the kernel size. For TCN, the size of each filter is (n s -n k +1). The complexity of the convolutional network can be expressed in terms of the number of features, the number of filters, the kernel size, and the output length L after the convolutional layer. outIn the above formula, the number of features in the first convolutional layer corresponds to the number of features in the original data. In subsequent layers, the number of features is updated according to the number of filters set in the previous convolutional layer. Each residual block contains two convolutional structures, each with a different number of filters and kernel size. To distinguish them, let n k and n f Represents the parameters of the first layer, n' k and n' f represents the parameters of the second layer. N represents the number of residual blocks. Therefore, the complexity of the proposed TFM neural network equalizer can be written as:

[0074]

[0075] Generally speaking, floating point operations per second (FLOPS) is a measure of hardware complexity. Based on the parameterization outlined in this letter, the complexity of the proposed algorithm is calculated to be 4571 FLOPS.

[0076] In view of the limitation of hardware resources, the proposed signal nonlinear compensation model (i.e., TFM neural network model) is experimentally verified in a 1024 / 4096-QAM signal transmission system at a rate of 4GBd. It is important to clarify that, Figure 4 The experimental setup shown and the designed TFM neural network model are independent of the signal transmission rate and exhibit wide applicability.

[0077] At the transmitter, the probability shaping encoder uses a pseudo-random binary sequence as input and generates the corresponding amplitude sequence through a sophisticated bit-to-symbol mapping strategy. Subsequently, pilot symbols are added in the digital signal processing (DSP) module to enhance signal quality. The data is resampled and pulse shaped by a root cosine (RRC) finite impulse response (FIR) filter with a roll-off factor of 0.05 to ensure that the signal's spectral characteristics and time domain waveform meet the transmission requirements.

[0078] The processed digital signal is fed into an arbitrary waveform generator (AWG) with a sampling rate of 8-GSa / s and a resolution of 14 bits, and converted into an analog signal through a digital-to-analog conversion (DAC) process. This analog signal drives the four-channel electrical signal input to the Mach-Zehnder modulator (MZM), which in turn generates a high-quality 1024-QAM optical signal. After power amplification by an erbium-doped fiber amplifier (EDFA), the signal strength is significantly improved. At this point, the input power into the 80-km standard single-mode fiber (SSMF) can be precisely controlled by adjusting the variable optical attenuator (VOA), where the 4096-QAM signal exhibits optimal performance at an input power of about -4dBm.

[0079] In the signal transmission path, a second VOA is set up to adjust the optical signal-to-noise ratio (OSNR) to meet the needs of bit error rate measurement; while the third VOA ensures that the optical power received by the coherent receiver remains stable. At the receiving end, an optical bandpass filter (OBPF) is first used to effectively suppress the amplified spontaneous emission (ASE) noise introduced by the EDFA, and then the signal is sent to the coherent receiver together with the light wave emitted by the local oscillator (LO). The receiver is equipped with a laser with a narrow linewidth (less than 1kHz) and a wavelength set to 1550.112nm, as well as two 90-degree optical hybrids and four balanced photodetectors to achieve efficient photoelectric conversion.

[0080] The analog-to-digital conversion (ADC) process is completed by a 50-GSa / s real-time oscilloscope, and the collected data is then subjected to a series of preprocessing steps in the offline DSP module, including traditional algorithms such as dispersion compensation, clock recovery, polarization demultiplexing, and carrier recovery to optimize signal quality. Finally, after fine equalization and judgment by the TFM NN equalizer, the original signal is successfully restored, thus verifying the effectiveness and reliability of the signal nonlinear compensation model described in the present invention.

[0081] In this invention, the performance of PS 1024 / 4096-QAM signals using ideal rate adaptive forward error correction (FEC) coding is evaluated. For probability shaped signals with different entropy values, normalized generalized mutual information (NGMI) can be used as a useful metric for channel measurement. Regardless of the modulation format used, a specific soft decision forward error correction (SD-FEC) scheme can be evaluated. This means that whether it is 1024-QAM or 4096-QAM, or even other higher-order modulation methods that may be used in the future, their error correction capabilities and system throughput under a specific SD-FEC scheme can be predicted and optimized by calculating NGMI.

[0082] Among them, the definition of NGMI is as follows:

[0083] NGMI=1-(H-GMI) / (4m)

[0084] Among them, m for 1024-QAM is 5, and m for 4096-QAM is 6. Using 20% ​​and 25% LDPC overhead, the corresponding NGMI thresholds are 0.881 and 0.778 respectively. Generally speaking, as long as the NGMI exceeds these thresholds, the QAM signal can achieve error-free transmission.

[0085] like Figure 5As shown in (a) and (c) of Figure 1, as the signal entropy value increases, the NGMI (normalized generalized mutual information) performance shows a significant downward trend. However, thanks to the excellent performance of the proposed TFM NN equalizer, the NGMI performance has been significantly improved and successfully reached the required NGMI threshold. In order to intuitively demonstrate this improvement, Figure 5 Figures (a) and (c) of Figure 1 show the constellation diagrams of 1024-QAM and 4096-QAM signals with and without the TFM NN equalizer. It can be clearly seen that with the help of the TFM NN equalizer, the originally fuzzy and scattered constellation points become converged and clear, which directly reflects the significant improvement in signal quality.

[0086] Likewise, observe Figure 4 We can also find similar performance improvement trends in (b) and (d) of Figure 1. These results show that the TFM NN equalizer has a significant effect on improving the transmission performance of high-order QAM signals.

[0087] Finally, for 1024-QAM signals, we achieved a raw spectral efficiency (SE) of up to 16.190 bits / second / Hz using an entropy of 17.0 bits / symbol while keeping the NGMI after LDPC (low-density parity-check code) encoding above the threshold of 20%. For 4096-QAM signals, when the NGMI after LDPC encoding reached the threshold of 25%, we obtained a raw SE of 21.188 bits / second / Hz by using an entropy of 21.4 bits / symbol. These results not only verify the effectiveness of the TFM NN equalizer in improving the transmission performance of high-order QAM signals, but also provide strong support for the design of future high-speed and high-density optical fiber communication systems.

[0088] In summary, the present invention innovatively proposes a temporal feature memory (TFM) neural network for PS-PDM (probability shaping-polarization multiplexing) ultra-high-order QAM (quadrature amplitude modulation) coherent optical transmission system. The experimental verification on PS-PDM 1024 / 4096-QAM signals transmitted on 80 km standard single-mode optical fiber (SSMF) fully demonstrates the advancement and practicality of the technology.

[0089] The TFM neural network equalizer designed by the present invention cleverly combines the temporal convolutional network feature extraction layer with the bidirectional long short-term memory network, significantly optimizing the performance of the latter. This innovative design enables the PS-PDM 1024 / 4096-QAM signal to be efficiently transmitted over 80 km SSMF, with original spectrum efficiencies of up to 16.190 bits / second / Hz and 21.188 bits / second / Hz, respectively.

[0090] Compared with the traditional neural network plus linear equalizer solution, the TFM neural network equalizer of the present invention not only has a simpler structure, but also shows excellent performance in compensating the nonlinear distortion of ultra-high-order QAM signals. Compared with the classic Volterra equalizer, the TFM neural network equalizer also shows significant performance advantages.

[0091] Therefore, the method proposed in the present invention provides an economical and efficient solution for the generation and transmission of ultra-high-order signals in future commercial large-capacity coherent optical transmission systems. This innovative achievement not only promotes the development of optical communication technology, but also lays a solid foundation for building a faster and more reliable optical fiber communication network.

[0092] Embodiment 2

[0093] Based on the same inventive concept as that of the first embodiment, the present invention also provides a nonlinear compensation system for ultra-high-order QAM signals based on a time-feature memory neural network. Figure 6 As shown, the system includes the following modules:

[0094] The data acquisition and preprocessing module 100 is used to acquire multiple groups of quadrature amplitude modulation signals under different optical signal-to-noise ratio conditions, and perform zero-mean normalization processing on the quadrature amplitude modulation signals to obtain preprocessed quadrature amplitude modulation signals;

[0095] The model building and training module 200 is used to divide the preprocessed quadrature amplitude modulation signal into a training set and a test set, build a signal nonlinear compensation model, and train the signal nonlinear compensation model using the training set to obtain a trained signal nonlinear compensation model;

[0096] The signal nonlinear compensation module 300 is used to perform nonlinear compensation on the orthogonal amplitude modulation signal to be compensated based on the trained signal nonlinear compensation model.

[0097] The present embodiment proposes a nonlinear compensation system for ultra-high-order QAM signals based on a time feature memory neural network, which is used to implement the aforementioned nonlinear compensation method for ultra-high-order QAM signals based on a time feature memory neural network. Therefore, the specific implementation method of the nonlinear compensation system for ultra-high-order QAM signals based on a time feature memory neural network can be seen from the embodiment part of the aforementioned nonlinear compensation method for ultra-high-order QAM signals based on a time feature memory neural network. For example, the data acquisition and preprocessing module 100, the model building and training module 200 and the signal nonlinear compensation module 300 are respectively used to implement the steps S1, S2 and S3 in the nonlinear compensation method for ultra-high-order QAM signals based on a time feature memory neural network in the first embodiment. Therefore, its specific implementation method can refer to the description of the corresponding embodiments of each part. In order to avoid redundancy, it will not be repeated here.

[0098] Embodiment 3

[0099] The present invention also provides a computer storage medium, which stores a computer software product. The computer software product includes several instructions for enabling a computer device to execute the instructions of the ultra-high-order QAM signal nonlinear compensation method based on time characteristic memory neural network described in Example 1.

[0100] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0101] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0102] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0103] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0104] Obviously, the above embodiments are merely examples for clear explanation and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived from these are still within the protection scope of the invention.

Claims

1. A nonlinear compensation method for ultra-high-order QAM signals based on a time-feature memory neural network, characterized in that: The following steps are involved: S1: Under different optical signal-to-noise ratio conditions, multiple groups of orthogonal amplitude modulation signals are acquired, and zero-mean normalization processing is performed on the orthogonal amplitude modulation signals to obtain pre-processed orthogonal amplitude modulation signals; S2: dividing the preprocessed orthogonal amplitude modulation signal into a training set and a test set, constructing a signal nonlinear compensation model, and training the signal nonlinear compensation model using the training set to obtain a trained signal nonlinear compensation model; S3: Based on the trained signal nonlinear compensation model, perform nonlinear compensation on the quadrature amplitude modulation signal to be compensated.

2. The nonlinear compensation method for ultra-high-order QAM signals based on a time-characteristic memory neural network according to claim 1, characterized in that: The signal nonlinear compensation model includes an input layer, a temporal convolutional network layer, a bidirectional long short-term memory neural network layer and an output layer; The input layer processes the training set to obtain a modulated signal in the form of an embedded vector; The temporal convolutional network layer extracts features from the modulation signal in the embedded vector form to obtain feature sequence data; The bidirectional long short-term memory neural network layer processes the feature sequence data in a forward and reverse order respectively, captures the bidirectional time dependency therein, and obtains compensated feature sequence data; The output layer outputs the compensated feature sequence data.

3. The nonlinear compensation method for ultra-high-order QAM signals based on time-feature memory neural network according to claim 2 is characterized in that: The temporal convolutional network layer is configured to include: a plurality of residual units connected in sequence, each residual unit includes a first branch and a second branch, the first branch and the second branch respectively extract features of the modulated signal in the form of an embedded vector, and perform a splicing operation on the output features of the two branches to obtain the feature sequence data.

4. The nonlinear compensation method for ultra-high-order QAM signals based on time-feature memory neural network according to claim 3 is characterized in that: The first branch includes a dilated convolution layer, a batch normalization layer, an activation layer, a Dropout layer, a dilated convolution layer, a batch normalization layer, an activation layer and a Dropout layer connected in sequence.

5. The nonlinear compensation method for ultra-high-order QAM signals based on time-feature memory neural network according to claim 4 is characterized in that: The expression of the processing process of the dilated convolutional layer on data is as follows: Among them, s represents the element of the input sequence x, d is the dilation factor, k is the size of the filter kernel, sd·i represents the past direction, and f is the filter set.

6. The nonlinear compensation method for ultra-high-order QAM signals based on time-feature memory neural network according to claim 4 is characterized in that: The Tanh function used by the activation layer.

7. The nonlinear compensation method for ultra-high-order QAM signals based on time-feature memory neural network according to claim 3 is characterized in that: The second branch is a residual skip branch including a 1×1 point-by-point convolutional layer.

8. The nonlinear compensation method for ultra-high-order QAM signals based on time-feature memory neural network according to claim 2 is characterized in that: The bidirectional long short-term memory neural network layer includes a forward long short-term memory network unit, a reverse long short-term memory network unit and a splicing operation layer; Among them, the forward long short-term memory network unit and the reverse long short-term memory network unit process the feature sequence data in forward and reverse order respectively to obtain a forward feature sequence and a reverse feature sequence, and the splicing operation layer splices the forward feature sequence and the reverse feature sequence to obtain the compensated feature sequence data.

9. A nonlinear compensation system for ultra-high-order QAM signals based on a time-characteristic memory neural network, characterized in that: Includes the following modules: A data acquisition and preprocessing module, used to acquire multiple groups of orthogonal amplitude modulation signals under different optical signal-to-noise ratios, perform zero-mean normalization processing on the orthogonal amplitude modulation signals, and obtain preprocessed orthogonal amplitude modulation signals; A model building and training module, used for dividing the preprocessed orthogonal amplitude modulation signal into a training set and a test set, building a signal nonlinear compensation model, and training the signal nonlinear compensation model using the training set to obtain a trained signal nonlinear compensation model; The signal nonlinear compensation module is used to perform nonlinear compensation on the orthogonal amplitude modulation signal to be compensated based on the trained signal nonlinear compensation model.

10. A computer storage medium, characterized in that: The computer storage medium stores a computer software product, which includes a number of instructions for enabling a computer device to execute the ultra-high-order QAM signal nonlinear compensation method based on a time characteristic memory neural network as described in any one of claims 1 to 8.

Images