A method and system for monitoring optical performance in a PON

Abstract

The application provides a PON optical performance monitoring method and system, the method comprising: obtaining a received signal transmitted through a PON network to reach a receiving end and after digital signal processing, obtaining system parameters related to a PON network communication system; performing fractional Fourier transform on the received signal to construct a three-dimensional spectrum diagram, and collecting a two-dimensional projection image of the three-dimensional spectrum diagram on an "amplitude-order" plane; taking the two-dimensional projection image and the system parameters as inputs of a pre-trained multi-modal fusion convolutional neural network-full connection neural network CNN-FCNN, a convolutional neural network branch performing feature extraction on the input two-dimensional projection image, and a full connection network branch performing feature extraction on the input system parameters; a fusion layer fusing the features of different modalities extracted, and outputting an optical signal-to-noise ratio estimation value and a received optical power estimation value at an output layer.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of optical fiber communication technology, and in particular to a PON optical performance monitoring method and system. BACKGROUND

[0002] With the explosive growth of data traffic and the rapid development of cloud computing and big data applications, people's demand for high-speed data center optical interconnection is increasing. Passive optical interconnection (POI) has become an important development direction of current data center network architecture due to its unique advantages. In the POI architecture, passive optical network (PON) has become an important development direction of current data center network architecture due to its unique advantages. As an efficient optical fiber access technology, passive optical network can meet people's demand for high-speed, low-cost broadband access services, and has been widely used in home broadband, enterprise access and other scenarios, becoming a key technology to promote the upgrading of communication infrastructure and support the development of smart cities and Internet of Things.

[0003] However, due to the passive nature of PON, the internal link of data center is intensive, and the signal is inevitably affected by loss, dispersion and nonlinear effects during transmission. After long-distance transmission and passive splitting of optical signal, the optical signal-to-noise ratio (OSNR) may be severely affected, which in turn affects the reliability of the communication system. Therefore, real-time monitoring and evaluation of link quality is crucial to ensure the stable operation of data center.

[0004] As an important indicator in optical performance monitoring (OPM), optical signal-to-noise ratio (OSNR) is crucial for optimizing PON network resource allocation and improving communication system transmission performance.

[0005] Researchers have proposed various OSNR monitoring methods for optical communication systems, including Statistical Moments-Based (SMB), Error Vector Magnitude (EVM), Asynchronous Amplitude Histogram (AAH), differential pilot method, spectral analysis method, and deep learning-based OSNR monitoring method. The SMB method only relies on the amplitude envelope of the received signal, effectively extracts high-dimensional features of the signal, and efficiently separates the signal and noise, which has attracted widespread attention from many scholars at home and abroad. However, the SMB method does not fully consider the impact of nonlinear phase noise on OSNR monitoring, resulting in larger OSNR monitoring errors for high-order modulation format systems such as QAM. At the same time, the existing SMB method usually uses a fitting method to obtain a closed solution, which severely limits the robustness of the method in OSNR monitoring under different system parameters PON.

[0006] In recent years, deep learning has been widely applied in many fields due to its powerful feature extraction capability. Domestic and foreign scholars have proposed many deep learning-based OSNR monitoring methods. However, the lack of interpretability of neural networks and their low accuracy on out-of-sample data have triggered widespread controversy in the academic community about their scientificity and reliability.

[0007] Therefore, how to design a PON optical performance monitoring method and system to overcome the problem of poor interpretability of traditional neural networks and low accuracy on out-of-sample data is a technical problem to be solved. SUMMARY

[0008] In view of this, the embodiments of the present application provide a PON optical performance monitoring method to eliminate or improve one or more defects in the prior art.

[0009] In one aspect of the present application, a PON optical performance monitoring method is provided, which comprises the following steps: obtaining a received signal after digital signal processing of a signal transmitted through a PON network to a receiving end; obtaining system parameters related to a PON network communication system; performing fractional Fourier transform on the received signal to construct a three-dimensional spectrum graph, and collecting a two-dimensional projection image of the three-dimensional spectrum graph on an "amplitude-order" plane; taking the two-dimensional projection image and the system parameters as inputs of a pre-trained multi-modal fusion convolutional neural network-full connection neural network (CNN-FCNN), a convolutional neural network branch of the CNN-FCNN extracts features from the input two-dimensional projection image, and a full connection network branch of the CNN-FCNN extracts features from the input system parameters; a fusion layer of the CNN-FCNN fuses the features of different modalities extracted by the convolutional neural network branch and the full connection network branch, and outputs an optical signal-to-noise ratio (OSNR) estimation value and a received optical power estimation value at an output layer; wherein, in the training process of the CNN-FCNN, an OSNR reference value measured by an optical spectrum analyzer located at the receiving end and a received optical power reference value measured by an optical power meter are taken as labels to construct a training set; and a loss in the PON network signal transmission is calculated by a nonlinear Schrödinger equation as an equation loss, an OSNR estimation error is calculated based on the OSNR estimation value and the OSNR reference value, a received optical power estimation error is calculated based on the received optical power estimation value and the received optical power reference value, and a weighted sum of the equation loss, the OSNR estimation error and the received optical power estimation error is taken as a loss function for training the CNN-FCNN.

[0010] In some embodiments of the present application, the system parameters include baud rate, transmission power and transmission distance.

[0011] In some embodiments of the present application, the convolutional neural network branch includes a preset number of convolution-maximum pooling units, each group of convolution-maximum pooling units includes a preset number of convolution layers and a maximum pooling layer, the preset number of convolution layers are used to extract features from the input two-dimensional projection image, and the maximum pooling layer is used for down-sampling; and the full connection network branch includes a single full connection layer containing a preset number of neurons.

[0012] In some embodiments of the present application, the fractional Fourier transform is performed on the received signal to construct a three-dimensional spectrum graph, and a two-dimensional projection image of the three-dimensional spectrum graph on an "amplitude-order" plane is collected, which comprises: performing fractional Fourier transform of different orders on the received signal to construct a three-dimensional spectrum graph; for the three-dimensional spectrum graphs constructed by fractional Fourier transform of different orders, the maximum amplitude corresponding to the order is taken, thereby obtaining a two-dimensional projection image of the three-dimensional spectrum graph on an "amplitude-order" plane.

[0013] In some embodiments of the present application, the method further comprises: subtracting the received optical power estimation value and the optical signal-to-noise ratio estimation value to obtain a noise optical power estimation value.

[0014] In some embodiments of the present application, the expression of the equation loss is:

[0015] ;

[0016] wherein, and respectively represent the real part and the imaginary part of a given complex number, the expression of the equation loss is:

[0017] ;

[0018] wherein, represents a loss coefficient, represents a nonlinear coefficient, represents a distance of the optical signal propagating along an axial direction in the optical fiber, represents a received optical power reference value, represents a noise optical power reference value, represents an imaginary unit.

[0019] In some embodiments of the present application, the loss function for training the CNN-FCNN is mathematically expressed as:

[0020] ;

[0021] wherein, respectively are weight coefficients of the equation loss, the optical signal-to-noise ratio estimation error and the received optical power estimation error, represents the equation loss, represents the optical signal-to-noise ratio estimation error, represents the received optical power estimation error.

[0022] Corresponding to the above method, the present application further provides a PON optical performance monitoring system, comprising a processor, a memory and computer programs / instructions stored in the memory, located at a receiving end of a PON network, the processor being configured to execute the computer programs / instructions, and when the computer programs / instructions are executed, the system implements the steps of the method according to any one of the above embodiments.

[0023] Corresponding to the above method, the present application further provides a computer readable storage medium having computer programs / instructions stored thereon, and when the computer programs / instructions are executed by a processor, the steps of the method according to any one of the above embodiments are implemented.

[0024] Corresponding to the above method, the application also provides a computer program product comprising computer programs / instructions which, when executed by a processor, implement the steps of the method according to any one of the above embodiments.

[0025] The PON optical performance monitoring method and system provided by the application construct a loss function based on a nonlinear Schrodinger equation, thereby introducing physical information into a neural network model, which is beneficial to overcoming the poor interpretability of the neural network in the prior art, and by collecting a two-dimensional projection image of a three-dimensional spectrum on an 'amplitude-order' plane as an input of a CNN-FCNN network model, it is beneficial to reducing the complexity of the model without losing key information, and by introducing multi-modal data containing system parameters and the two-dimensional projection image to extract a multi-modal feature map, it is beneficial to solving the problem of low accuracy of the neural network on out-of-sample data.

[0026] Additional advantages, objects, and features of the application will be set forth in part by the description that follows, and will become apparent to those skilled in the art upon examination of the following detailed description and drawings in which

[0027] Those skilled in the art will appreciate that the objects and advantages of the application can be realized and attained by means summarized fully in the detailed description that follows, particularly when considered in light of the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0028] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description, serve to explain the principles of the application. The components in the drawings are not necessarily to scale, emphasis instead being placed upon illustrating the principles of the application. Portions of some embodiments may be exaggerated relative to others for the purpose of better illustration.

[0029] Figure 1 A flow chart of a PON optical performance monitoring method in an embodiment of the application.

[0030] Figure 2 A flow chart of a coherent optical fiber communication system for realizing optical performance monitoring in an embodiment of the application.

[0031] Figure 3 A CNN-FCNN neural network model structure for realizing PON optical performance monitoring in an embodiment of the application.

[0032] Fig. 4(a) is a schematic diagram of a received signal after digital signal processing of a PON network transmission to a receiving end in an embodiment of the present application.

[0033] Fig. 4(b) is a schematic diagram of a three-dimensional spectrum constructed after fractional Fourier transform of a received signal in an embodiment of the present application.

[0034] Fig. 4(c) is a schematic diagram of a two-dimensional projection image of a three-dimensional spectrum on an "amplitude-order" plane in an embodiment of the present application.

[0035] Figure 5 Fig. 5 is a loss function variation curve of PON optical performance monitoring in a coherent fiber communication system in an embodiment of the present application.

[0036] Fig. 6(a) is an optical signal-to-noise ratio estimation error and received optical power estimation error of PON optical performance monitoring in a coherent fiber communication system with a baud rate of 25 Gbaud and a DP-QSPK modulation mode in an embodiment of the present application.

[0037] Fig. 6(b) is an optical signal-to-noise ratio estimation error and received optical power estimation error of PON optical performance monitoring in a coherent fiber communication system with a baud rate of 30 Gbaud and a DP-QSPK modulation mode in an embodiment of the present application.

[0038] Fig. 6(c) is an optical signal-to-noise ratio estimation error and received optical power estimation error of PON optical performance monitoring in a coherent fiber communication system with a baud rate of 40 Gbaud and a DP-QSPK modulation mode in an embodiment of the present application.

[0039] Fig. 6(d) is an optical signal-to-noise ratio estimation error and received optical power estimation error of PON optical performance monitoring in a coherent fiber communication system with a baud rate of 50 Gbaud and a DP-QSPK modulation mode in an embodiment of the present application.

[0040] Fig. 7(a) is a comparison curve of optical signal-to-noise ratio monitoring performance of PON optical performance monitoring and a conventional algorithm in a coherent fiber communication system with a baud rate of 25 Gbaud in an embodiment of the present application.

[0041] Fig. 7(b) is a comparison curve of optical signal-to-noise ratio monitoring performance of PON optical performance monitoring and a conventional algorithm in a coherent fiber communication system with a baud rate of 30 Gbaud in an embodiment of the present application.

[0042] Fig. 7(c) is a comparison curve of optical signal-to-noise ratio monitoring performance of PON optical performance monitoring and a conventional algorithm in a coherent fiber communication system with a baud rate of 40 Gbaud in an embodiment of the present application.

[0043] FIG7( d ) is a comparison curve of the optical signal-to-noise ratio monitoring performance of the PON optical performance monitoring in a coherent optical fiber communication system with a baud rate of 50 Gbaud according to an embodiment of the present invention and that of the traditional algorithm.

[0044] FIG8( a ) shows the optical signal-to-noise ratio estimation error and received optical power estimation error of PON optical performance monitoring in a coherent optical fiber communication system with a baud rate of 25 Gbaud and a modulation mode of DP-16QAM in one embodiment of the present invention.

[0045] FIG8( b ) shows the optical signal-to-noise ratio estimation error and received optical power estimation error of PON optical performance monitoring in a coherent optical fiber communication system with a baud rate of 30 Gbaud and a modulation mode of DP-16QAM in one embodiment of the present invention.

[0046] FIG8( c ) shows the optical signal-to-noise ratio estimation error and received optical power estimation error of PON optical performance monitoring in a coherent optical fiber communication system with a baud rate of 40 Gbaud and a modulation mode of DP-16QAM in one embodiment of the present invention.

[0047] FIG8( d ) shows the optical signal-to-noise ratio estimation error and received optical power estimation error of PON optical performance monitoring in a coherent optical fiber communication system with a baud rate of 50 Gbaud and a modulation mode of DP-16QAM in one embodiment of the present invention. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments and the accompanying drawings. Here, the exemplary embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0049] It should also be noted that, in order to avoid obscuring the present invention due to unnecessary details, the accompanying drawings only show structures and / or processing steps closely related to the solutions according to the present invention, while other details that are not closely related to the present invention are omitted.

[0050] It should be emphasized that the term "include / comprises" when used herein refers to the existence of features, elements, steps or components, but does not exclude the existence or addition of one or more other features, elements, steps or components.

[0051] It should also be noted that, unless otherwise specified, the term "connection" herein may refer not only to a direct connection but also to an indirect connection involving an intermediate.

[0052] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. In the accompanying drawings, the same reference numerals represent the same or similar components, or the same or similar steps.

[0053] In order to solve the problems of insufficient explainability of the existing deep learning network model optical signal-to-noise ratio monitoring method and low accuracy on out-of-sample data, the method based on statistical moments has a large OSNR monitoring error of high-order modulation format system due to the neglect of the phase factor, which can be solved by constructing a physical information driven optical signal-to-noise ratio monitoring model, using a physics-informed neural network (PINN) to integrate physical constraints into a neural network model, and then realizing intelligent optical performance diagnosis. Among them, the physics-informed neural network (PINN) can be used to integrate physical constraints into the neural network model, which can be widely used in the solution of Navier-Stokes equation in fluid mechanics, the solution of Maxwell equation set in electromagnetism, and the channel modeling of optical fiber communication system based on nonlinear Schrödinger equation. By constructing a physical information driven optical signal-to-noise ratio monitoring model, the explainability of the model can be significantly enhanced, and the accurate prediction of out-of-sample data is possible, which provides an efficient, reliable and physically interpretable optical performance intelligent diagnosis scheme for PON network.

[0054] Figure 1 The PON optical performance monitoring method flowchart in an embodiment of the present application. The optical performance monitoring method comprises the following steps:

[0055] Step S110: acquiring the received signal transmitted through the PON network to the receiving end and processed by the digital signal processing, and acquiring the system parameters related to the PON network communication system.

[0056] Step S120: performing fractional Fourier transform on the received signal to construct a three-dimensional spectrum, and collecting a two-dimensional projection image of the three-dimensional spectrum on the "amplitude-order" plane.

[0057] Step S130: taking the two-dimensional projection image and the system parameters as the input of the pre-trained multi-modal fusion convolutional neural network-full connected neural network (in the specification, it is abbreviated as CNN-FCNN network, Convolutional Neural Network-Fully Connected Network), the convolutional neural network branch of CNN-FCNN extracts features from the input two-dimensional projection image, and the fully connected network branch of CNN-FCNN extracts features from the input system parameters.

[0058] Step S140: The fusion layer of the CNN-FCNN fuses the features of different modalities extracted by the convolutional neural network branch and the fully connected network branch, and outputs the optical signal-to-noise ratio (OSNR) estimation value and the receiver of power (ROP) estimation value at the output layer.

[0059] In the training process of the CNN-FCNN, the optical signal-to-noise ratio reference value measured by the spectrum analyzer located at the receiving end and the received optical power reference value measured by the optical power meter are used as labels to construct a training set; and the loss in the signal transmission of the PON network is calculated by the nonlinear Schrödinger equation as an equation loss, the optical signal-to-noise ratio estimation error is calculated based on the optical signal-to-noise ratio estimation value and the optical signal-to-noise ratio reference value, the received optical power estimation error is calculated based on the received optical power estimation value and the received optical power reference value, and the weighted sum of the equation loss, the optical signal-to-noise ratio estimation error and the received optical power estimation error is used as a loss function for training the CNN-FCNN.

[0060] In the specific implementation process, the execution subject of the present scheme is located at the receiving end, for example, it can be a server end, and more specifically, it can be a server end optical network unit (ONU) device located at the server end.

[0061] The PON optical performance monitoring method and system proposed in the present application construct a loss function based on the nonlinear Schrödinger equation, thereby introducing physical information into the neural network model, which is helpful to overcome the poor interpretability of the neural network in the prior art, and by collecting a two-dimensional projection image of the three-dimensional spectrum on the "amplitude-order" plane as the input of the CNN-FCNN network model, it is helpful to reduce the complexity of the model without losing key information, and by introducing multi-modal data containing system parameters and two-dimensional projection images to extract multi-modal feature maps, it is helpful to solve the problem of low accuracy of the neural network on out-of-sample data.

[0062] The key point of the present scheme is that in the neural network of the CNN-FCNN architecture, a loss function is constructed based on the nonlinear Schrödinger equation, which is constructed based on physical information, thereby introducing physical information into the neural network model.

[0063] In some embodiments of the present application, the system parameters include baud rate, transmission power and transmission distance. It should be noted that the present scheme is not limited thereto, and the above system parameters refer to system parameters that can be used to characterize the signal transmission characteristics of the PON network, for example, the system parameters can also include channel bandwidth, etc.

[0064] The embodiment of the application can introduce multi-modal data, thereby extracting multi-modal features, and the introduction of the multi-modal features is beneficial to solving the problem of low accuracy of the neural network on out-of-sample data.

[0065] In some embodiments of the application, the convolutional neural network branch includes a preset number of convolution-maximum pooling units, each group of convolution-maximum pooling units includes a preset number of convolution layers and a maximum pooling layer, the preset number of convolution layers are used for feature extraction on the input two-dimensional projection image, and the maximum pooling layer is used for down-sampling.

[0066] In some embodiments of the application, the fully connected network branch includes a single fully connected layer including a preset number of neurons.

[0067] In the specific implementation process, the convolutional neural network branch includes two groups of convolution-maximum pooling units, each group of convolution-maximum pooling units includes 64 convolution layers with a 3*3 convolution kernel and a maximum pooling layer with a step of 2*2. The fully connected network branch can use a single fully connected layer including 32 neurons.

[0068] The application is not limited to this, and the number of groups of the two groups of convolution-maximum pooling units or the size of the maximum pooling unit can be flexibly adjusted as needed.

[0069] The above two embodiments of the application can be freely combined, and a specific implementation mode of the CNN-FCNN network is given.

[0070] In some embodiments of the application, the step S120 includes: performing fractional Fourier transform on the received signal to construct a three-dimensional spectrum, and collecting a two-dimensional projection image of the three-dimensional spectrum on an "amplitude-order" plane, including: (1) performing fractional Fourier transform on the received signal to construct a three-dimensional spectrum; (2) for the three-dimensional spectrum constructed by the fractional Fourier transform of different orders, taking the maximum amplitude corresponding to the order, thereby obtaining the two-dimensional projection image of the three-dimensional spectrum on the "amplitude-order" plane.

[0071] In the embodiment of the application, the two-dimensional projection image of the three-dimensional spectrum on the "amplitude-order" plane is collected as the input of the CNN-FCNN network model, which is beneficial to reducing the complexity of the model without losing key information.

[0072] In some embodiments of the application, the method further includes: subtracting the received optical power estimation value and the optical signal-to-noise ratio estimation value to obtain a noise optical power estimation value.

[0073] Based on the above inventive embodiments, the noise optical power estimation value can also be used to represent the optical performance monitoring situation of the PON network, and the parameters used to represent the optical performance can be any one or combination of the optical signal-to-noise ratio, the received optical power and the noise optical power.

[0074] Further, in order to better embody the present application, the mathematical expression of the equation loss is given as follows:

[0075] ;

[0076] wherein, and respectively represent the real part and the imaginary part of a given complex number, The expression of is:

[0077] ;

[0078] wherein, represents a loss coefficient, represents a nonlinear coefficient, represents the distance of the optical signal propagating along the axial direction in the optical fiber, represents a received optical power reference value, represents a noise optical power reference value, represents an imaginary unit.

[0079] Further, the loss function used for training the CNN-FCNN is given, and the mathematical expression is:

[0080] ;

[0081] wherein, respectively are the weight coefficients of the equation loss, the optical signal-to-noise ratio estimation error and the received optical power estimation error, represents the equation loss, represents the optical signal-to-noise ratio estimation error, represents the received optical power estimation error.

[0082] In order to better verify the present application, the optical performance monitoring method proposed by the present application is verified through a simulation experiment.

[0083] Firstly, the optical fiber communication system needs to be built, and the original data required for training the CNN-FCNN network is collected.

[0084] Generally, an optical fiber communication system includes a sending end, an optical fiber transmission link, and a receiving end. Their functions are as follows: (1) the sending end mainly includes a bit sequence generation module, a digital-to-analog conversion module, and an optoelectronic modulator; (2) the optical fiber transmission link includes an optical fiber, an optical power amplifier, and an adjustable optical attenuator; and (3) the receiving end mainly includes an optical spectrum analyzer, an optical power meter, a receiver, and a digital signal processing module. The process of signal propagation through the optical fiber communication system includes: (1) at the sending end, a series of binary bit sequences randomly generated by the bit sequence generation module are first modulated into digital symbols of a specific modulation format, then the digital-to-analog conversion module converts these digital symbol sequences into analog signals, and the optoelectronic modulator modulates the analog signals onto an optical carrier to realize optoelectronic conversion; (2) at the optical fiber transmission link, after the optical signal is transmitted through a certain span of optical fiber, the optical power amplifier is used to preamplify the optical signal, and the adjustable optical attenuator is used to adjust the size of the signal optical power; (3) at the receiving end, the optical spectrum analyzer is used to collect the optical signal-to-noise ratio (OSNR) of the signal light, the optical power meter is used to collect the received optical power of the signal light, then the receiver restores the received optical signal into an electrical signal, and the digital signal processing module performs digital signal processing; the digital signal processing module includes, in sequence, IQ imbalance compensation and quadrature normalization, dispersion compensation, clock recovery, channel dynamic equalization, and polarization demultiplexing.

[0085] A coherent optical fiber communication system is a specific implementation of an optical fiber communication system, and the coherent optical fiber communication system includes, in structure, a core switching node, an optical fiber transmission link, and a server-side optical network unit (ONU) device. Their functions are as follows: (1) the core switching node is responsible for processing data traffic and transmitting data to the ONU device through a passive optical network; (2) the PON transmission link distributes signals to multiple server racks through the optical fiber transmission link and a passive optical splitter; and (3) the server-side ONU device is responsible for receiving downlink optical signals, completing optoelectronic conversion, and performing signal demodulation, and mainly includes a receiver, a digital signal processing module, an optical power measurement module, and an OSNR monitoring module.

[0086] The process of signal propagation through a coherent optical fiber communication system includes: (1) At the core switching node, a series of binary bit sequences randomly generated by a bit sequence generation module are first modulated into digital symbols of a specific modulation format. Subsequently, the digital-to-analog conversion module converts these digital symbol sequences into analog signals and modulates them onto an optical carrier using an optoelectronic modulator to achieve optoelectronic conversion; (2) On the optical fiber transmission link, the optical signal is distributed to multiple server ends through a passive optical network. When the optical signal reaches the server end, the optical performance may be attenuated (characterized by the optical signal-to-noise ratio). Therefore, it is necessary to monitor the optical performance of the received optical signal in the server-end ONU device to evaluate the transmission quality of the link; (3) At the server end, the optical power measurement module and the optical performance monitoring module are used to collect the received optical power and optical signal-to-noise ratio of the signal light, respectively. The received optical signal is then restored to an electrical signal through the receiver and digital signal processing is performed through the digital signal processing module (digital signal processing includes steps such as IQ imbalance compensation and orthogonal normalization, dispersion compensation, clock recovery, channel dynamic equalization, and polarization demultiplexing). The received signal after dynamic channel equalization and polarization demultiplexing and its corresponding system parameters such as baud rate, transmit power, and transmission distance are collected. During the training process, system parameters that also need to be collected include received optical power and optical signal-to-noise ratio to facilitate the subsequent construction of the data set.

[0087] Figure 2 This is a flowchart for implementing optical performance monitoring in a coherent optical fiber communication system according to one embodiment of the present invention. First, the coherent optical fiber communication system transmits and receives signals. At the receiving end, data is collected to obtain a received signal after dynamic channel equalization and polarization demultiplexing, along with corresponding system parameters such as baud rate, transmit power, transmission distance, received optical power, and optical signal-to-noise ratio (OSNR). Using the received optical power and OSNR as labels, other system parameters related to the PON network communication system and the received signal are obtained as input to a CNN-FCNN network. The received signal is processed using a Fourier transform of varying order to obtain a two-dimensional projection image. The labels, 2D projection image, and system parameters are then input into a multimodal fusion convolutional neural network (CNN-FCNN) that integrates physical information. The CNN-FCNN network is then trained using a loss function based on the nonlinear Schrödinger equation.

[0088] In the specific implementation process, in step S120, it is necessary to perform a fractional Fourier transform on the received signal and collect a two-dimensional projection image of its three-dimensional spectrum on the "amplitude-order" plane. The process specifically includes:

[0089] The schematic diagram of fractional domain Fourier transform two-dimensional projection image generation is as follows Figure 3 Shown: The signal collected by the receiving end Perform fractional Fourier transform, which The order fractional Fourier transform is defined as:

[0090] (1)

[0091] wherein, is the kernel function of fractional Fourier transform, defined as:

[0092] (2)

[0093] wherein, , by performing fractional Fourier transform of different orders on the , a three-dimensional spectrum is constructed:

[0094] (3)

[0095] For fractional Fourier transform of different orders, take the maximum amplitude corresponding to the order,

[0096] (4)

[0097] is the two-dimensional projection image of the three-dimensional spectrum obtained by fractional Fourier transform in the "amplitude-order" plane.

[0098] In step S130, the system parameters including baud rate, transmission power and transmission distance and the acquired two-dimensional projection image are taken as input, the optical signal-to-noise ratio measured by the optical spectrum analyzer and the received optical power measured by the optical power meter are taken as labels, a data set is constructed, and the CNN-FCNN network is trained based on the constructed data set to predict the optical signal-to-noise ratio and the received optical power.

[0099] In one embodiment of the present application, the structure of the CNN-FCNN network includes a convolutional neural network branch (referred to as CNN branch) and a fully connected network branch (referred to as FCNN branch), and the acquired two-dimensional projection image is taken as the input of the CNN branch, and the system parameters including baud rate, transmission power and transmission distance are taken as the input of the FCNN branch. Different branches are trained and features are extracted respectively, and then multi-modal learning is realized by fusion training.

[0100] Figure 3 is the CNN-FCNN neural network model structure for realizing PON optical performance monitoring in one embodiment of the present application. The specific structure includes:

[0101] (1) For the CNN branch, first, a convolutional layer containing 64 3x3 convolutional kernels is used to perform preliminary feature extraction on the image input, i.e., two-dimensional picture data, and then a max-pooling layer with a step size of 2x2 is used to downsample the feature map obtained by preliminary feature extraction, thereby reducing the computational overhead as much as possible without losing key features, and then another set of convolution-max-pooling units is used to further extract high-dimensional features. The extracted high-dimensional features are flattened through a hidden layer and input to the fusion layer for fusion processing with the high-dimensional features extracted by the FCNN branch.

[0102] (2) For the FCNN branch, a single fully connected layer containing 32 neurons is used to achieve efficient extraction of data input features. Among them, Rs, LP and D represent the baud rate, transmission power and transmission distance in the system parameters, respectively.

[0103] (3) The fusion layer (Dropout) adopts a fully connected structure to splice the different modal features extracted by the CNN branch and the FCNN branch, fully excavate and correlate the cross-modal information through linear transformation, realize multi-modal learning through fusion training, and then obtain the OSNR estimation value and the estimation value of the received optical power in the output layer. Moreover, the model introduces a dynamic learning rate and an early stopping mechanism. During the training process, when the test set loss does not decrease for 5 consecutive training periods, the learning rate is reduced to 50% of the current value, and when the validation set loss is lower than 0.2, the early stopping mechanism is triggered. If the test set loss does not decrease for 8 consecutive training periods, the training is terminated in advance. The combination of dynamic learning rate and early stopping mechanism effectively prevents model overfitting, reduces redundant training iterations, reduces computational overhead, and ensures the optimal performance of the model.

[0104] (4) The constructed data set is input into the network for training, and the trained model can output the optical signal-to-noise ratio estimation value and the received optical power estimation value. Furthermore, the noise power estimation value can be calculated based on . .

[0105] As shown in Figure 3 , the input data collected in the coherent optical fiber communication system includes two-dimensional projection images and system parameters. The input data is input into the multi-modal fusion convolutional neural network-full connection neural network that fuses physical information, a loss function is constructed based on the nonlinear Schrödinger equation, and the multi-modal fusion convolutional neural network-full connection neural network is trained based on the loss feedback.

[0106] In the specific implementation process, the derivation process of the equation loss in the loss function based on the nonlinear Schrödinger equation is as follows:

[0107] First, the noise power estimate can be obtained from the optical signal-to-noise ratio estimate and the received optical power estimate output by the CNN-FCNN model as follows:

[0108] (5)

[0109] wherein are the estimates of the optical signal-to-noise ratio, the received optical power and the noise optical power, respectively.

[0110] The Nonlinear Schrödinger Equation (NLSE) provides a theoretical basis for studying the loss, dispersion and nonlinear effects of optical signals in the process of optical fiber transmission, is essential for understanding the signal evolution in optical fiber, and has become an indispensable analysis tool in optical fiber communication systems. The mathematical expression of the NLSE is:

[0111] (6)

[0112] Since the received signal is collected after dispersion compensation, the NLSE can be rewritten as:

[0113] (7)

[0114] wherein represent the loss coefficient, the dispersion coefficient and the nonlinear coefficient, respectively, represents the distance of the optical signal propagating along the axial direction in the optical fiber, is the complex envelope of the signal optical field, and the group delay expression is:

[0115] (8)

[0116] wherein is the ratio of the signal optical power to the noise power, and are the real part and the imaginary part of the standard normalized optical field amplitude , respectively, and the standard normalized optical field amplitude is defined as:

[0117] (9)

[0118] wherein and linear and nonlinear parts of the phase noise, respectively; when the amplifier spontaneous emission noise (ASE) propagates independently in the fiber, the nonlinear effect does not change the statistical properties of the ASE noise, when the signal and the ASE noise propagate together in the fiber link, the nonlinear distortion caused by the interaction between the signal and the ASE noise makes the statistical properties of the signal deviate from the Gaussian distribution, meanwhile, when the fiber is modeled as a zero-memory nonlinear (ZMNL) system, it can be considered that the ASE noise is uniformly distributed in the fiber link, and the signal and the noise are statistically independent of each other; therefore, the phase noise caused by the signal and the ASE noise can be separated by a stochastic differential equation:

[0119] (10)

[0120] wherein the first term represents deterministic non-phase noise caused by the Kerr effect, and the second term represents random phase noise introduced by the ASE noise, including linear and nonlinear phase noise introduced by the Kerr effect represented by formula (11), and linear and nonlinear phase noise introduced by the ASE noise represented by formula (12):

[0121] (11)

[0122] (12)

[0123] wherein γ represents a nonlinear coefficient, represents a received optical power reference value, and details are shown in formula (2) , represents a distance of the optical signal propagating along an axial direction in the fiber, represents linear phase noise introduced by the ASE noise, represents nonlinear phase noise introduced by the ASE noise.

[0124] Formula (12) is an Itô integral, and a closed solution cannot be directly obtained; a traditional solving method is to use Monte Carlo simulation to numerically calculate the statistical properties of the Itô integral by generating a large number of random samples, which has a large calculation overhead in a high-dimensional scenario, converges slowly, and may be affected by insufficient sample size or pseudo-randomness; without loss of generality, the Itô integral is approximately determined, and formula (12) can be rewritten as:

[0125] ; (13)

[0126] Further, the signal light field complex envelope expression of formula (8) can be rewritten as:

[0127] ; (14)

[0128] Substitute the above formula into formula (7), the expression of equation loss is:

[0129] (15)

[0130] wherein, and respectively represent the real part and the imaginary part of a given complex number, The expression of is:

[0131] (16)

[0132] The neural network loss includes equation loss , optical signal-to-noise ratio estimation error and received optical power estimation error three parts, defined as:

[0133] (17)

[0134] are weight coefficients of equation loss, optical signal-to-noise ratio estimation error and received optical power estimation error, respectively, and the optical signal-to-noise ratio estimation error and the received optical power estimation error are defined as:

[0135] (18)

[0136] wherein, and are the optical signal-to-noise ratio reference value measured by the optical spectrum analyzer and the received optical power reference value measured by the optical power meter, and respectively represent the optical signal-to-noise ratio estimation value and the received optical power estimation value; by calculating the model loss, repeatedly iterating and constantly optimizing the network parameters, the ideal optical signal-to-noise ratio estimation value and the received optical power estimation value can be obtained.

[0137] For the method of collecting two-dimensional projection images proposed in step S120, refer to FIGS. 4(a)-(c) for details, wherein FIG. 4(a) is a schematic diagram of the received signal after the PON network transmission reaches the receiving end and is processed by a digital signal in an embodiment of the present application, FIG. 4(b) is a schematic diagram of the three-dimensional spectrum constructed after the received signal is subjected to fractional Fourier transform in an embodiment of the present application, and FIG. 4(c) is a schematic diagram of the two-dimensional projection image of the three-dimensional spectrum collected in the "amplitude-order" plane in an embodiment of the present application.

[0138] Firstly, Fig. 4(a) shows the signal form of the PON communication system under two modulation formats (16QAM and QPSK respectively). Fig. 4(b) shows the signal collected by the receiving end after fractional Fourier transform, showing the three-dimensional structure in the dimensions of fractional order p, fractional domain u and amplitude. Further, Fig. 4(c) shows the projection result in the "fractional order p-amplitude" plane.

[0139] Figure 5 Fig. 8 is the comparison result of the optical performance monitoring method proposed in the present application and other methods through experiments. The specific description is as follows:

[0140] Figure 5 The loss function change curve of the PON optical performance monitoring in the coherent optical fiber communication system in an embodiment of the present application includes three curves of training set loss, test set loss and equation loss. It can be seen that after the 68th training, the validation set loss fails to decrease for 5 consecutive training periods, the dynamic learning rate mechanism is triggered, and the learning rate is reduced to 50% of the original, effectively realizing the further decrease of the validation set loss. When the validation set loss decreases below the threshold of 0.2, the early stopping mechanism is activated, and the model terminates the training in advance under the condition that the loss does not improve for 8 consecutive training periods. After the early stopping mechanism is activated, the best model is saved. At the same time, it can be seen from the equation loss change curve that in the initial stage of training, the model pays more attention to reducing the equation loss, but as the training proceeds, the optical signal-to-noise ratio estimation error is continuously reduced, resulting in a certain degree of increase in the equation loss and then tending to be stable. However, the equation loss of the model is very low overall, and considering that the equation modeling, data acquisition and other factors inevitably lead to certain errors, it can be considered that the modeling of the nonlinear Schrödinger equation signal light field complex envelope is reasonable and relatively accurate.

[0141] It should be noted that 400km, 600km, 800km, 1000km and 1200km in Figs. 6-8 refer to different system parameter coherent optical fiber communication systems, which are distinguished by transmission distance here.

[0142] Figs. 6(a)-(d) are the optical signal-to-noise ratio estimation error (OSNR estimation error for short) and the received optical power estimation error (received power estimation error for short) of the PON optical performance monitoring in the coherent optical fiber communication system with the baud rate of 25 Gbaud, 30 Gbaud, 40 Gbaud and 50 Gbaud and the modulation mode of DP-QSPK in an embodiment of the present application. The estimation error is in units of Db, and the value is an absolute value. As shown in FIG. 6, it can be seen that the model achieves ideal estimation performance at 25, 30, 40 Gbaud, and the average OSNR estimation error is 0.11 dB, 0.13 dB and 0.14 dB respectively. Further, 50 Gbaud data is collected to test the OSNR monitoring performance of the PON optical performance monitoring method based on the physical information neural network for out-of-sample data. The experimental data shows that the average OSNR estimation error is 0.32 dB. The PON optical performance monitoring method based on the physical information neural network embeds the nonlinear Schrödinger equation into the loss function to guide the neural network training with physical constraints, enhances the interpretability of the neural network, and improves the optical performance monitoring accuracy for out-of-sample data.

[0143] FIGS. 7(a)-(d) are curves of the optical signal-to-noise ratio monitoring performance comparison between the PON optical performance monitoring method and the conventional algorithm under the coherent optical fiber communication system with baud rates of 25 Gbaud, 30 Gbaud, 40 Gbaud and 50 Gbaud, wherein the horizontal coordinate is the OSNR reference value, and the vertical coordinate is the OSNR estimation error. The optical performance monitoring method disclosed in the present application is compared with the conventional statistical moment-based (SMB) method and the baud rate tolerant OSNR monitoring method (BRTM), and different PINN models are used to distinguish different system parameter coherent optical fiber communication systems (400 km, 600 km, 800 km, 1000 km and 1200 km). It can be seen that the optical performance monitoring method disclosed in the present application considers the influence of the phase noise caused by the nonlinear effect on the OSNR monitoring, and introduces physical constraints in the training process, which shows higher OSNR estimation accuracy compared with the conventional method in the strong nonlinear and different baud rate scenarios.

[0144] FIGS. 8(a)-(d) are curves of the optical signal-to-noise ratio estimation error (OSNR estimation error) and the received optical power estimation error (received optical power estimation error) of the PON optical performance monitoring method in the coherent optical fiber communication system with baud rates of 25 Gbaud, 30 Gbaud, 40 Gbaud and 50 Gbaud and the modulation mode of DP-16QAM. The optical performance monitoring method disclosed in the present application is used to test the OSNR estimation error and the received optical power estimation error of the optical performance monitoring method under different baud rates and system parameters DP-16QAM system. The estimation error). The optical performance monitoring method disclosed in the present application is used to test the OSNR estimation error and the received optical power estimation error of the optical performance monitoring method under different baud rates and system parameters DP-16QAM system. Estimate error. It can be seen that the model also shows ideal estimation performance under the 16QAM system, and the average estimation error of OSNR is 0.21 dB, 0.18 dB, 0.19 dB and 0.22 dB under 25, 30, 40 and 50 Gbaud. The experimental results show that the PON optical performance monitoring method based on the physical information neural network disclosed in the application embeds the nonlinear Schrodinger equation into the loss function, analyzes the influence of the nonlinear phase noise on the OSNR monitoring, solves the problem that the traditional method has a large OSNR monitoring error for the QAM high-order modulation system due to only considering the signal amplitude envelope and ignoring the phase factor, and shows ideal OSNR monitoring performance in the DP-QPSK and high-order DP-16QAM based PON network.

[0145] The PON optical performance monitoring method and system provided in the application construct a loss function based on the nonlinear Schrodinger equation, thereby introducing physical information into the neural network model, which is beneficial to overcome the poor interpretability of the neural network in the prior art, and by collecting a two-dimensional projection image of a three-dimensional spectrum on an 'amplitude-order' plane as the input of the CNN-FCNN network model, it is beneficial to reduce the complexity of the model without losing key information, and by introducing multi-modal data containing system parameters and two-dimensional projection images to extract a multi-modal feature map, it is beneficial to solve the problem of low accuracy of the neural network on out-of-sample data.

[0146] The application embeds the nonlinear Schrodinger equation into the loss function, analyzes the influence of the nonlinear phase noise on the OSNR monitoring, solves the problem that the traditional method has a large OSNR monitoring error for the QAM high-order modulation system due to only considering the signal amplitude envelope and ignoring the phase factor, at the same time, guides the neural network training through physical constraints, thereby enhancing the interpretability of the model and improving the OSNR monitoring accuracy on out-of-sample data; in addition, by performing fractional Fourier transform on the input signal, and taking the two-dimensional projection image of the obtained three-dimensional spectrum as the model input, the model can perform multi-scale analysis on the signal through time-frequency joint analysis, thereby further improving the OSNR monitoring accuracy of the PON network.

[0147] Experiments prove that the PON optical performance monitoring method disclosed in the application shows high OSNR estimation accuracy under various baud rates and system parameters of the dual-polarization quadrature phase shift keying (DP-QPSK) and 16 quadrature amplitude modulation (DP-16QAM) system.

[0148] Corresponding to the above method, the present application also provides a PON optical performance monitoring system, comprising a computer device, the computer device comprising a processor and a memory, located at a receiving end of a PON network, the memory storing computer instructions, the processor being configured to execute the computer instructions stored in the memory, and the system implementing the steps of the method as described above when the computer instructions are executed by the processor.

[0149] Corresponding to the above method, the present application also provides a computer readable storage medium, storing computer programs / instructions thereon, the computer programs / instructions being executed by a processor to implement the steps of the method as described in any one of the above embodiments. The computer readable storage medium can be a tangible storage medium, such as a random access memory (RAM), a memory, a read only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a floppy disk, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0150] Corresponding to the above method, the present application also provides a computer program product, comprising computer programs / instructions, the computer programs / instructions being executed by a processor to implement the steps of the method as described in any one of the above embodiments.

[0151] It should be apparent to one of ordinary skill in the art that the example components, systems and methods described in connection with the embodiments disclosed herein can be implemented in hardware, software, or a combination thereof. The choice of hardware or software implementation is dependent on the particular application and design constraints imposed on the overall system. Skilled artisans can employ a variety of approaches to implement the described functionality, and the application is not limited by the choice of hardware or software implementations. When implemented in hardware, the functionality can be provided as an electronic circuit, a special-purpose integrated circuit (ASIC), a suitable firmware, a plug-in, a functional card, etc. When implemented in software, the elements of the application are the program or code segments that carry out the necessary tasks. The program or code segments can be stored in a machine-readable medium or transmitted through a data signal carried in a carrier wave over a transmission medium or communication link.

[0152] It should be apparent to one of ordinary skill in the art that the example components, systems and methods described in connection with the embodiments disclosed herein can be implemented in hardware, software, or a combination thereof. The choice of hardware or software implementation is dependent on the particular application and design constraints imposed on the overall system. Skilled artisans can employ a variety of approaches to implement the described functionality, and the application is not limited by the choice of hardware or software implementations. When implemented in hardware, the functionality can be provided as an electronic circuit, a special-purpose integrated circuit (ASIC), a suitable firmware, a plug-in, a functional card, etc. When implemented in software, the elements of the application are the program or code segments that carry out the necessary tasks. The program or code segments can be stored in a machine-readable medium or transmitted through a data signal carried in a carrier wave over a transmission medium or communication link.

[0153] In this disclosure, features described and / or illustrated with respect to one implementation can be used in the same manner or in an analogous manner in one or more other implementations, and / or in combination with or in place of features of other implementations.

[0154] The above descriptions are only the preferred embodiments of the present application, not intended to limit the present application. The embodiments of the present application can be variously changed and modified by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall fall within the scope of the present application.

Claims

1. A PON optical performance monitoring method, characterized in that: include: Obtaining a received signal that has been transmitted through the PON network and reached the receiving end and has undergone digital signal processing, and obtaining system parameters related to the PON network communication system; Performing a fractional Fourier transform on the received signal to construct a three-dimensional spectrum, and collecting a two-dimensional projection image of the three-dimensional spectrum on an "amplitude-order" plane; The two-dimensional projection image and the system parameters are used as inputs of a pre-trained multimodal fusion convolutional neural network-fully connected neural network CNN-FCNN, the convolutional neural network branch of the CNN-FCNN performs feature extraction on the input two-dimensional projection image, and the fully connected network branch of the CNN-FCNN performs feature extraction on the input system parameters; The fusion layer of CNN-FCNN fuses the features of different modalities extracted by the convolutional neural network branch and the fully connected network branch, and outputs the estimated optical signal-to-noise ratio and received optical power at the output layer; During the training process of CNN-FCNN, the optical signal-to-noise ratio reference value measured by the optical spectrum analyzer at the receiving end and the received optical power reference value measured by the optical power meter are used as labels to construct a training set. In addition, the loss in PON network signal transmission is calculated as the equation loss using the nonlinear Schrödinger equation. The optical signal-to-noise ratio estimation error is calculated based on the optical signal-to-noise ratio estimation value and the optical signal-to-noise ratio reference value. The received optical power estimation error is calculated based on the received optical power estimation value and the received optical power reference value. The weighted sum of the equation loss, optical signal-to-noise ratio estimation error, and received optical power estimation error is used as the loss function for CNN-FCNN training.

2. The method according to claim 1, characterized in that The system parameters include baud rate, transmission power and transmission distance.

3. The method according to claim 1, characterized in that The convolutional neural network branch includes a preset number of convolution-maximum pooling units, each group of convolution-maximum pooling units includes a preset number of convolution layers and maximum pooling layers, the preset number of convolution layers are used to extract features from the input two-dimensional projection image, and the maximum pooling layer is used for downsampling; The fully connected network branch includes a single fully connected layer containing a preset number of neurons.

4. The method according to claim 1, wherein Performing a fractional Fourier transform on the received signal to construct a three-dimensional spectrum, and collecting a two-dimensional projection image of the three-dimensional spectrum on an "amplitude-order" plane, including: Performing fractional Fourier transform of different orders on the received signal to construct a three-dimensional spectrogram; For the three-dimensional spectra constructed by fractional Fourier transform of different orders, the maximum amplitude corresponding to the order is taken to obtain a two-dimensional projection image of the three-dimensional spectra on the "amplitude-order" plane.

5. The method according to claim 1, wherein The loss expression of the equation is: ； in, and represent the real and imaginary parts of a given complex number, respectively. The expression is: ； in, represents the loss coefficient, represents the nonlinear coefficient, Indicates the distance that the optical signal propagates along the axial direction in the optical fiber. Indicates the reference value of received optical power. Indicates the reference value of noise optical power, Represents an imaginary unit.

6. The method according to claim 1, characterized in that The loss function used for CNN-FCNN training is mathematically expressed as: ； in, are the weight coefficients of equation loss, optical signal-to-noise ratio estimation error, and received optical power estimation error, respectively. represents the equation loss, represents the optical signal-to-noise ratio estimation error, Indicates the received optical power estimation error.

7. A PON optical performance monitoring system comprising a processor, a memory, and a computer program / instruction stored in the memory, characterized in that: Located at the receiving end of the PON network, the processor is used to execute the computer program / instructions. When the computer program / instructions are executed, the system implements the steps of the method according to any one of claims 1 to 6.

8. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images