Method and system for monitoring optical performance of high-capacity optical communication system of passive optical network

Abstract

The invention provides a method and a system for monitoring optical performance of a high-capacity optical communication system of a passive optical network, and relates to the technical field of optical fiber communication and machine learning. Receiving signal data and corresponding distortion data of the few-mode optical fiber communication system are collected at the same time; processing the received signal data and the distortion data to obtain a corresponding data set; constructing a processing model, and training and testing the processing model by using the data set; acquiring actual received signal data of an actual few-mode optical fiber communication system, and inputting the actual received signal data into the trained and tested processing model for processing; according to the invention, real-time monitoring of the optical signal-to-noise ratio in the multimode optical fiber transmission system is realized.

Description

Technical Field

[0001] The present invention relates to the technical fields of optical fiber communication and machine learning, and more specifically, to an optical performance monitoring method and system for a large-capacity optical communication system in a passive optical network. Background Art

[0002] At present, multi-dimensional mode division multiplexing optical fiber communication technology can significantly improve the bandwidth utilization rate and capacity of communication systems, and can meet the urgent demand for large-capacity information transmission due to the continuous rapid growth of network traffic in the context of the global computing power era. It is an important technical direction to break through the capacity limit of single-mode optical fibers and solve the bandwidth crisis.

[0003] However, compared with traditional single-mode optical fibers, the optical fiber communication links of mode division multiplexing optical fiber technology are more complex. Therefore, it is crucial to monitor the optical performance of mode division multiplexing optical fiber communication systems to measure the transmission quality. However, there are no relevant monitoring methods in the existing technologies.

[0004] In a passive optical network system, the optical signal-to-noise ratio is a key indicator for measuring the transmission performance of an optical link. It reflects the degree of noise interference suffered by an optical signal during transmission. Monitoring the change of the optical signal-to-noise ratio can timely detect potential problems in the link. Real-time monitoring of the optical signal-to-noise ratio is crucial for ensuring system performance and stable operation, and provides important support for network maintenance, performance optimization, and resource allocation.

[0005] Researchers have proposed a variety of technologies for monitoring the optical signal-to-noise ratio in optical communication systems, including algorithms based on statistical moments, error vector magnitude method, asynchronous amplitude histogram method, differential pilot method, spectral analysis method, etc. However, the above traditional methods have a high computational complexity in processing a large amount of high-noise signal data, and the detection accuracy is limited.

[0006] Therefore, how to provide an optical performance monitoring method for an optical communication system that can solve the above problems is an urgent problem to be solved by those skilled in the art. Summary of the Invention

[0007] In view of this, the present invention provides an optical performance monitoring method and system for a large-capacity optical communication system in a passive optical network, and proposes a method based on an improved convolutional neural network model to realize real-time monitoring of the optical signal-to-noise ratio in a multi-mode optical fiber transmission system. This method combines deep learning technology with a multi-task learning framework and can process multiple optical signal features simultaneously.

[0008] To achieve the above object, the present invention adopts the following technical solutions:

[0009] An optical performance monitoring method for a large-capacity optical communication system in a passive optical network, comprising the following steps:

[0010] Build and run a few-mode fiber optic communication system, and simultaneously collect the received signal data and corresponding distortion data of the few-mode fiber optic communication system;

[0011] Process the received signal data and the distortion data to obtain corresponding data sets;

[0012] Construct a processing model, and use the data sets to train and test the processing model;

[0013] Obtain the actual received signal data of the actual few-mode fiber optic communication system, and input the actual received signal data into the processed and tested processing model for processing.

[0014] Preferably, the specific process of obtaining the corresponding data sets includes:

[0015] Perform normalization processing on the received signal data;

[0016] Extract the optical signal-to-noise ratio from the normalized received signal data;

[0017] Use the optical signal-to-noise ratio and the normalized received signal data as data sets.

[0018] Preferably, the specific process of constructing the processing model includes:

[0019] The processing model includes: a shared network and multiple sub-networks, and the shared network is connected to the multiple sub-networks.

[0020] Preferably, the specific process of constructing the processing model further includes:

[0021] The shared network includes a convolutional layer, a pooling layer, a flattening layer, and a fully connected layer, and the sub-network includes a dense layer.

[0022] Preferably, the specific process of using the data sets to train and test the processing model includes:

[0023] Divide the data sets into a training set and a test set according to a ratio;

[0024] Use the training set to train the processing model;

[0025] Construct the loss function of the sub-network and the loss function of the shared network, and obtain the corresponding total model loss function;

[0026] Use the test set to test the processing model, and adjust the parameters of the processing model according to the calculation results of the total model loss function and the test results to obtain the corresponding optimal processing model.

[0027] The present invention also provides an optical performance monitoring system for an optical communication system, comprising:

[0028] An acquisition module, configured to establish and operate a few-mode fiber communication system, and simultaneously acquire received signal data and corresponding distortion data of the few-mode fiber communication system;

[0029] An obtaining module, configured to process the received signal data and the distortion data to obtain corresponding data sets;

[0030] A construction module, configured to construct a processing model, and use the data sets to train and test the processing model;

[0031] A processing module, configured to obtain actual received signal data of an actual few-mode fiber communication system, and input the actual received signal data into the trained and tested processing model for processing

[0032] As can be seen from the above technical solutions, compared with the prior art, the present invention discloses an optical performance monitoring method and system for a large-capacity optical communication system of a passive optical network, and proposes a data-driven optical performance monitoring method for a mode-division multiplexing fiber communication system based on a multi-task convolutional neural network. By combining a convolutional neural network with a multi-task learning framework, it can extract transmission link information and complete synchronous prediction of the optical signal-to-noise ratio of multi-mode channels. On the basis of retaining the fine-grained feature extraction of the convolutional neural network, a multi-task learning mechanism is introduced to construct a mapping relationship between mode-division multiplexing fiber channel parameters and received-end distortion signals, realize multi-mode signal feature processing, and complete synchronous prediction of the optical signal-to-noise ratio of multi-mode channels. The proposed solution improves the generalization ability and prediction performance of the model by sharing different mode signal damage feature information, and reduces the computational complexity while maintaining high-accuracy optical signal-to-noise ratio prediction performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present invention, and those of ordinary skill in the art can also obtain other drawings according to the provided drawings without creative efforts.

[0034] Figure 1 It is the overall flowchart of an optical performance monitoring method for a large-capacity optical communication system of a passive optical network provided by the present invention;

[0035] Figure 2 It is the overall structural principle block diagram of a few-mode fiber communication system and a processing model provided by an embodiment of the present invention;

[0036] Figure 3Schematic diagram of the principle implementation of the processing model provided by the embodiment of the present invention;

[0037] Figure 4 Schematic block diagram of the structural principle of an optical performance monitoring system for an optical communication system provided by the present invention;

[0038] Figure 5 Schematic diagram of the changes in the training loss value and MAE value of different mode tasks provided by the embodiment of the present invention. Detailed implementation manners

[0039] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0040] Refer to Figure 1 As shown, the embodiment of the present invention discloses an optical performance monitoring method for a large-capacity optical communication system of a passive optical network, including the following steps:

[0041] Build and run a few-mode fiber optic communication system, and simultaneously collect the received signal data of the few-mode fiber optic communication system and the corresponding distortion data;

[0042] Process the received signal data and the distortion data to obtain the corresponding data set;

[0043] Build a processing model, and use the data set to train and test the processing model;

[0044] Obtain the actual received signal data of the actual few-mode fiber optic communication system, and input the actual received signal data into the trained and tested processing model for processing.

[0045] Specifically, for the overall schematic diagram of the few-mode fiber optic system, refer to Figure 2 As shown, it mainly includes three modules: a signal transmission module, a signal transmission module, and a signal reception module. At the transmitting end, the data goes through steps such as electro-optic conversion, IQ modulator, and digital-to-analog conversion, and different mode signals are obtained by modulating different modes; the different mode signals are coupled together through a mode multiplexer, and then transmitted to the few-mode fiber through a transmitter; during the transmission process, the few-mode fiber can support the transmission of multiple optical modes, but effects such as mode coupling and dispersion will cause damage to the signal; at the receiving end, the transmitted signal goes through the corresponding processing process and the corresponding digital signal processing module to compensate for the link damage; collect the transmitted different mode signals and the corresponding optical signal-to-noise ratio values as subsequent training and verification data.

[0046] In a specific embodiment, the specific process of obtaining the corresponding data set includes:

[0047] Normalize the received signal data;

[0048] Extract the optical signal-to-noise ratio from the normalized received signal data;

[0049] Use the optical signal-to-noise ratio and the normalized received signal data as the data set.

[0050] In a specific embodiment, as shown in Figure 3 The specific process of constructing the processing model includes:

[0051] The processing model includes: a shared network and multiple sub-networks, and the shared network is connected to the multiple sub-networks.

[0052] In a specific embodiment, the specific process of constructing the processing model further includes:

[0053] The shared network includes a convolutional layer, a pooling layer, a flattening layer, and a fully connected layer, and introduces non-linearity into the neural network through a non-linear activation function; the shared network is designed to extract common features, and the extracted feature data is simultaneously input into different independent sub-networks to obtain specific loss values for different tasks; each sub-network contains a dense layer and uses a linear activation function to obtain the output optical signal-to-noise ratio data.

[0054] Specifically, the convolutional kernel size of the convolutional layer can be 3, and batch normalization and a Dropout layer are performed after each pass through the convolutional layer. The pooling size of the pooling layer can be 4, and the stride is 4. The flattening layer is the Flatten layer, and the Flatten layer is subsequently used to flatten the two-dimensional data into one-dimensional. The fully connected layer is used to map the output to the specified number of nodes and finally output the regression result data corresponding to the network.

[0055] In a specific embodiment, the specific process of training and testing the processing model using the data set includes:

[0056] Divide the data set into a training set and a test set according to a ratio;

[0057] Use the training set to train the processing model;

[0058] Construct the loss function of the sub-network and the loss function of the shared network, and obtain the corresponding total model loss function;

[0059] Use the test set to test the processing model, and adjust the parameters of the processing model according to the calculation result of the total model loss function and the test result to obtain the corresponding optimal processing model.

[0060] Specifically, the loss function of the sub-network is defined as the mean squared error (MSE), and the specific expression is:

[0061]

[0062] In the formula, n represents the number of samples in the training set. y i is the true value of the OSNR corresponding to the i-th group of distorted signals. is the predicted value of the OSNR corresponding to the i-th group of distorted signals; the weight of the loss value of different sub-networks is adjusted according to requirements, and the total loss function is obtained. The specific expression is:

[0063]

[0064] In the formula, m represents the number of sub-networks. w 1. w 2. …, w m represents the weight values corresponding to sub-networks 1, 2, …, m. Loss 1. Loss 2. …, Loss m represents the loss values corresponding to sub-networks 1, 2, …, m, where the number of sub-networks can specifically be 2.

[0065] When multiple tasks (corresponding to multiple different sub-networks) of the network target have different tendencies, the weight values can be adjusted according to the actual situation. For example, when there are two sub-networks, if the accuracy requirement for the task corresponding to the first sub-network is higher, and the accuracy requirement for the second sub-network is relatively lower, the weight value of the first sub-network can be increased. For example, the weight of the first sub-network is set to 0.6, and the weight of the second sub-network is set to 0.4.

[0066] The mean absolute error (MAE) is used as an index to evaluate the prediction accuracy of the OSNR corresponding to different sub-networks of the model. The expression is:

[0067]

[0068] The constructed processing model is adjusted by selecting appropriate parameters such as the learning rate, batch size, and number of iterations, and the processing model is trained online and tested offline; when training the processing model online, a large number of received signals of different patterns collected are used as the input of the training set, and the monitored optical signal-to-noise ratio data is used as the label to complete the training of the processing model, so that the processing model has the ability to detect optical signal-to-noise ratio data; during the offline test of the processing model, the validation set data signal is used as the input, and then the processing model will output the optical signal-to-noise ratio data corresponding to different patterns to verify the detection performance of the processing model.

[0069] When training the network online, different modes of I and Q signals are used as inputs, and the optical signal-to-noise ratio detected by OSA is used as the label. When the network is tested offline, I and Q signals are used as the test inputs, and the processing model will output the optical signal-to-noise ratios corresponding to different modes.

[0070] See Figure 4 As shown, an embodiment of the present invention also provides a system for monitoring the optical performance of a large-capacity optical communication system of a passive optical network using the method according to any one of the above embodiments, including:

[0071] An acquisition module, configured to form and run a few-mode fiber communication system, and simultaneously acquire the received signal data and corresponding distortion data of the few-mode fiber communication system;

[0072] An obtaining module, configured to process the received signal data and the distortion data to obtain corresponding data sets;

[0073] A construction module, configured to construct a processing model, and use the data set to train and test the processing model;

[0074] A processing module, configured to obtain the actual received signal data of the actual few-mode fiber communication system, and input the actual received signal data into the processed and tested processing model for processing.

[0075] An embodiment of the present invention uses OPTISYSTEM to simulate an optical communication system based on FMF. The parameters of the processing model are trained on the TensorFlow platform using NVIDIA GeForce RTX 3090 and two-mode tasks. The main parameter settings are shown in Table 1.

[0076] Table 1 Parameter Settings

[0077]

[0078] Figure 5 Shows the training loss values and MAE values of different mode tasks, as well as the validation loss and MAE values. The training loss values corresponding to these two modes both show a downward trend and finally tend to be stable. The smaller the MAE value, the higher the prediction accuracy. The training losses are all close to 1.0, and the validation MAE is close to 0.8.

[0079] In the embodiments of the present invention, the processing model is adjusted by selecting appropriate parameters such as the learning rate, batch size, number of iterations, etc., and the network is trained online and tested offline; when training the network online, the training set input is used, and the monitored optical signal-to-noise ratio value is used as a label to complete the training of the network, so that the network has the ability to detect the optical signal-to-noise ratio value; during the offline test of the network, the validation set data signal is used as the input, and then the processing model will output the optical signal-to-noise ratio values corresponding to different modes to verify the detection performance of the network. The experimental results show that: the processing model proposed by the present invention can monitor the optical performance and effectively estimate the optical signal-to-noise ratio of the transmission signal.

[0080] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The same or similar parts among the various embodiments can be referred to each other. For the devices disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0081] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for monitoring optical performance of a large-capacity optical communication system of a passive optical network, characterized in that: The following steps are involved: Establishing and operating a few-mode optical fiber communication system, and collecting received signal data and corresponding distortion data of the few-mode optical fiber communication system; Processing the received signal data and the distorted data to obtain corresponding data sets; Building a processing model, and using the data set to train and test the processing model; The actual received signal data of the actual few-mode optical fiber communication system is obtained, and the actual received signal data is input into the trained and tested processing model for processing.

2. The optical performance monitoring method of a large-capacity optical communication system of a passive optical network according to claim 1, characterized in that: The specific process of obtaining the corresponding data set includes: Performing normalization processing on the received signal data; Extracting an optical signal-to-noise ratio value from the normalized received signal data; The optical signal-to-noise ratio value and the normalized received signal data are taken as a data set.

3. The optical performance monitoring method of a large-capacity optical communication system of a passive optical network according to claim 2 is characterized in that: The specific process of building a processing model includes: The processing model includes: a shared network and a plurality of sub-networks, wherein the shared network is connected to the plurality of sub-networks.

4. The optical performance monitoring method of a large-capacity optical communication system of a passive optical network according to claim 3 is characterized in that: The specific process of building a processing model also includes: The shared network includes convolutional layers, pooling layers, flattening layers and fully connected layers, and the sub-network includes dense layers.

5. The optical performance monitoring method of a large-capacity optical communication system of a passive optical network according to claim 2, characterized in that: The specific process of training and testing the processing model using the data set includes: Dividing the data set into a training set and a test set according to a certain ratio; Using the training set to train the processing model; Construct the loss function of the sub-network and the loss function of the shared network, and obtain the corresponding total loss function of the model; The processing model is tested using the test set, and the parameters of the processing model are adjusted according to the calculation results of the total loss function of the model and the test results to obtain the corresponding optimal processing model.

6. A system using the optical performance monitoring method of a large-capacity optical communication system of a passive optical network according to any one of claims 1 to 5, characterized in that: include: An acquisition module, used to establish and operate a few-mode optical fiber communication system, and to collect received signal data and corresponding distortion data of the few-mode optical fiber communication system; An acquisition module, used for processing the received signal data and the distorted data to obtain a corresponding data set; A construction module, used to construct a processing model, and train and test the processing model using the data set; The processing module is used to obtain actual received signal data of an actual few-mode optical fiber communication system, and input the actual received signal data into the trained and tested processing model for processing.

Images