Link dispersion management configuration optimization method and system

Abstract

The invention provides a link dispersion management configuration optimization method and system, and the method comprises the steps: S1, determining an optimization scene, configuring corresponding channel parameters, obtaining transmission signal data through a numerical simulation method, and constructing a training set of a feature decoupling distributed optical fiber channel model and dispersion management configuration optimization data based on the transmission signal data; s2, preprocessing the training set, training a single-span feature decoupling distributed fiber channel model by using the preprocessed training set, and cascading and constructing a dispersion management fiber link model based on the trained single-span feature decoupling distributed fiber channel model; and S3, optimizing the data by using the dispersion management configuration, optimizing the dispersion management optical fiber link model through a gradient descent algorithm, and optimizing the dispersion management configuration.

Description

Technical Field

[0001] The present invention relates to the field of optical fiber channel modeling and neural network application technology, and in particular, to a link dispersion management configuration optimization method and system, and more particularly, to a link dispersion management configuration optimization method and system based on a feature-decoupled distributed optical fiber channel model. Background Art

[0002] Link dispersion management is an important aspect of managing and optimizing fiber-optic communication systems. Adding a dispersion management module to fiber-optic transmission aims to offset the impact of dispersion effects and improve the performance and reliability of fiber-optic transmission. Existing dispersion management mostly uses uniform dispersion management, that is, the management intensity of each span is the same, and the management intensity of each span is not optimized, resulting in the damage caused by nonlinear effects being higher than that of a link without dispersion management, thereby reducing overall performance. The key reason for the inability to optimize management configuration is the lack of efficient channel simulation tools and optimization methods. Traditional numerical channel simulation has high computational complexity, slow running speed, and is non-differentiable. It can only use brute force search. For multi-span long-distance transmission links, this optimization method has too large a search space and too high optimization complexity.

[0003] Compared with traditional simulation algorithms, the advantages of the fiber channel modeling scheme based on deep learning are that it takes into account high operating efficiency and low complexity, and is differentiable. Patent document CN114139324A (application number: 202111478750.6) discloses a method for fast and accurate modeling of distributed fiber channels based on feature decoupling, including the following steps: S1: collecting input and output data under different fiber channel conditions as a training data set; step S2: compensating the output data of the channel including nonlinear characteristics through a linear compensation algorithm; step S3: performing data preprocessing and arrangement to obtain a nonlinear data set suitable for neural network training; step S4: modeling the nonlinear channel characteristics based on the neural network structure and deep learning algorithm, and obtaining a nonlinear channel model through training; step S5: during the transmission process, the nonlinear model, linear model, and amplifier model are modeled in turn, and then multiple iterations are performed to finally achieve long-distance transmission of the fiber channel model. Summary of the invention

[0004] In view of the defects in the prior art, an object of the present invention is to provide a method and system for optimizing link dispersion management configuration.

[0005] A link dispersion management configuration optimization method provided by the present invention includes:

[0006] Step S1: Determine the optimization scenario and configure the corresponding channel parameters, obtain the transmission signal data using the numerical simulation method, and construct the training set of the feature decoupled distributed optical fiber channel model and the dispersion management configuration optimization data based on the transmission signal data;

[0007] Step S2: preprocessing the training set, and using the preprocessed training set to train a single-span feature-decoupled distributed optical fiber channel model, and constructing a dispersion-managed optical fiber link model based on the trained single-span feature-decoupled distributed optical fiber channel model by cascading;

[0008] Step S3: Utilizing the dispersion management configuration optimization data, optimizing the dispersion management optical fiber link model through a gradient descent algorithm, and optimizing the dispersion management configuration.

[0009] Preferably, the step S1 includes: determining an optimization scenario and configuring corresponding channel parameters, wherein the channel parameters include: rate, number of channels, channel spacing, number of spans, and span positions for dispersion management.

[0010] Preferably, step S2 comprises:

[0011] Step S2.1: Perform linear feature separation, data rearrangement, and normalization on the training set to obtain a preprocessed training set;

[0012] Step S2.2: using the preprocessed training set to train a single-span feature decoupled distributed optical fiber channel model, to obtain a trained single-span feature decoupled distributed optical fiber channel model;

[0013] Step S2.3: Obtain the number of spans according to the configured channel parameters, decouple the distributed optical fiber channel model based on the single span feature characteristics after span data cascade training, and obtain the dispersion managed optical fiber link model.

[0014] Preferably, the single-span feature-decoupled distributed optical fiber channel model comprises:

[0015]

[0016] Among them, A represents the light field, z represents the distance variable, ω represents the frequency domain angular frequency variable, i represents the imaginary number, β 2 represents the group velocity dispersion parameter, L span Represents the distance of a single span, R DDM Indicates the dispersion management strength parameter to be optimized.

[0017] Preferably, step S3 comprises:

[0018] Step S3.1: inputting the dispersion management configuration optimization data into the dispersion management optical fiber link model for forward propagation to obtain signal data;

[0019] Step S3.2: Processing the obtained signal data including down sampling and carrier phase recovery to obtain a processed single symbol;

[0020] Step S3.3: Calculate the mean square error loss function between the processed single symbol and the original transmitted single symbol in the transmission signal data, perform gradient back propagation from the loss function, update the dispersion management fiber link model, repeatedly trigger steps S3.1 to S3.3, perform multiple gradient descents, until the loss function meets the preset requirements, obtain the optimized dispersion management fiber link model, and achieve optimized dispersion management configuration;

[0021] The loss function includes:

[0022]

[0023] Among them, y i , Represent the original transmitted symbol and the received symbol respectively; N data Indicates the number of symbols transmitted at one time.

[0024] A link dispersion management configuration optimization system provided according to the present invention includes:

[0025] Module M1: Determine the optimization scenario and configure the corresponding channel parameters, obtain the transmission signal data using the numerical simulation method, and construct the training set of the feature-decoupled distributed optical fiber channel model and the dispersion management configuration optimization data based on the transmission signal data;

[0026] Module M2: preprocessing the training set, and using the preprocessed training set to train a single-span feature decoupled distributed optical fiber channel model, and constructing a dispersion-managed optical fiber link model based on the cascade of the trained single-span feature decoupled distributed optical fiber channel model;

[0027] Module M3: Utilize the dispersion management configuration optimization data, optimize the dispersion management fiber link model through the gradient descent algorithm, and optimize the dispersion management configuration.

[0028] Preferably, the module M1 includes: determining an optimization scenario and configuring corresponding channel parameters, wherein the channel parameters include: rate, number of channels, channel spacing, number of spans, and span positions for dispersion management.

[0029] Preferably, the module M2 comprises:

[0030] Module M2.1: Perform linear feature separation, data rearrangement, and normalization on the training set to obtain the preprocessed training set;

[0031] Module M2.2: using the preprocessed training set to train a single-span feature decoupled distributed optical fiber channel model, to obtain a trained single-span feature decoupled distributed optical fiber channel model;

[0032] Module M2.3: Obtain the number of spans according to the configured channel parameters, decouple the distributed optical fiber channel model based on the single span feature characteristics after span data cascade training, and obtain the dispersion managed optical fiber link model.

[0033] Preferably, the single-span feature-decoupled distributed optical fiber channel model comprises:

[0034]

[0035] Among them, A represents the light field, z represents the distance variable, ω represents the frequency domain angular frequency variable, i represents the imaginary number, β 2 represents the group velocity dispersion parameter, L span Represents the distance of a single span, R DDM Indicates the dispersion management strength parameter to be optimized.

[0036] Preferably, the module M3 comprises:

[0037] Module M3.1: Input the dispersion management configuration optimization data into the dispersion management fiber link model for forward propagation to obtain signal data;

[0038] Module M3.2: Processing the obtained signal data including down sampling and carrier phase recovery to obtain a processed single symbol;

[0039] Module M3.3: Calculate the mean square error loss function between the processed single symbol and the original transmitted single symbol in the transmission signal data, perform gradient backpropagation from the loss function, update the dispersion management fiber link model, repeatedly trigger modules M3.1 to M3.3, perform multiple gradient descents until the loss function meets the preset requirements, obtain the optimized dispersion management fiber link model, and realize the optimized dispersion management configuration;

[0040] The loss function includes:

[0041]

[0042] Among them, y i , Represent the original transmitted symbol and the received symbol respectively; N data Indicates the number of symbols transmitted at one time.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] 1. The present invention models linear dispersion through principle formulas, so that the model has the generalization capability of dispersion management configuration and can flexibly construct link models with different dispersion management configurations;

[0045] 2. The present invention improves the system transmission performance by optimizing the configuration of distributed dispersion management parameters;

[0046] 3. By building a differentiable dispersion management link model, automatic optimization based on the gradient descent algorithm is realized, avoiding global brute force search and improving optimization efficiency;

[0047] 4. The present invention is used to optimize the dispersion management parameters of the optical fiber transmission link. A differentiable dispersion management link model is constructed based on a characteristic decoupled distributed optical fiber channel model. Dispersion management is performed at the end of each span. A gradient descent algorithm is used to optimize the dispersion management parameters. The nonlinear damage generated during the signal transmission process is reduced and the system transmission performance is improved through optimization.

[0048] 5. The present invention reduces the transmission accumulated dispersion through the management in the middle of the link, and can reduce the power consumption of the receiving end to compensate for the residual dispersion. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings:

[0050] Figure 1 The present invention is a flowchart of a method for optimizing link dispersion management configuration based on a feature-decoupled distributed optical fiber channel model according to an embodiment of the present invention.

[0051] Figure 2 A schematic diagram of constructing a channel model training data set and a dispersion management configuration optimization data set according to an embodiment of the present invention.

[0052] Figure 3 A schematic diagram of the structure of a dispersion-managed optical fiber link model constructed by a trained channel model and a cascade model according to an embodiment of the present invention.

[0053] Figure 4 The flowchart is a method for optimizing dispersion management configuration by using a gradient descent algorithm according to an embodiment of the present invention.

[0054] Figure 5 It is a schematic diagram of the test results of the optimized configuration of an embodiment of the present invention. DETAILED DESCRIPTION

[0055] The present invention is described in detail below in conjunction with specific embodiments. The following embodiments will help those skilled in the art to further understand the present invention, but are not intended to limit the present invention in any form. It should be noted that, for those of ordinary skill in the art, several changes and improvements can also be made without departing from the concept of the present invention. These all belong to the protection scope of the present invention.

[0056] Example 1

[0057] A link dispersion management configuration optimization method provided by the present invention includes:

[0058] Step 1: Determine the parameter configuration and generate data used in the training and optimization processes, most of which are used for channel model training and the rest are used for the dispersion management configuration optimization process.

[0059] Specifically, step 1 comprises the following steps:

[0060] Step 1.1: Determine the optimization scenario and the corresponding system parameter configuration, including rate, number of channels, channel spacing, number of spans, and span positions for dispersion management;

[0061] Step 1.2: Under the channel conditions specified in step 1.1, without performing dispersion management, the transmission signal data is generated through numerical simulation to construct the channel model training set and the data used for dispersion management configuration optimization.

[0062] Step 2: Use the data generated in step S1 to build a model training data set, train the feature-decoupled distributed fiber channel model, and cascade to build a link model.

[0063] Specifically, step 2 comprises the following steps:

[0064] Step 2.1: Use the data generated in step S1 to construct a model training data set to train a feature-decoupled distributed fiber channel model;

[0065] Step 2.2: According to the number of spans determined in step 1.1, the single span feature decoupled distributed optical fiber channel model trained in cascade step 2.1 is obtained to construct a link model; for the spans specified in step 1.1 for dispersion management, the dispersion management strength parameter of the linear part of the span model is set to be variable. The dispersion management strength is defined as the ratio of the dispersion generated by the single-span single-mode optical fiber transmission to the managed dispersion size. The parameter value of the span model without dispersion management is 0. The dispersion management strength parameter of each span can be regarded as a set of dispersion management configurations.

[0066] Step 3: Based on the dispersion managed fiber link model constructed in step S2, the dispersion management configuration is optimized by a gradient descent algorithm using the data generated in step S1.

[0067] Specifically, step 3 comprises the following steps:

[0068] Step 3.1: Input the transmission signal for link dispersion management configuration optimization obtained in step 1 into the link model constructed in step 2 for forward propagation;

[0069] Step 3.2: The signal data obtained in step 3.1 is processed by the receiving end signal, including down sampling and carrier phase recovery to obtain a single symbol;

[0070] Step 3.3: Calculate the mean square error loss function between the single symbol obtained in step 3.2 and the original transmitted single symbol. Perform gradient backpropagation from the loss function to update the link dispersion management strength configuration.

[0071] Step 3.4: Repeat steps 3.1-3.3 and perform multiple gradient descents until the loss function basically stops decreasing, and finally obtain the optimized dispersion management configuration.

[0072] The present invention also provides a link dispersion management configuration optimization system, which can be implemented by executing the process steps of the link dispersion management configuration optimization method, that is, those skilled in the art can understand the link dispersion management configuration optimization method as a preferred implementation of the link dispersion management configuration optimization system.

[0073] Example 2

[0074] Embodiment 2 is a preferred embodiment of Embodiment 1

[0075] like Figure 1 As shown, the present invention proposes a link dispersion management configuration optimization method based on a feature-decoupled distributed optical fiber channel model, comprising the following steps:

[0076] Step S1: Set the channel parameters to 5 channels, 30 GBaud, 5.0 dBm transmit power per channel, 50 GHz channel spacing, 800 km link transmission distance, 10 spans in total, 80 km per span; set the dispersion management configuration mode to perform dispersion management at the fiber entry and at the end of the 1st to 4th spans; collect data under the above channel conditions, such as Figure 2 As shown, the channel input data is the data after wavelength division multiplexing, and the output data is the data after optical fiber transmission, both of which are high-multiple sampled data; for the data used for dispersion management configuration optimization, the originally sent single-symbol data is also saved for the calculation of the loss function; dispersion management is not performed during the data collection process.

[0077] Step S2: Using the data generated in step S1, construct a data set for channel model training, train a feature-decoupled distributed optical fiber channel model, and cascade to construct a dispersion-managed optical fiber link model.

[0078] This step specifically includes the following sub-steps:

[0079] Step S2.1: Using the data generated in step S1, a training data set is constructed through a process of linear feature separation, data rearrangement, and normalization, and a feature-decoupled distributed optical fiber channel model is obtained through training;

[0080] Step S2.2: According to the number of spans determined in step S1.1, cascade the characteristic decoupled distributed fiber channel models of 10 single spans to build a link model. The model diagram is shown in FIG. Figure 3 As shown, the expression of the linear one-step model is Where R DDM is the dispersion management strength parameter to be optimized; for the 1st to 4th spans specified in step S1.1, the parameter R DDM It is set to be variable, and only the linear model is included at the fiber entry. For the 5th to 10th spans without management, this parameter is fixed to 0;

[0081] Step S3: Using the dispersion management fiber link model constructed in step S2, the dispersion management configuration is optimized by a gradient descent algorithm. The flow chart is as follows: Figure 4 This step specifically includes the following sub-steps:

[0082] Step S3.1: Input the transmission signal for link dispersion management configuration optimization obtained in step S1 into the link model constructed in step S2 for forward propagation, and set the total number of symbols for optimization to 2^17, divided into 3 batches, that is, 2^14 are transmitted at a time; the initial dispersion management configuration is all 0, that is, no dispersion management;

[0083] Step S3.2: The signal generated in step S3.1 is processed by the receiving end signal, including down sampling and carrier phase recovery, to obtain a single symbol;

[0084] Step S3.3: Calculate the mean square error loss function between the single symbol obtained in step S3.2 and the original transmitted single symbol. The loss function expression is: where y i , Represent the original transmitted symbols and received symbols respectively, N data Represents the number of symbols transmitted at one time. Perform gradient backpropagation from the loss function to update the link dispersion management strength configuration parameters;

[0085] Step S3.4: Repeat steps S3.1-S3.3, set the initial learning rate to 0.01, perform gradient descent 1000 times, and decay the learning rate according to cosine in the process. Finally, the optimized dispersion management configuration is obtained. The transmission performance of optimized management and no management under different transmission power conditions is compared as shown in the figure. Figure 5As shown in the figure, the transmission performance of the link with and without dispersion management configuration optimization under different transmission power conditions is compared, with the quality factor (Q factor) as the measurement standard. The higher the Q value, the lower the signal damage level and the higher the transmission quality. Compared with no management, the optimal fiber input power increased by about 0.5dB and the Q factor increased by about 0.3dB with optimized management, achieving the same performance as no management combined with the receiving end nonlinear compensation algorithm, proving that the nonlinear damage caused by the transmission process can be reduced by optimizing the dispersion management configuration.

[0086] The present invention also provides a link dispersion management configuration optimization system, which can be implemented by executing the process steps of the link dispersion management configuration optimization method, that is, those skilled in the art can understand the link dispersion management configuration optimization method as a preferred implementation of the link dispersion management configuration optimization system.

[0087] The link dispersion management configuration optimization method based on the characteristic decoupled distributed optical fiber channel model provided by the present invention realizes the dispersion management configuration optimization by utilizing the channel model, reduces the nonlinear damage generated in the transmission process, and improves the signal transmission quality; at the same time, through the management in the middle of the link, the transmission accumulated dispersion is reduced, and the power consumption of the receiving end to compensate for the residual dispersion can be reduced.

[0088] Those skilled in the art know that, in addition to realizing the system and its various devices, modules, and units provided by the present invention in a purely computer-readable program code, it is entirely possible to realize the same functions in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers by logically programming the method steps. Therefore, the system and its various devices, modules, and units provided by the present invention can be considered as a hardware component, and the devices, modules, and units included therein for realizing various functions can also be regarded as structures within the hardware component; the devices, modules, and units for realizing various functions can also be regarded as both software modules for realizing the method and structures within the hardware component.

[0089] The above describes the specific embodiments of the present invention. It should be understood that the present invention is not limited to the above specific embodiments, and those skilled in the art can make various changes or modifications within the scope of the claims, which does not affect the essence of the present invention. In the absence of conflict, the embodiments of the present application and the features in the embodiments can be combined with each other arbitrarily.

Claims

1. A method for optimizing link dispersion management configuration, characterized in that: include: Step S1: Determine the optimization scenario and configure the corresponding channel parameters, obtain the transmission signal data using the numerical simulation method, and construct the training set of the feature decoupled distributed optical fiber channel model and the dispersion management configuration optimization data based on the transmission signal data; Step S2: preprocessing the training set, and using the preprocessed training set to train a single-span feature-decoupled distributed optical fiber channel model, and constructing a dispersion-managed optical fiber link model based on the trained single-span feature-decoupled distributed optical fiber channel model by cascading; Step S3: Utilizing the dispersion management configuration optimization data, optimizing the dispersion management optical fiber link model through a gradient descent algorithm, and optimizing the dispersion management configuration.

2. The link dispersion management configuration optimization method according to claim 1, characterized in that: The step S1 includes: determining an optimization scenario and configuring corresponding channel parameters, wherein the channel parameters include: rate, number of channels, channel spacing, number of spans, and span positions for dispersion management.

3. The link dispersion management configuration optimization method according to claim 1, characterized in that: The step S2 comprises: Step S2.1: Perform linear feature separation, data rearrangement, and normalization on the training set to obtain a preprocessed training set; Step S2.2: using the preprocessed training set to train a single-span feature decoupled distributed optical fiber channel model, to obtain a trained single-span feature decoupled distributed optical fiber channel model; Step S2.3: Obtain the number of spans according to the configured channel parameters, decouple the distributed optical fiber channel model based on the single span feature characteristics after span data cascade training, and obtain the dispersion managed optical fiber link model.

4. The link dispersion management configuration optimization method according to claim 1, characterized in that: The single-span feature decoupled distributed optical fiber channel model includes: Where A represents the light field, z represents the distance variable, ω represents the frequency domain angular frequency variable, i represents an imaginary number, β2 represents the group velocity dispersion parameter, and L span Represents the distance of a single span, R DDM Indicates the dispersion management strength parameter to be optimized.

5. The link dispersion management configuration optimization method according to claim 1, characterized in that: The step S3 comprises: Step S3.1: inputting the dispersion management configuration optimization data into the dispersion management optical fiber link model for forward propagation to obtain signal data; Step S3.2: Processing the obtained signal data including down sampling and carrier phase recovery to obtain a processed single symbol; Step S3.3: Calculate the mean square error loss function between the processed single symbol and the original transmitted single symbol in the transmission signal data, perform gradient back propagation from the loss function, update the dispersion management fiber link model, repeatedly trigger steps S3.1 to S3.3, perform multiple gradient descents, until the loss function meets the preset requirements, obtain the optimized dispersion management fiber link model, and achieve optimized dispersion management configuration; The loss function includes: in, Represent the original transmitted symbol and the received symbol respectively; N data Indicates the number of symbols transmitted at one time.

6. A link dispersion management configuration optimization system, characterized in that: include: Module M1: Determine the optimization scenario and configure the corresponding channel parameters, obtain the transmission signal data using the numerical simulation method, and construct the training set of the feature-decoupled distributed optical fiber channel model and the dispersion management configuration optimization data based on the transmission signal data; Module M2: preprocessing the training set, and using the preprocessed training set to train a single-span feature decoupled distributed optical fiber channel model, and constructing a dispersion-managed optical fiber link model based on the cascade of the trained single-span feature decoupled distributed optical fiber channel model; Module M3: Utilize the dispersion management configuration optimization data, optimize the dispersion management fiber link model through the gradient descent algorithm, and optimize the dispersion management configuration.

7. The link dispersion management configuration optimization system according to claim 6, characterized in that: The module M1 includes: determining the optimization scenario and configuring corresponding channel parameters, wherein the channel parameters include: rate, number of channels, channel interval, number of spans, and span positions for dispersion management.

8. The link dispersion management configuration optimization system according to claim 6, characterized in that: The module M2 comprises: Module M2.1: Perform linear feature separation, data rearrangement, and normalization on the training set to obtain the preprocessed training set; Module M2.2: using the preprocessed training set to train a single-span feature decoupled distributed optical fiber channel model, to obtain a trained single-span feature decoupled distributed optical fiber channel model; Module M2.3: Obtain the number of spans according to the configured channel parameters, decouple the distributed optical fiber channel model based on the single span feature characteristics after span data cascade training, and obtain the dispersion managed optical fiber link model.

9. The link dispersion management configuration optimization system according to claim 6, characterized in that: The single-span feature decoupled distributed optical fiber channel model includes: Where A represents the light field, z represents the distance variable, ω represents the frequency domain angular frequency variable, i represents an imaginary number, β2 represents the group velocity dispersion parameter, and L span Represents the distance of a single span, R DDM Indicates the dispersion management strength parameter to be optimized.

10. The link dispersion management configuration optimization system according to claim 6, characterized in that: The module M3 comprises: Module M3.1: Input the dispersion management configuration optimization data into the dispersion management fiber link model for forward propagation to obtain signal data; Module M3.2: Processing the obtained signal data including down sampling and carrier phase recovery to obtain a processed single symbol; Module M3.3: Calculate the mean square error loss function between the processed single symbol and the original transmitted single symbol in the transmission signal data, perform gradient backpropagation from the loss function, update the dispersion management fiber link model, repeatedly trigger modules M3.1 to M3.3, perform multiple gradient descents until the loss function meets the preset requirements, obtain the optimized dispersion management fiber link model, and realize the optimized dispersion management configuration; The loss function includes: in, Represent the original transmitted symbol and the received symbol respectively; N data Indicates the number of symbols transmitted at one time.

Images