CN111934755A - SDN controller and optical signal-to-noise ratio prediction method of optical communication equipment - Google Patents

Abstract

The invention discloses an SDN controller and an optical signal-to-noise ratio prediction method of an optical communication device. The method includes: collecting current network state information related to OSNR prediction of the optical communication device under test; for the collected current network state information, The algorithm based on prior knowledge performs OSNR operation; the collected current network state information and the OSNR prior prediction result obtained by the operation are input into the neural network model to obtain the final OSNR prediction result of the optical communication device under test; wherein, The neural network model is pre-trained through the training set. The application of the present invention can have lower prediction error, and can greatly reduce the training difficulty of the neural network model.

Description

SDN controller and optical signal-to-noise ratio prediction method for optical communication equipment

technical field

The invention relates to the technical field of optical communication, in particular to an SDN controller and an optical signal-to-noise ratio prediction method for optical communication equipment.

Background technique

With the vigorous development of Internet technology, optical networks are developing towards ultra-high-speed and large-capacity. The scale of the network is increasing, the structure of the optical network is becoming more and more complex, and the optical path in the network is continuously extended. This will inevitably lead to the transmission of optical signals in the network. Various physical The effect of layer damage. Optical signal damage results in a high end-to-end bit error rate, which not only affects the transmission quality, but even leads to a high blocking rate. Therefore, it is very important to ensure the reliability of communication. Optimize the allocation of optical transmission network resources and service deployment, and improve the stability and reliability of the optical transmission network.

Due to the accumulation of transmission impairments such as power loss, optical signal and dispersion in the optical domain caused by long-distance transmission, the transmission quality of the optical signal will be degraded, which may not meet the requirements of the receiver for the quality of the transmitted signal, and thus cannot be resolved in the optical signal. correct relevant information.

In optical communication equipment, the most important parameter of the physical layer is the optical signal-to-noise ratio. Its value directly determines whether the service can operate normally. Once the optical signal-to-noise ratio (OSNR) in the receiver cannot meet the requirements, it will cause Transmission errors or failures, reduced quality of service, increased transmission consumption, etc.

To sum up, accurate and efficient OSNR prediction is very necessary. Therefore, it is necessary to design an effective OSNR prediction method to achieve more accurate and efficient performance monitoring, thereby laying a foundation for ensuring the transmission quality of communication.

There are two main directions for OSNR prediction of optical communication equipment. One is a priori knowledge scheme based on analytical formulas, and the other is a posteriori knowledge scheme based on historical data and machine learning algorithms. Algorithm schemes based on analytical formulas are generally difficult Taking the influencing factors of OSNR into consideration will reduce the prediction accuracy. The solution based on machine learning algorithm has the problem of lack of training set, which will also lead to the decrease of prediction performance.

In fact, AI algorithms currently occupy half of the OSNR prediction field. However, it is difficult to directly apply these artificial intelligence models to actual scenarios, because they all require a large amount of data sets for model training, but in actual scenarios, there will be very little effective device data, which cannot meet the requirements of model prediction accuracy .

SUMMARY OF THE INVENTION

In view of this, the purpose of the present invention is to propose an optical signal-to-noise ratio prediction method for an SDN controller and an optical communication device. Accurate analysis and prediction capabilities, with lower prediction errors; compared with the existing posterior knowledge schemes based on historical data and machine learning algorithms, the number of training sets required is less and the training time is shorter, which can greatly reduce the neural network model. training difficulty.

Based on the above purpose, the present invention provides an optical signal-to-noise ratio prediction method for optical communication equipment, including:

Collect current network status information related to OSNR prediction of the optical communication device under test;

According to the collected current network state information, the algorithm based on prior knowledge performs OSNR operation;

Input the collected current network state information and the OSNR prior prediction result obtained by the operation into the neural network model to obtain the final OSNR prediction result of the optical communication device under test;

Wherein, the neural network model is pre-trained through a training set; wherein, the training set is based on the OSNR of the optical communication device under test to predict the relevant historical network state information, and for the optical communication device under test The historical network state information related to the OSNR prediction is generated from the prediction result obtained by performing the OSNR operation with an algorithm based on prior knowledge.

Wherein, the neural network model is specifically obtained by pre-training according to the following method:

Collect historical network state information related to OSNR prediction of the optical communication device under test;

For each group of collected historical network state information, the algorithm based on prior knowledge performs OSNR operation to obtain the corresponding OSNR prior prediction result;

For each group of historical network state information, the group of historical network state information and the corresponding OSNR a priori prediction result are formed into a group of input data; the actual OSNR detection result corresponding to the group of historical network state information is used as the input of the group The label data corresponding to the data;

generating the training set according to multiple sets of input data and their corresponding label data;

Input each set of input data in the training set into the neural network model in turn, and train the neural network model: for each set of input data in the training set, after inputting the set of input data into the neural network model, according to the output of the neural network model and the neural network model. The difference between the label data corresponding to the input data of the group is used to adjust the parameters of the neural network model.

The present invention also provides an SDN controller, comprising:

The data acquisition module is used to acquire the current network status information related to the OSNR prediction of the optical communication device under test;

The OSNR a priori prediction result calculation module is used to perform OSNR calculation based on the algorithm of prior knowledge for the current network state information related to the OSNR prediction of the optical communication device under test, and obtain the OSNR prior prediction result;

The OSNR final prediction module is used to input the current network state information related to the OSNR prediction of the optical communication device under test and the OSNR prior prediction result obtained by the operation into the neural network model, and obtain the information of the optical communication device under test. Final OSNR prediction result;

Wherein, the neural network model is pre-trained through a training set; wherein, the training set is based on the OSNR of the optical communication device under test to predict the relevant historical network state information, and for the optical communication device under test The historical network state information related to the OSNR prediction is generated from the prediction result obtained by performing the OSNR operation with an algorithm based on prior knowledge.

In the technical scheme of the present invention, current network state information related to OSNR prediction of the optical communication device under test is collected; for the collected current network state information, an algorithm based on prior knowledge performs OSNR calculation; the collected current network state information, and The OSNR prior prediction result obtained by the operation is input into the neural network model, and the final OSNR prediction result of the optical communication device under test is obtained; wherein, the neural network model is pre-trained through the training set; wherein, the training The set is obtained by performing the OSNR calculation based on the prior knowledge algorithm according to the historical network state information related to the OSNR prediction of the optical communication device under test and the historical network state information related to the OSNR prediction of the optical communication device under test. Prediction results are generated. Therefore, by introducing prior knowledge into the neural network model based on a posteriori knowledge, the neural network model can effectively reduce the training complexity, so that compared with the existing analytical learning model, the technical solution of the present invention can fully combine the neural network. The precise analysis and prediction ability of the complex problems of the model has lower prediction error; at the same time, compared with the existing neural network learning model, the technical solution of the present invention requires fewer training sets and shorter training time, which can greatly reduce the The training difficulty of the neural network model.

Description of drawings

In order to explain the embodiments of the present invention or the technical solutions in the prior art more clearly, the following briefly introduces the accompanying drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only These are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can also be obtained according to these drawings without creative efforts.

1 is a flowchart of a training method of a neural network model applied to optical signal-to-noise ratio prediction of optical communication equipment according to an embodiment of the present invention;

FIG. 2 is a schematic diagram of the architecture of an SDN network according to an embodiment of the present invention;

3 is a schematic diagram of a training deep neural network according to an embodiment of the present invention;

4 is a flowchart of a method for predicting an optical signal-to-noise ratio of an optical communication device according to an embodiment of the present invention;

FIG. 5 is a block diagram of an internal structure of an SDN controller according to an embodiment of the present invention.

Detailed ways

In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to specific embodiments and accompanying drawings.

It should be noted that, unless otherwise defined, the technical or scientific terms used in the embodiments of the present invention shall have the usual meanings understood by those with ordinary skill in the art to which the present disclosure belongs. As used in this disclosure, "first," "second," and similar terms do not denote any order, quantity, or importance, but are merely used to distinguish the various components. "Comprises" or "comprising" and similar words mean that the elements or things appearing before the word encompass the elements or things recited after the word and their equivalents, but do not exclude other elements or things. Words like "connected" or "connected" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "Down", "Left", "Right", etc. are only used to represent the relative positional relationship, and when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

The technical scheme of the present invention proposes a hybrid machine learning scheme, which uses prior knowledge to improve the weakness of the posterior knowledge scheme. OSNR prediction.

In the technical scheme of the present invention, the prior knowledge and the posterior knowledge are combined to establish a hybrid machine learning model. By introducing the prior knowledge into the neural network model based on the posterior knowledge, the neural network model can effectively reduce the Therefore, compared with the existing analytical learning model, the technical solution of the present invention can fully combine the accurate analysis and prediction capabilities of the complex problems of the neural network model, and has lower prediction error; compared with the existing neural network learning model In comparison, the technical solution of the present invention requires fewer training sets and shorter training time, which can greatly reduce the training difficulty of the neural network model.

The technical solutions of the embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

A neural network model applied to optical signal-to-noise ratio prediction of an optical communication device provided by an embodiment of the present invention may be obtained by pre-training through a training set; wherein the training set is based on the optical communication device under test. The historical network state information related to the OSNR prediction, and the prediction result obtained by performing the OSNR operation based on the prior knowledge for the collected historical network state information are generated. Specifically, the training method of the neural network model, as shown in Figure 1, includes the following steps:

Step S101: Collect historical network state information related to OSNR prediction of the optical communication device under test.

Specifically, the software-defined network (SDN) shown in FIG. 2 includes a control plane, a data plane, and an application plane; the SDN controller in the control plane is implemented by the mainstream controller ONOS (Open Network Operating System), and introduces The artificial intelligence engine Tensorflow endows the control plane with the ability to implement artificial intelligence algorithms; the underlying network of the data plane includes two optical communication devices A1, A2; an Erbium-Doped Fiber Amplifier (EDFA) between A1 and A2 for signal loss compensation;

The data acquisition module in the optical communication equipment of the data plane can capture network data information according to the statistical policy issued by the SDN controller of the control plane, and perform necessary data preprocessing, and then report the data preprocessing results to the SDN control device:

The data acquisition module in the optical communication device can collect the historical network state information related to OSNR prediction according to the statistical policy issued by the SDN controller of the control plane, which can include: modulation format of historical transmission data, link length, noise figure of EDFA, WDM channel occupancy rate, power, measured channel flag bit and other related data.

The data acquisition module in the optical communication device can preprocess the collected data of the historical network state information; the data preprocessing includes data standardization and data storage operations. Data standardization refers to the standardization of data information, including the translation of data indicators, the removal of redundant data, format conversion, and making them into data types that can be directly used by neural network models. Data storage refers to storing the standardized data information in the database and reporting it to the SDN controller at the control layer.

Step S102: For the collected historical network state information, an algorithm based on prior knowledge performs an OSNR operation to obtain a corresponding OSNR prior prediction result.

Specifically, after obtaining the preprocessed historical network status information reported by the optical communication device, the SDN controller at the control layer calculates the corresponding OSNR value using an algorithm based on prior knowledge for each group of historical network status information, and obtains Corresponding OSNR prior prediction results; there are many algorithms based on prior knowledge, such as frequency sweep analysis method suitable for application in the field of common low-speed, static, short-distance optical fiber communication OSNR testing, suitable for long-distance, high-speed, high-capacity scenarios. The polarization correlation test method and the formula method in various specific scenarios can be calculated according to the applicable algorithm according to the application scenario; for example, the OSNR a priori prediction result can be calculated according to the following formulas 1 and 2:

OSNR=P _out -∑P _ASE (Equation 1)

P _ASE = -58+N _F +G (Equation 2)

Among them, OSNR is the value calculated by the algorithm based on prior knowledge, P _out is the output power of the optical communication device, ∑P _ASE is the output noise power of the optical communication device, and the single amplifier noise power PASE is determined by the amplifier noise factor _NF and the amplifier gain G is calculated.

Step S103: Generate a training set based on the collected historical network state information and the OSNR prior prediction result obtained by the operation.

Specifically, for each group of historical network status information reported by the optical communication device, the SDN controller forms a set of input data from the group of historical network status information and the OSNR prior prediction result obtained by computing the group of historical network status information; The actual OSNR detection result corresponding to the group of historical network state information is used as the label data corresponding to the group of input data.

The training set is generated according to multiple sets of input data and their corresponding label data, that is, the SDN controller divides several sets of input data and their corresponding label data into a training set, and divides other sets of input data and their corresponding label data into a training set. validation set.

Step S104: Train the neural network model based on the obtained training set.

Specifically, the SDN controller trains the neural network model based on the obtained training set: input each group of input data in the training set into the neural network model in turn, and train the neural network model until the neural network model has converged:

For each set of input data in the training set, after inputting the set of input data into the neural network model, adjust the parameters of the neural network model according to the difference between the output of the neural network model and the label data corresponding to the set of input data . The neural network model may specifically be a deep neural network, or a convolutional neural network, or a long-term and short-term neural network.

The training method of the neural network model of the present invention is consistent with the traditional machine learning model, and the commonly used methods such as gradient descent can be used for model training. For example, as shown in Figure 3, taking the deep neural network as an example, the input data in the training set, that is, a set of historical network state information, and the OSNR prior prediction result obtained by the operation of the set of historical network state information, are obtained from the deep neural network. The input layer of the network is input, and the OSNR prediction value output by the deep neural network is obtained from the output layer of the deep neural network; according to the OSNR prediction value output by the deep neural network and the label data corresponding to the input data, that is, the OSNR of the group of historical network state information is first. The parameters of the deep neural network are adjusted according to the difference between the experimental prediction results.

A method for predicting an optical signal-to-noise ratio of an optical communication device provided by an embodiment of the present invention, as shown in FIG. 4 , includes the following steps:

Step S401: Collect current network state information related to OSNR prediction of the optical communication device under test.

Specifically, the optical communication device under test sends the collected current network state information related to the OSNR prediction to the SDN controller.

Step S402: For the collected current network state information, perform OSNR calculation based on prior knowledge.

Specifically, the SDN controller performs an OSNR calculation based on an algorithm based on a priori knowledge for the current network state information sent by the optical communication device under test, and obtains a corresponding OSNR prior prediction result.

Step S403: Input the collected current network state information and the OSNR prior prediction result obtained by the operation into the neural network model obtained by pre-training to obtain the final OSNR prediction result of the optical communication device under test.

Specifically, the SDN controller inputs the current network status information sent by the optical communication device under test and the corresponding OSNR a priori prediction result calculated according to the current network status information sent by the optical communication device under test into the In the trained neural network model;

The SDN controller obtains the final OSNR prediction result of the optical communication device under test according to the output of the neural network model.

Finally, the application plane in the software-defined network (SDN) can perform performance optimization according to the final OSNR prediction result of the optical communication device obtained by the SDN controller.

An internal structure of the above-mentioned SDN controller is shown in FIG. 5 , including: a data acquisition module 501 , an OSNR prior prediction result calculation module 502 , and an OSNR final prediction module 503 .

The data acquisition module 501 is configured to acquire current network state information related to OSNR prediction of the optical communication device under test;

The OSNR a priori prediction result calculation module 502 is configured to perform an OSNR calculation based on an algorithm based on a priori knowledge for the current network state information related to the OSNR prediction of the optical communication device under test to obtain an OSNR a priori prediction result;

The OSNR final prediction module 503 is configured to input the current network state information related to the OSNR prediction of the optical communication device under test and the OSNR prior prediction result obtained by the operation into the neural network model to obtain the information of the optical communication device under test. Final OSNR prediction result;

Wherein, the neural network model is pre-trained through a training set; wherein, the training set is based on the OSNR of the optical communication device under test to predict the relevant historical network state information, and for the optical communication device under test The historical network state information related to the OSNR prediction is generated from the prediction result obtained by performing the OSNR operation with an algorithm based on prior knowledge.

Further, the above-mentioned SDN controller may further include: a neural network model training module 504 .

The neural network model training module 504 is used to perform an OSNR calculation based on the prior knowledge algorithm for the OSNR prediction related historical network state information of the optical communication device under test to obtain a corresponding OSNR prior prediction result; for each group of historical network Status information, the group of historical network status information and the corresponding OSNR a priori prediction results are formed into a group of input data; the actual OSNR detection result corresponding to the group of historical network status information is used as the label data corresponding to the group of input data ; Generate the training set according to multiple groups of input data and their corresponding label data; Input each group of input data in the training set into the neural network model in turn, and train the neural network model: for each group of input data in the training set, the After the set of input data is input into the neural network model, the parameters of the neural network model are adjusted according to the difference between the output of the neural network model and the label data corresponding to the set of input data.

For the implementation method of the specific functions of each module in the apparatus for predicting a transaction trend provided by the embodiment of the present invention, reference may be made to the method in each step of the flow shown in FIGS.

The apparatuses in the foregoing embodiments are used to implement the corresponding methods in the foregoing embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

In the technical scheme of the present invention, current network state information related to OSNR prediction of the optical communication device under test is collected; for the collected current network state information, an algorithm based on prior knowledge performs OSNR calculation; the collected current network state information, and The OSNR prior prediction result obtained by the operation is input into the neural network model, and the final OSNR prediction result of the optical communication device under test is obtained; wherein, the neural network model is pre-trained through the training set; wherein, the training The set is obtained by performing the OSNR calculation based on the prior knowledge algorithm according to the historical network state information related to the OSNR prediction of the optical communication device under test and the historical network state information related to the OSNR prediction of the optical communication device under test. Prediction results are generated. Therefore, by introducing prior knowledge into the neural network model based on a posteriori knowledge, the neural network model can effectively reduce the training complexity, so that compared with the existing analytical learning model, the technical solution of the present invention can fully combine the neural network. The precise analysis and prediction ability of the complex problems of the model has lower prediction error; at the same time, compared with the existing neural network learning model, the technical solution of the present invention requires fewer training sets and shorter training time, which can greatly reduce the The training difficulty of the neural network model.

The computer readable medium of this embodiment includes both permanent and non-permanent, removable and non-removable media and can be implemented by any method or technology for information storage. Information may be computer readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read only memory (ROM), Electrically Erasable Programmable Read Only Memory (EEPROM), Flash Memory or other memory technology, Compact Disc Read Only Memory (CD-ROM), Digital Versatile Disc (DVD) or other optical storage, Magnetic tape cartridges, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission medium that can be used to store information that can be accessed by a computing device.

Those of ordinary skill in the art should understand that the discussion of any of the above embodiments is only exemplary, and is not intended to imply that the scope of the present disclosure (including the claims) is limited to these examples; under the spirit of the present invention, the above embodiments or There may also be combinations between technical features in different embodiments, steps may be carried out in any order, and there are many other variations of the different aspects of the invention as described above, which are not provided in detail for the sake of brevity.

Additionally, well known power/ground connections to integrated circuit (IC) chips and other components may or may not be shown in the figures provided in order to simplify illustration and discussion, and in order not to obscure the present invention. . Furthermore, devices may be shown in block diagram form in order to avoid obscuring the present invention, and this also takes into account the fact that the details regarding the implementation of these block diagram devices are highly dependent on the platform on which the invention will be implemented (i.e. , these details should be fully within the understanding of those skilled in the art). Where specific details (eg, circuits) are set forth to describe exemplary embodiments of the invention, it will be apparent to those skilled in the art that these specific details may be used without or with changes The present invention is carried out below. Accordingly, these descriptions are to be considered illustrative rather than restrictive.

Although the present invention has been described in conjunction with specific embodiments thereof, many alternatives, modifications, and variations to these embodiments will be apparent to those of ordinary skill in the art from the foregoing description. For example, other memory architectures (eg, dynamic RAM (DRAM)) may use the discussed embodiments.

Embodiments of the present invention are intended to cover all such alternatives, modifications and variations that fall within the broad scope of the appended claims. Therefore, any omission, modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims (10)

1. An osnr prediction method for an optical communication device, comprising:

acquiring current network state information related to OSNR prediction of a tested optical communication device;

aiming at the collected current network state information, carrying out OSNR operation based on an algorithm of prior knowledge;

inputting the collected current network state information and the OSNR prior prediction result obtained by operation into a neural network model to obtain a final OSNR prediction result of the tested optical communication equipment;

wherein, the neural network model is obtained by training in advance through a training set; the training set is generated according to historical network state information related to OSNR prediction of the tested optical communication device and a prediction result obtained by carrying out OSNR operation on the historical network state information related to OSNR prediction of the tested optical communication device based on a priori knowledge algorithm.

2. The method of claim 1, wherein the neural network model is pre-trained according to the following method:

collecting historical network state information related to OSNR prediction of the tested optical communication equipment;

carrying out OSNR operation based on a priori knowledge algorithm aiming at each group of collected historical network state information to obtain a corresponding OSNR priori prediction result;

for each group of historical network state information, forming a group of input data by the group of historical network state information and a corresponding OSNR priori prediction result; taking an actual OSNR detection result corresponding to the set of historical network state information as tag data corresponding to the set of input data;

generating the training set according to the multiple groups of input data and the corresponding label data;

and (3) sequentially inputting each group of input data in the training set into the neural network model, and training the neural network model: and for each group of input data in the training set, after the group of input data is input into the neural network model, adjusting parameters of the neural network model according to the difference between the output of the neural network model and the label data corresponding to the group of input data.

3. The method according to claim 2, wherein the collecting historical network state information related to OSNR prediction of the measured optical communication device specifically includes:

and the tested optical communication equipment sends the collected historical network state information related to OSNR prediction to an SDN controller.

4. The method according to claim 3, wherein the OSNR computation is performed based on a priori knowledge algorithm for each set of collected historical network state information to obtain a corresponding OSNR prior prediction result, specifically:

and the SDN controller performs OSNR operation on each group of historical network state information sent by the tested optical communication equipment based on a priori knowledge algorithm to obtain a corresponding OSNR priori prediction result.

5. The method according to claim 4, wherein the training of the neural network model is specifically:

the SDN controller trains a neural network model.

6. The method according to any one of claims 1 to 5, wherein the collecting of the current network state information related to the OSNR prediction of the measured optical communication device is specifically:

and the tested optical communication equipment sends the collected current network state information related to OSNR prediction to an SDN controller.

7. The method according to claim 6, wherein the OSNR operation is performed by a priori knowledge based algorithm with respect to the collected current network state information, specifically:

the SDN controller carries out OSNR operation on current network state information sent by the tested optical communication equipment based on a priori knowledge algorithm; and

the method comprises the following steps of inputting the collected current network state information and an OSNR priori prediction result obtained by operation into a neural network model, specifically:

and the SDN controller inputs current network state information sent by the tested optical communication equipment and an OSNR priori prediction result obtained by operation into a neural network model.

8. An SDN controller, comprising:

the data acquisition module is used for acquiring current network state information related to OSNR prediction of the tested optical communication equipment;

the OSNR priori prediction result operation module is used for carrying out OSNR operation on the current network state information related to OSNR prediction of the tested optical communication equipment based on a priori knowledge algorithm to obtain an OSNR priori prediction result;

the OSNR final prediction module is configured to input current network state information related to OSNR prediction of the measured optical communication device and an OSNR prior prediction result obtained through calculation into a neural network model, so as to obtain a final OSNR prediction result of the measured optical communication device;

wherein, the neural network model is obtained by training in advance through a training set; the training set is generated according to historical network state information related to OSNR prediction of the tested optical communication device and a prediction result obtained by carrying out OSNR operation on the historical network state information related to OSNR prediction of the tested optical communication device based on a priori knowledge algorithm.

9. The SDN controller of claim 8, further comprising:

the neural network model training module is used for carrying out OSNR operation on historical network state information related to OSNR prediction of the tested optical communication equipment based on a priori knowledge algorithm to obtain a corresponding OSNR priori prediction result; for each group of historical network state information, forming a group of input data by the group of historical network state information and a corresponding OSNR priori prediction result; taking an actual OSNR detection result corresponding to the set of historical network state information as tag data corresponding to the set of input data; generating the training set according to the multiple groups of input data and the corresponding label data; and (3) sequentially inputting each group of input data in the training set into the neural network model, and training the neural network model: for each group of input data in the training set, after the group of input data is input into the neural network model, adjusting parameters of the neural network model according to the difference between the output of the neural network model and the label data corresponding to the group of input data; and

the data acquisition module is further used for acquiring historical network state information related to OSNR prediction of the tested optical communication equipment.

10. An SDN network, comprising: the SDN controller of claim 8 or 9.

Images