Data processing method and device and storage medium

Abstract

The embodiment of the invention discloses a data processing method and device and a storage medium, relates to the technical field of communication, and solves the problem of how to improve the acquisition efficiency and accuracy of fine-grained data. The method comprises the following steps: an NMS obtains network performance data of a first granularity of a communication network, and obtains time sequence characteristic data of the network performance data; wherein the time sequence feature data can comprise one or more of the following data: a feature vector of the network performance data, statistical data of the network performance data and network performance data of a second granularity smaller than a preset number. The feature vector is used for representing the data content of the network performance data from multiple dimensions, and the data acquisition frequency corresponding to the second granularity is greater than the data acquisition frequency corresponding to the first granularity. And then, the NMS performs super-resolution processing on the network performance data of the first granularity according to the time sequence feature data to obtain target data. Wherein the target data comprises the network performance data of the second granularity larger than the preset number.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of communication technology, and in particular, to a data processing method, device, and storage medium. Background Art

[0002] In the process of building a digital twin network, how to obtain fine-grained data of the network to be built is a key issue in building a digital twin network.

[0003] At present, the common data collection method is usually to directly collect fine-grained data, or to perform super-resolution processing on the collected coarse-grained data through super-resolution technology to obtain fine-grained data.

[0004] However, directly collecting fine-grained data puts a lot of pressure on data collection, while the accuracy of fine-grained data obtained through super-resolution technology is low. Summary of the invention

[0005] The embodiments of the present application provide a data processing method, device, and storage medium, which can reduce the pressure of data collection and improve the accuracy of fine-grained data.

[0006] To achieve the above objectives, the present application provides the following technical solutions:

[0007] In a first aspect, a data processing method is provided, which can be executed by a network management system (NMS), or by a component of the NMS, such as a processor, chip, or chip system of the NMS, or by a logic module or software that can implement all or part of the NMS. The following is an example of the method being executed by the NMS. The data processing method includes:

[0008] First, the NMS can obtain network performance data of a first granularity of the communication network, and obtain time series characteristic data of the network performance data. Among them, the time series characteristic data may include one or more of the following data: a characteristic vector of the network performance data, statistical data of the network performance data, and network performance data of a second granularity less than a preset number. The above-mentioned characteristic vector is used to represent the data content of the network performance data from multiple dimensions, and the data collection frequency corresponding to the second granularity is greater than the data collection frequency corresponding to the first granularity. Then, the NMS can perform super-resolution processing on the network performance data of the first granularity according to the time series characteristic data to obtain the target data. Among them, the target data includes network performance data of a second granularity greater than a preset number.

[0009] As can be seen from the above, compared with directly collecting a large amount of fine-grained network performance data, in the technical solution provided by the embodiment of the present application, the first-granularity network performance data and timing characteristic data acquired by the NMS are data with a smaller data volume. The NMS can generate a large amount of fine-grained network performance data (i.e., target data) based on the first-granularity network performance data and timing characteristic data with a smaller data volume. In this way, the technical solution provided by the embodiment of the present application does not need to obtain a large amount of fine-grained network performance data, reducing the pressure of data collection.

[0010] Secondly, compared with the general technology that only performs super-resolution processing on coarse-grained network performance data to obtain fine-grained network performance data, NMS can refer to the time series feature data of the network performance data to perform super-resolution processing on the first-grained network performance data. In this way, the time series feature data of the network performance data can be used to supplement the missing data of the first-grained network performance data, thereby improving the accuracy of the obtained fine-grained data.

[0011] In a possible implementation, the above method for obtaining time series characteristic data of network performance data specifically includes: the NMS may receive the time series characteristic data from an Equipment Management System (EMS).

[0012] Since the EMS is used to manage the original full network performance data of each network device in the communication network, the EMS can determine the timing characteristic data based on the original full network performance data of each network device in the communication network, and send the timing characteristic data to the NMS. Accordingly, the NMS can receive the timing characteristic data from the EMS. In this way, the NMS can generate target data based on the timing characteristic data and the obtained first granularity network performance data.

[0013] In a possible implementation, before receiving the time series characteristic data from the EMS, the data processing method further includes: the NMS may send a first data acquisition request to the EMS, wherein the first data acquisition request is used to request to acquire the time series characteristic data.

[0014] The NMS may send a first data acquisition request to the EMS for requesting to acquire the time series characteristic data. Thus, after receiving the first data acquisition request, the EMS may send the time series characteristic data to the NMS for the NMS to acquire fine-grained network performance data.

[0015] In one possible implementation, the first data acquisition request includes: a data type of the network performance data corresponding to the required timing characteristic data, and / or a time period of the network performance data corresponding to the required timing characteristic data.

[0016] As can be seen from the above, since the data types and / or time periods of the network performance data corresponding to the timing characteristic data are relatively large, in order to generate target data that meets the requirements, the first data acquisition request may include: the data type of the network performance data corresponding to the required timing characteristic data, and / or the time period of the network performance data corresponding to the required timing characteristic data. In this way, the EMS can determine the timing characteristic data required to generate the target data based on the information included in the first data acquisition request (i.e., the data type of the network performance data corresponding to the required timing characteristic data, and / or the time period of the network performance data corresponding to the required timing characteristic data), and send the characteristic data to the NMS, thereby improving the flexibility and accuracy of generating the target data.

[0017] In a possible implementation, the method of performing super-resolution processing on the network performance data of the first granularity according to the time series feature data to obtain the target data specifically includes: first, the NMS can obtain super-resolution parameters. The super-resolution parameters include one or more of the following parameters: the type of data to be super-resolution processed, the time period of the data to be super-resolution processed, and the super-resolution ratio. Then, the NMS can perform super-resolution processing on the network performance data of the first granularity according to the super-resolution parameters according to the time series feature data to obtain the target data.

[0018] As can be seen from the above, when super-resolution processing is performed on the network performance data of the first granularity, since there are many types of super-resolution parameters, in order to generate target data that meets the requirements, the NMS can obtain super-resolution parameters that meet the requirements. In this way, the NMS can perform super-resolution processing on the network performance data of the first granularity according to the obtained super-resolution parameters and time series feature data to obtain target data corresponding to the super-resolution parameters, thereby improving the flexibility and accuracy of generating target data.

[0019] In a possible implementation, the method of performing super-resolution processing on the network performance data of the first granularity according to the super-resolution parameters to obtain the target data specifically includes: first, the NMS can obtain a pre-trained super-resolution model corresponding to the super-resolution parameters. Among them, the super-resolution model has the ability to obtain fine-grained network performance data based on time series feature data and coarse-grained network performance data. Then, the NMS can input the time series feature data and the network performance data of the first granularity into the super-resolution model to obtain the target data.

[0020] As can be seen from the above, multiple super-resolution models can be pre-deployed in NMS, and different super-resolution models can correspond to different super-resolution parameters. After obtaining the super-resolution parameters, NMS can input the time series feature data and the first-granularity network performance data into the corresponding super-resolution model to obtain the target data, which provides a specific implementation method for determining the target data.

[0021] In a possible implementation, the method of performing super-resolution processing on the first-granularity network performance data according to the time series feature data to obtain the target data specifically includes: the NMS can input the time series feature data and the first-granularity network performance data into a pre-trained super-resolution model to obtain the target data. The super-resolution model has the ability to obtain fine-grained network performance data based on the time series feature data and the coarse-grained network performance data.

[0022] In this way, after obtaining the time series feature data and the first-granularity network performance data, the NMS can input the time series feature data and the first-granularity network performance data into a pre-trained super-resolution model to obtain the target data, providing another specific implementation method for determining the target data.

[0023] In one possible implementation, the above-mentioned network performance data includes one or more of the following data: central processing unit (CPU) utilization rate, memory occupancy rate, queries per second (QPS), network request delay, network alarm information, network traffic, number of terminal connections, and data transmission rate.

[0024] In a possible implementation, after super-resolution processing is performed on the network performance data of the first granularity according to the time series feature data to obtain the target data, the data processing method further includes: the NMS can obtain verification data of the target data. The verification data includes network performance data of the second granularity corresponding to the target data and greater than a preset number. When the verification data and the target data meet preset conditions, the NMS can update the super-resolution model. The preset conditions include: the matching degree between the verification data and the target data is less than the preset matching degree, and / or the result difference between the data application result corresponding to the verification data and the data application result corresponding to the time series feature data is less than the preset difference.

[0025] As can be seen from the above, after generating the target data, NMS can also verify the target data by obtaining the verification data of the target data. When the verification data and the target data meet the preset conditions, it means that the accuracy of the super-resolution model is low. In this case, NMS can update the super-resolution model to improve the accuracy of the super-resolution model.

[0026] In a possible implementation, the method for obtaining verification data specifically includes: the NMS may receive the verification data from the EMS.

[0027] Since the EMS is used to manage the original full network performance data of each network device in the communication network, the EMS can determine the verification data based on the original full network performance data of each network device in the communication network and send the verification data to the NMS. Accordingly, the NMS can receive the verification data from the EMS. In this way, the NMS can determine whether to update the super-resolution model based on whether the verification data and the target data meet the preset conditions.

[0028] In a possible implementation, before receiving the verification data from the EMS, the data processing method further includes: the NMS may send a second data acquisition request to the EMS. The second data acquisition request is used to request to obtain the verification data, and the second data acquisition request includes one or more of the following information: the second granularity, the data type of the target data, and the amount of required verification data.

[0029] The NMS may send a second data acquisition request to the EMS for requesting to acquire verification data. Thus, after receiving the second data acquisition request, the EMS may send verification data to the NMS for the NMS to determine whether to update the super-resolution model.

[0030] In a possible implementation, the data processing method further includes: the NMS may obtain a feature extraction model. The feature extraction model has the ability to extract feature vectors of network performance data based on network performance data. When the verification data and the target data meet preset conditions, the NMS may update the feature extraction model and send the updated feature extraction model to the EMS.

[0031] As can be seen from the above, since the time series feature data includes the feature vector of the network performance data, and the feature vector of the network performance data is obtained by the EMS according to the feature extraction model deployed in the EMS, when the verification data and the target data meet the preset conditions, it means that the accuracy of the feature extraction model deployed in the EMS is low. In this case, the NMS can update the feature extraction model to improve the accuracy of the feature extraction model.

[0032] In a possible implementation, the above-mentioned acquisition of the network performance data of the first granularity of the communication network specifically includes: the NMS may send a third data acquisition request to the EMS. The third data acquisition request is used to request to obtain the network performance data of the first granularity. The third data acquisition request includes the first granularity. Then, the NMS may receive the network performance data of the first granularity from the EMS.

[0033] In a possible implementation, the third data acquisition request also includes: the data type of the required network performance data, and / or the time period of the required network performance data.

[0034] Since the EMS is used to manage the original full amount of network performance data of each network device in the communication network, the EMS can determine the network performance data of the first granularity based on the original full amount of network performance data of each network device in the communication network. The NMS can send a third data acquisition request to the EMS for requesting to obtain the network performance data of the first granularity. In this way, after receiving the third data acquisition request, the EMS can send the network performance data of the first granularity to the NMS, so that the NMS can generate the target data based on the network performance data of the first granularity and the acquired timing feature data.

[0035] In the second aspect, a model training method is provided, which can be executed by a model training device, or by a component of the model training device, such as a processor, chip, or chip system of the model training device, or by a logic module or software that can implement all or part of the model training device. The model training device can be integrated with the NMS or EMS in the same device, or it can be independently set. The following is an example of the method being executed by an independently set model training device. The model training method includes:

[0036] First, the model training device can obtain the first sample data. The first sample data includes sample network performance data of the first network and time series feature data of the sample network performance data, and the time series feature data of the sample network performance data includes one or more of the following data: feature vectors of the sample network performance data, statistical data of the sample network performance data, and sample network performance data of a second granularity less than a preset number. Then, the model training device can obtain a super-resolution model based on the first sample data training. The super-resolution model has the ability to obtain fine-grained network performance data based on time series feature data and coarse-grained network performance data. Then, the model training device can send the super-resolution model to the NMS.

[0037] As can be seen from the above, since the first sample data includes the sample network performance data of the first network and the time series feature data of the sample network performance data, the super-resolution model obtained by the model training device based on the first sample data training has the ability to obtain fine-grained network performance data based on the time series feature data and coarse-grained network performance data. In this way, the model training device can send the super-resolution model to the NMS, so that the NMS can obtain the target data based on the super-resolution model, the first-granularity network performance data of the communication network, and the time series feature data, without the need for the NMS to collect a large amount of fine-grained network performance data, thereby reducing the pressure of data collection on the NMS.

[0038] Secondly, since the super-resolution model has the ability to obtain fine-grained network performance data based on time series feature data and coarse-grained network performance data, compared with the general technology that only obtains fine-grained network performance data through super-resolution processing of coarse-grained network performance data, the NMS can use the super-resolution model sent by the model training device to supplement the missing data of the first-grained network performance data through the time series feature data of the network performance data, thereby improving the accuracy of the target data.

[0039] Again, after the super-resolution model is trained, in other communication network scenarios, the NMS only needs to obtain the first-granularity network performance data and timing feature data of other communication networks to obtain the target data based on the super-resolution model. Compared with the general technology that needs to retrain the super-resolution model based on the sample data of the current communication network, this application does not need to repeatedly train the super-resolution model, which improves the generalization of the super-resolution model and enriches the application scenarios of the super-resolution model.

[0040] In a possible implementation, the above method for obtaining the first sample data specifically includes: the model training device can receive the first sample data from the EMS.

[0041] Since EMS is used to manage the original and complete network performance data of each network device in the communication network, the model training device can obtain the first sample data from EMS.

[0042] In a possible implementation, before receiving the first sample data from the EMS, the model training method further includes: the model training device may send a fourth data acquisition request to the EMS, wherein the fourth data acquisition request is used to request to acquire the first sample data.

[0043] The model training device may send a fourth data acquisition request to the EMS for requesting to obtain the first sample data. Thus, after receiving the fourth data acquisition request, the EMS may send the first sample data to the model training device for the model training device to train the super-resolution model.

[0044] In a third aspect, a data processing method is provided, which can be executed by an EMS, or by a component of the EMS, such as a processor, a chip, or a chip system of the EMS, or by a logic module or software that can implement all or part of the EMS. The following is an example of the method being executed by the EMS. The data processing method includes:

[0045] The EMS may send network performance data of the first granularity of the communication network to the NMS, and send time series characteristic data of the network performance data to the NMS. The time series characteristic data includes one or more of the following data: feature vectors of the network performance data, statistical data of the network performance data, and network performance data of the second granularity less than a preset number, the feature vector is used to represent the data content of the network performance data from multiple dimensions, and the data collection frequency corresponding to the second granularity is greater than the data collection frequency corresponding to the first granularity; the time series characteristic data is used by the NMS to perform super-resolution processing on the received network performance data of the first granularity to obtain target data, and the target data includes network performance data of the second granularity greater than a preset number.

[0046] In a possible implementation, before sending the time series characteristic data of the network performance data to the NMS, the data processing method further includes: the EMS may receive a first data acquisition request from the NMS, wherein the first data acquisition request is used to request to acquire the time series characteristic data.

[0047] In one possible implementation, the first data acquisition request includes: a data type of the network performance data corresponding to the required timing characteristic data, and / or a time period of the network performance data corresponding to the required timing characteristic data.

[0048] In a possible implementation, when the time series feature data includes a feature vector of network performance data, the data processing method further includes: the EMS may input the network performance data into a feature extraction model to obtain the feature vector of the network performance data. The feature extraction model has the ability to extract the feature vector of the network performance data based on the network performance data, and the feature extraction model is trained based on the network performance data of the second network.

[0049] In a possible implementation, the data processing method further includes: the EMS may send verification data of the target data to the NMS, wherein the verification data is used by the NMS to determine whether to update the super-resolution model and the feature extraction model, and the verification data includes network performance data of a second granularity greater than a preset number corresponding to the target data.

[0050] In a possible implementation, before sending the verification data to the NMS, the data processing method further includes: the EMS may receive a second data acquisition request from the NMS. The second data acquisition request is used to request the verification data, and the second data acquisition request includes one or more of the following information: the second granularity, the data type of the target data, and the amount of the required verification data.

[0051] In a possible implementation, the data processing method further includes: the EMS may receive an updated feature extraction model sent from the NMS.

[0052] In a fourth aspect, a data processing device is provided for implementing the above-mentioned various methods. The data processing device may be the NMS in the above-mentioned first aspect or any implementation of the first aspect, or a device including the above-mentioned NMS, or a device included in the above-mentioned NMS, such as a chip; or, the data processing device may be the model training device in the above-mentioned second aspect or any implementation of the second aspect, or a device including the above-mentioned model training device, or a device included in the above-mentioned model training device, such as a chip; or, the data processing device may be the EMS in the above-mentioned third aspect or any implementation of the third aspect, or a device including the above-mentioned EMS, or a device included in the above-mentioned EMS, such as a chip. The data processing device includes a module, unit, or means corresponding to the implementation of the above-mentioned method, and the module, unit, or means may be implemented by hardware, software, or by hardware executing the corresponding software implementation. The hardware or software includes one or more modules or units corresponding to the above-mentioned functions.

[0053] In some possible designs, the data processing device may include a processing module and a transceiver module. The transceiver module, which may also be referred to as a transceiver unit, is used to implement the sending and / or receiving functions in any of the above aspects and any possible implementations thereof. The transceiver module may be composed of a transceiver circuit, a transceiver, a transceiver or a communication interface. The processing module may be used to implement the processing functions in any of the above aspects and any possible implementations thereof.

[0054] In some possible designs, the transceiver module includes a sending module and a receiving module, which are respectively used to implement the sending and receiving functions in any of the above aspects and any possible implementation methods thereof.

[0055] In a fifth aspect, a data processing device is provided, comprising: a processor. The processor is connected to a memory, the memory is used to store computer-executable instructions, and the processor executes the computer-executable instructions stored in the memory, thereby implementing any one of the methods provided in any one of the first to third aspects. The memory and the processor may be integrated together or may be independent devices. If the memory is an independent device, the memory may be located inside the data processing device or outside the data processing device.

[0056] In a possible implementation, the processor includes a logic circuit and an input interface and / or an output interface, wherein the output interface is used to execute the sending action in the corresponding method, and the input interface is used to execute the receiving action in the corresponding method.

[0057] In a possible implementation, the data processing device further includes a data processing interface and a data processing bus, and the processor, the memory and the data processing interface are connected via the data processing bus. The data processing interface is used to perform the sending and receiving actions in the corresponding method. The data processing interface may also be referred to as a transceiver. Optionally, the data processing interface includes a transmitter and a receiver, in which case the transmitter is used to perform the sending action in the corresponding method, and the receiver is used to perform the receiving action in the corresponding method.

[0058] In a possible implementation, the data processing device exists in the form of a chip product.

[0059] In a sixth aspect, a computer-readable storage medium is provided, comprising instructions, which, when executed on a computer, enable the computer to execute any one of the methods provided in any one of the first to third aspects.

[0060] In a seventh aspect, a computer program product comprising instructions is provided. When the instructions are executed on a computer, the computer is caused to execute any one of the methods provided in any one of the first to third aspects.

[0061] In an eighth aspect, a chip is provided, comprising: a processor and an interface, wherein the processor is coupled to a memory via the interface, and when the processor executes a computer program or instruction in the memory, any one of the methods provided in any one of the first to third aspects is executed.

[0062] In a ninth aspect, a data processing system is provided, comprising: an NMS for executing the data processing method described in the first aspect, and an EMS for executing the data processing method described in the third aspect.

[0063] In a possible implementation, the data processing system may further include: a model training device for executing the model training method described in the second aspect above.

[0064] The technical effects brought about by any implementation method of the third to ninth aspects can refer to the technical effects brought about by the corresponding implementation methods of the first to second aspects, and will not be repeated here.

[0065] It should be noted that various possible implementation methods of any one of the above aspects can be combined under the premise that there is no contradiction between the solutions. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 It is a schematic diagram of network performance data with different granularities;

[0067] Figure 2 Provides a schematic diagram of data transmission rates under different environments;

[0068] Figure 3 Provides another schematic diagram of network performance data at a different granularity;

[0069] Figure 4 A schematic diagram of the structure of a data processing system provided in an embodiment of the present application;

[0070] Figure 5 A flowchart of a data processing method provided in an embodiment of the present application;

[0071] Figure 6 A flowchart of another data processing method provided in an embodiment of the present application;

[0072] Figure 7 A flowchart of a model training method provided in an embodiment of the present application;

[0073] Figure 8 A flowchart of another model training method provided in an embodiment of the present application;

[0074] Fig. 9 A flowchart of another data processing method provided in an embodiment of the present application;

[0075] Fig.10 A schematic diagram of the composition of a data processing device provided in an embodiment of the present application;

[0076] Fig.11 A schematic diagram of the composition of another data processing device provided in an embodiment of the present application;

[0077] Fig.12 A schematic diagram of the hardware structure of a data processing device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0078] In the description of this application, unless otherwise specified, " / " indicates that the objects associated with each other are in an "or" relationship, for example, A / B can represent A or B; "and / or" in this application is merely a description of the association relationship between associated objects, indicating that three relationships may exist, for example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone, where A and B can be singular or plural.

[0079] In the description of this application, unless otherwise specified, "plurality" means two or more than two. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b, or c can mean: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple.

[0080] In addition, in order to clearly describe the technical solutions of the embodiments of the present application, in the embodiments of the present application, words such as "first" and "second" are used to distinguish the same items or similar items with substantially the same functions and effects. Those skilled in the art can understand that words such as "first" and "second" do not limit the quantity and execution order, and words such as "first" and "second" do not necessarily limit the difference.

[0081] In the embodiments of the present application, words such as "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a concrete way for easy understanding.

[0082] It is understood that the "embodiment" mentioned throughout the specification means that the specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the present application. Therefore, the various embodiments in the entire specification do not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. It is understood that in various embodiments of the present application, the size of the sequence number of each process does not mean the order of execution, and the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present application.

[0083] It can be understood that in the present application, "when" and "if" both mean that corresponding processing will be carried out under certain objective circumstances, and do not limit the time, nor do they require any judgment action when implementing, nor do they mean the existence of other limitations.

[0084] It can be understood that some optional features in the embodiments of the present application may be implemented independently in certain scenarios without relying on other features, such as the solution on which they are currently based, to solve corresponding technical problems and achieve corresponding effects, or may be combined with other features according to needs in certain scenarios. Accordingly, the devices provided in the embodiments of the present application may also realize these features or functions accordingly, which will not be elaborated here.

[0085] In this application, unless otherwise specified, the same or similar parts between the various embodiments can refer to each other. In the various embodiments in this application, and the various implementation methods / implementation methods / implementation methods in each embodiment, if there is no special explanation and logical conflict, the terms and / or descriptions between different embodiments and the various implementation methods / implementation methods / implementation methods in each embodiment are consistent and can be referenced to each other. The technical features in different embodiments and the various implementation methods / implementation methods / implementation methods in each embodiment can be combined to form new embodiments, implementation methods, implementation methods, or implementation methods according to their inherent logical relationships. The implementation methods of this application described below do not constitute a limitation on the scope of protection of this application.

[0086] In order to make the embodiments of the present application clearer, some concepts involved in the present application are first briefly introduced.

[0087] 1. Network digital twin:

[0088] A network digital twin is a virtual network entity (also called a network digital twin) built based on digital technology and simulation models, which aims to achieve refined modeling and management of real communication networks. A network digital twin can achieve real-time monitoring, prediction and optimization of communication networks by mapping the network data of real communication networks into a virtual environment.

[0089] The network data of the communication network may include: network structure data, network configuration data, network status data, network performance data, etc. The network structure data may be the connection relationship of each network device in the communication network, etc. The network configuration data may be the configuration parameters of each network device in the communication network (such as port configuration parameters, maximum CPU usage configuration parameters, etc.). The network status data may be the operating status of each network device in the communication network (such as high load status, low load status, abnormal status, etc.).

[0090] The network performance data may be a key performance indicator (KPI) of each network device in the communication network.

[0091] Among them, the KPI of the network device may include the CPU usage rate, memory occupancy rate, QPS, network request delay, network alarm information, network traffic, number of terminal connections, data transmission rate and other data used to represent the performance of the device.

[0092] CPU usage refers to the ratio of the CPU workload to the available processing capacity of the CPU in a certain period of time.

[0093] Memory usage refers to the ratio of memory usage to total memory of a network device within a certain period of time.

[0094] QPS refers to the amount of traffic processed by a network device within a certain period of time.

[0095] Network request latency refers to the latency of a network device requesting data within a certain period of time.

[0096] Network alarm information refers to the number, time and alarm content of network alarm information issued by network equipment within a certain time period.

[0097] Network traffic refers to the data traffic obtained by network devices within a certain period of time.

[0098] The number of terminal connections refers to the number of terminals connected to a network device within a certain period of time.

[0099] Data transmission rate refers to the efficiency of network equipment in sending and receiving information within a certain period of time.

[0100] In the following embodiments, for the description of network performance data, please refer to the specific data of the network performance data herein and no further details will be given.

[0101] In the process of building a network digital twin, by establishing a simulation model that is highly matched with the actual communication network, the operation status and changes of the communication network can be reflected in real time, that is, by collecting and synchronizing the network data of the actual communication network in real time, the network digital twin can accurately reflect the operation status and performance characteristics of the actual communication network. Therefore, the key point of building a network digital twin lies in the data collection and synchronization of the communication network. General technology can collect and synchronize the network data of the communication network in real time through network monitoring equipment, sensors and other data sources.

[0102] In general technology, the device for building a network digital twin can obtain the original full amount of network data. The original full amount of network data may include the original full amount of network structure data, network configuration data, network status data, network performance data, etc. The original full amount of network data may be the network data of each network device at each moment in the communication network. In actual applications, in order to reduce the pressure of data collection, it is not necessary to obtain the original full amount of network data, and partial network data can be collected for building a network digital twin.

[0103] Specifically, since the fluctuations of network structure data, network configuration data, and network status data are small within a certain period of time, in order to reduce the pressure of data collection during the construction of the network digital twin, these data can be collected at a lower frequency, that is, there is no need to collect fine-grained network structure data, network configuration data, and network status data. However, the fluctuations of network performance data within a certain period of time are large, so in order to ensure the accuracy and reliability of the network digital twin, network performance data needs to be collected at a higher frequency.

[0104] Figure 1 A schematic diagram of network performance data at different granularities is provided. Figure 1 As shown, network performance data of different granularities may include coarse-grained network performance data and fine-grained network performance data. In the embodiment of the present application, the coarse-grained network performance data may also be referred to as network performance data of the first granularity. The fine-grained network performance data may also be referred to as network performance data of the second granularity. For ease of description, the embodiment of the present application is described by taking coarse-grained network performance data and fine-grained network performance data as examples.

[0105] Coarse-grained network performance data refers to network performance data collected at a lower collection frequency. For example, network performance data collected once a minute is coarse-grained network performance data.

[0106] Fine-grained network performance data refers to network performance data collected at a higher collection frequency. For example, network performance data collected once per second is fine-grained network performance data.

[0107] In practical applications, a higher acquisition frequency usually refers to an acquisition frequency within the range of once per second to once every 5 minutes. A lower acquisition frequency usually refers to an acquisition frequency within the range of once per hour to once per day.

[0108] like Figure 1 As shown, when the collected data is coarse-grained network performance data (e.g. Figure 1 The data characteristics of the coarse-grained network performance data are stable and cannot reflect the specific network performance of the network equipment in the communication network at each moment, and thus cannot infer whether a communication network failure occurs.

[0109] When the collected data is fine-grained network performance data (such as Figure 1When the network performance data is collected at the second level as shown in the figure, the data characteristics of the fine-grained network performance data can clearly reflect the specific network performance of the network equipment in the communication network at each moment. In this way, the network digital twin constructed by the fine-grained network performance data can accurately infer the fault delay and time period of the communication network, and then accurately locate the fault in the communication network.

[0110] That is to say, through fine-grained network performance data, the network digital twin can more comprehensively perceive the operation of the communication network, and can accurately capture and analyze subtle changes and anomalies, providing an important foundation for real-time diagnosis of network problems, rapid response to anomalies, and effective network optimization and self-healing. In contrast, the network digital twin constructed only using coarse-grained network performance data often cannot capture the fine-grained features in the communication network, thus limiting the ability to diagnose and repair network anomalies in real time.

[0111] Therefore, in order to build an effective network digital twin, the ability to obtain fine-grained network performance data is crucial. By obtaining accurate and fine-grained network performance data, we can better perceive and understand the operation status of the communication network, realize real-time diagnosis and repair of abnormal situations, and thus improve the efficiency and reliability of autonomous operation and maintenance of the communication network.

[0112] 2. High frequency acquisition:

[0113] In order to build an effective network digital twin, general technology can collect network performance data through high-frequency collection to obtain fine-grained network performance data. High-frequency collection refers to collecting network performance data at a higher collection frequency (for example, once per second). Among them, network performance data can be divided into statistical network performance data and measurement network performance data according to its data characteristics. The standard specifies different maximum collection frequencies for different types of network performance data.

[0114] For measurement-type network performance data, the standard (3GPP TS28.552) defines a higher collection frequency, which can be as high as once per second (in actual applications, it is generally once per minute). Measurement-type network performance data usually includes CPU usage, memory occupancy, QPS, network request latency, data transmission rate and other real-time network performance data. By collecting data at a higher collection frequency, the network digital twin can more accurately capture the instantaneous changes and dynamic performance of network equipment in the communication network, providing more timely data support for real-time network monitoring and fault diagnosis.

[0115] As for statistical network performance data (such as network traffic, number of terminal connections, etc.), since such data needs to be based on data trends and changes to reflect the overall status and performance of the communication network, the standard (3GPPTS28.552) defines a maximum collection frequency of once every 5 minutes (in practical applications, it is generally once an hour). This means that the equipment used to build the network digital twin will regularly collect and synchronize statistical network performance data in the communication network, and by capturing data trends and changes in statistical network performance data over a longer time interval, it can more comprehensively reflect the overall status and performance of the communication network.

[0116] It should be noted that, for statistical network performance data, although the collection frequency of once every 5 minutes is lower than the collection frequency of measurement network performance data (once per second), in actual applications, the collection frequency of statistical network performance data is generally once per hour. Therefore, the collection frequency of once every 5 minutes is also considered high-frequency collection among the collection frequencies corresponding to statistical network performance data.

[0117] Following the acquisition frequency defined in the standard, the network digital twin can maintain consistency with the real communication network, and can obtain the latest network performance data in a timely manner, and apply it to the modeling, simulation and analysis process to maintain an accurate description and simulation of the communication network. At the same time, this also provides network operation and maintenance personnel with more timely and accurate information to support network decision-making and optimization work.

[0118] However, collecting fine-grained network performance data through high-frequency acquisition places high demands on system resources, especially in large-scale network environments, where the demand for resources such as transmission bandwidth, memory, and computing power grows exponentially. This leads to a sharp increase in the load of data collection, which cannot meet the requirements of real-time and accuracy, and limits the application scope and performance of network digital twins.

[0119] Secondly, the general high-frequency data acquisition method faces high data acquisition costs. High-frequency data acquisition usually requires a large amount of storage space and transmission bandwidth to process and transmit massive amounts of data, which brings huge cost pressure. Especially in large-scale networks, the unacceptable cost of data acquisition is further highlighted, limiting the feasibility of network digital twins in actual deployment.

[0120] In addition, the general high-frequency acquisition method also has challenges in data processing and analysis. The real-time processing and analysis of data obtained from large-scale high-frequency acquisition requires powerful computing resources and efficient algorithms, and general technology still has room for improvement in data processing speed and efficiency. This also limits the actual application capabilities of network digital twins in application scenarios such as real-time monitoring, fault prediction, and optimized decision-making.

[0121] In summary, the general high-frequency data acquisition method has the disadvantages of high requirements on system resources, high data acquisition cost, and low processing and analysis efficiency.

[0122] 3. Super-Resolution:

[0123] Super-resolution, abbreviated as super-resolution, is an application in the field of images, which aims to obtain a high-resolution image through a series of low-resolution images.

[0124] The basic idea of ​​super-resolution is to improve the resolution of images or data through mathematical algorithms and statistical models. Super-resolution can use the redundant information and structural features in images or data to reconstruct and output high-resolution images or data from input low-resolution images or data. This process can be seen as inferring missing detail information from incomplete images or data, thereby achieving data reconstruction beyond the resolution limit of hardware devices.

[0125] Super-resolution can be applied in various fields, such as image processing, video enhancement, medical imaging, remote sensing images, etc. In the field of image processing, super-resolution generates high-resolution images by fusing multiple low-resolution images or by interpolation and prediction. This can improve the details and clarity of the image, making the target objects in the image clearer and more identifiable.

[0126] In the process of realizing super-resolution, commonly used methods include interpolation-based methods, statistical modeling-based methods, and machine learning-based methods. Interpolation methods increase the resolution of images by interpolating pixels of low-resolution images. Statistical modeling methods use the statistical characteristics of images and prior knowledge to infer missing high-frequency information. Machine learning-based methods learn the mapping relationship between low-resolution images and high-resolution images by training models, thereby achieving super-resolution reconstruction. At present, super-resolution is gradually being applied to other fields besides image processing.

[0127] In short, super-resolution can achieve the acquisition of images or other data with higher resolution than the imaging system by improving data quality and reconstructing missing information without changing the hardware equipment. This provides a reference tool and method for collecting fine-grained network performance data when building a network digital twin.

[0128] At present, super-resolution has gradually been applied in the application scenario of building network digital twins. When building a network digital twin, the device can obtain a training data set and train a super-resolution model based on the training data set. Subsequently, the device can input the collected coarse-grained network performance data into the trained super-resolution model to obtain fine-grained network performance data. In this way, the device can build a network digital twin through fine-grained network performance data.

[0129] However, due to the limitations of the training dataset, the generalization ability of the super-resolution model is restricted, that is, the super-resolution model has limited generalization ability in different network scenarios.

[0130] Exemplarily, referring to the description of the above network performance data, taking the network performance data as the data transmission rate of the network device in the communication network as an example, Figure 2 A schematic diagram of data transmission rate under different environments is provided. Figure 2 As shown in the figure, the circular mark represents the data transmission rate under different channel qualities in the model training environment (i.e., the training data set). The triangular mark represents the data transmission rate under different channel qualities in the actual communication network (i.e., the actual fine-grained network performance data). Figure 2 It can be seen that most of the triangular marks are distributed on the right side of the coordinate system, while most of the circular marks are distributed on the left side of the coordinate system, that is, in the actual communication network, the data transmission rate under the same channel quality is significantly higher than the data transmission rate in the model training environment. Therefore, the training data set obtained by the device cannot cover the rich and diverse existing network data, and therefore cannot accurately represent all possible input data for the super-resolution model.

[0131] Secondly, in the field of communications, super-resolution models face the complexity and diversity of the real world. The environment and conditions in communication networks often change, such as the update and deployment of network equipment in communication networks, fluctuations in data transmission rates, changes in channel quality, and the presence of interference and noise. However, when general technology is used to train super-resolution models, the training data set obtained by the equipment usually only contains sample data in a specific environment and cannot cover the rich and diverse network performance data of communication networks. That is, general super-resolution models cannot fully adapt to the changes in network performance data in communication networks, resulting in the inability to accurately predict and reconstruct fine-grained network performance data with high accuracy in different scenarios.

[0132] Thirdly, when anomalies or mutations occur in the communication network, network performance data often mutates. Especially for those fine-grained network performance data containing key information, since such fine-grained network performance data carries important business signals and key parameters, and is the key data of downstream business, when anomalies or mutations occur in the communication network, fine-grained network performance data can reflect the network performance of network equipment mutations. The input data of the super-resolution model is usually coarse-grained network performance data, so the general super-resolution model cannot accurately restore fine-grained network performance data.

[0133] For example, Figure 3 Another diagram of network performance data with different granularity is provided. Figure 3 As shown in , the solid line represents the fine-grained network performance data at different times, and the dotted line represents the coarse-grained network performance data at different times. Figure 3 It can be seen that coarse-grained network performance data cannot reflect the network performance when network devices in the communication network are abnormal or mutated. If coarse-grained network performance data is input into the super-resolution model, the super-resolution model cannot obtain enough mutation or anomaly information to accurately predict and reconstruct the details and features of fine-grained network performance data. This may cause the fine-grained network performance data output by the super-resolution model to be less accurate or distorted.

[0134] In summary, general super-resolution models have low generalization when acquiring fine-grained network performance data and cannot be applied to a variety of application scenarios. In addition, the accuracy of super-resolution models is low.

[0135] In response to the above problems, an embodiment of the present application provides a data processing method, in which coarse-grained network performance data of a communication network and time series characteristic data of the network performance data can be obtained.

[0136] Among them, the time series feature data can reflect the key performance of the network performance data from the dimension of time series features. The time series feature data may include one or more of the following data: feature vectors of network performance data (also known as encoding vectors, embedding vectors, encoding matrices, etc.), statistical data of network performance data, and fine-grained network performance data less than a preset number.

[0137] The feature vector of network performance data can represent the data content of network performance data from multiple dimensions, and the data volume of the feature vector is small, and the network resources occupied are also small. The above multiple dimensions may include the dimension of data trend, the dimension of data fluctuation, the dimension of data anomaly, the dimension of data mutation, the dimension of data content at each moment, etc.

[0138] The statistical data of the network performance data may be obtained by statistically processing the network performance data. The statistical data may represent the changing trend of the network performance data from the degree of dispersion (also referred to as the degree of dispersion). The statistical data may be at least one of the variance, standard deviation or mean of the network performance data.

[0139] The fine-grained network performance data that is less than a preset number is a small amount of fine-grained network performance data, wherein the preset number can be set according to the requirements of the actual application scenario.

[0140] Next, super-resolution processing is performed on the coarse-grained network performance data according to the time series feature data to obtain target data, wherein the target data includes fine-grained network performance data greater than a preset amount, that is, a large amount of fine-grained network performance data.

[0141] As can be seen from the above, compared with directly collecting a large amount of fine-grained network performance data, in the technical solution provided by the embodiment of the present application, the coarse-grained network performance data and timing characteristic data obtained are data with a small amount of data. A large amount of fine-grained network performance data (i.e., target data) is generated through coarse-grained network performance data and timing characteristic data with a small amount of data. In this way, the technical solution provided by the embodiment of the present application does not need to obtain a large amount of fine-grained network performance data, reducing the pressure of data collection.

[0142] Secondly, compared with the general technology that only performs super-resolution processing on coarse-grained network performance data to obtain fine-grained network performance data, the coarse-grained network performance data can be super-resolution processed with reference to the temporal feature data of the network performance data. In this way, the missing data of the coarse-grained network performance data can be supplemented by the temporal feature data of the network performance data, thereby improving the obtained fine-grained data.

[0143] The above data processing method can be applied to a data processing system. Figure 4 A structural diagram of a data processing system is provided. Figure 4 As shown, the data processing system includes: an NMS 401 and an EMS 402 which are communicatively connected.

[0144] like Figure 4 As shown, NMS 401 may send a coarse-grained network performance data acquisition request to EMS 402 .

[0145] In an embodiment of the present application, the coarse-grained network performance data acquisition request may also be referred to as a third data acquisition request. For ease of description, the third data acquisition request is used as an example for explanation below.

[0146] The EMS 402 may collect coarse-grained network performance data in response to the third data acquisition request, and send the coarse-grained network performance data to the NMS 401 .

[0147] NMS 401 may also send a request for obtaining time series characteristic data of network performance data to the EMS.

[0148] In an embodiment of the present application, a request for obtaining timing characteristic data of network performance data may also be referred to as a first data acquisition request. For ease of description, the following description will be given using the first data acquisition request as an example.

[0149] EMS 402 may also generate time series characteristic data according to the collected original full amount of network performance data in response to the first data acquisition request, and send the time series characteristic data to NMS 401 .

[0150] In one possible implementation, Figure 4 As shown, when the time series feature data includes a feature vector of network performance data, a pre-trained feature extraction model is deployed in EMS 402. EMS 402 can determine the feature vector of the network performance data based on the feature extraction model.

[0151] After receiving the coarse-grained network performance data and the time series feature data of the network performance data, the NMS 401 may perform super-resolution processing on the coarse-grained network performance data to obtain target data.

[0152] In one possible implementation, Figure 4 As shown, a pre-trained super-resolution model may be deployed in NMS 401. NMS 401 may determine target data based on the super-resolution model.

[0153] In the case where a feature extraction model is deployed in the EMS 402 and a super-resolution model is deployed in the NMS 401 , the NMS 401 may further send a verification data acquisition request of the target data to the EMS 402 .

[0154] In the embodiment of the present application, the verification data acquisition request of the target data may also be referred to as a second data acquisition request. For ease of description, the second data acquisition request is taken as an example for explanation below.

[0155] In response to the second data acquisition request, the EMS 402 may generate verification data of the target data according to the collected original full amount of network performance data, and send the verification data to the NMS 401 .

[0156] NMS 401 can verify the target data based on the verification data, and update the super-resolution model when the matching degree between the verification data and the target data is less than a preset matching degree, and / or the difference between the data application results corresponding to the verification data and the data application results corresponding to the timing feature data is less than a preset difference.

[0157] In one practicable manner, the NMS 401 may also obtain and update the feature extraction model, and send the updated feature extraction model to the EMS 402 .

[0158] In one practicable manner, the feature extraction model deployed in the EMS 402 and the super-resolution model deployed in the NMS 401 may be pre-trained by a model training device and sent to the EMS 402 and the NMS 401. In this case, Figure 4 As shown, the data processing system may further include: a model training device 403 communicatively connected to the NMS 401 and the EMS 402 .

[0159] The model training device 403 can obtain training data for training the super-resolution model and the feature extraction model from the EMS 402, and train the super-resolution model and the feature extraction model based on the training data. After the training is completed, the model training device 403 can send the super-resolution model to the NMS 401 and send the feature extraction model to the EMS 402.

[0160] The training data may include first sample data and second sample data, the first sample data is training data for training a super-resolution model, and the second sample data is training data for training a feature extraction model.

[0161] In one possible implementation, the model training device 403, NMS 401 and EMS 402 may be independent and different physical devices, or the functions of the model training device 403, the functions of the NMS 401 and the functions of the EMS 402 may be integrated on the same physical device, or part of the functions of the model training device 403, part of the functions of the NMS 401 and part of the functions of the EMS 402 may be integrated on one physical device, and the embodiments of the present application are not limited to this.

[0162] For ease of description, the embodiment of the present application is described by taking the model training device 403, NMS 401 and EMS 402 as three independent physical devices as an example.

[0163] In addition, the network architecture and business scenarios described in the embodiments of the present application are intended to more clearly illustrate the technical solutions of the embodiments of the present application, and do not constitute a limitation on the technical solutions provided in the embodiments of the present application. Ordinary technicians in this field can know that with the evolution of network architecture and the emergence of new business scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0164] The data processing method provided in the embodiment of the present application is described below in conjunction with the accompanying drawings.

[0165] Figure 5 A flowchart of a data processing method provided in an embodiment of the present application. Figure 5 As shown, the data processing method includes:

[0166] S501. The NMS obtains coarse-grained network performance data of the communication network.

[0167] The communication network is a communication network for which a network digital twin needs to be constructed. For example, when managing the network of City A, it is necessary to construct a network digital twin of the communication network of City A to achieve refined modeling and management of the communication network of the real City A. In this case, the communication network is the communication network of City A.

[0168] In one practicable manner, the EMS is used to manage the original full network performance data of each network device in the communication network. The EMS can collect the coarse-grained network performance data of the communication network from the original full network performance data of the managed communication network, and send the collected coarse-grained network performance data to the NMS. Correspondingly, the NMS can receive the coarse-grained network performance data from the EMS. For example, the NMS can periodically obtain the coarse-grained network performance data from the EMS through a subscription mode.

[0169] In another achievable manner, the NMS may send a third data acquisition request to the EMS to request the EMS to send the coarse-grained network performance data of the communication network to the NMS. After receiving the third data acquisition request sent by the NMS, the EMS may directly collect the coarse-grained network performance data and send the coarse-grained network performance data to the NMS. Accordingly, the NMS may receive the coarse-grained network performance data from the EMS. For example, the NMS may obtain the coarse-grained network performance data from the EMS through a polling mode.

[0170] In this implementation, the third data acquisition request may include various types of information of the required coarse-grained network performance data. In this way, the NMS can obtain accurate coarse-grained network performance data and improve the accuracy of determining the target data.

[0171] Optionally, the third data acquisition request may include a specific collection granularity of coarse-grained network performance data (ie, the first granularity in the embodiment of the present application).

[0172] In this implementation, the third data acquisition request also includes: the data type of the required coarse-grained network performance data, and / or the time period of the required coarse-grained network performance data.

[0173] Exemplarily, referring to the description of the above network performance data, taking the network performance data as the CPU usage of the network equipment in the communication network as an example, when the coarse-grained network performance data that the NMS needs to obtain is: the CPU usage collected once every 1 minute during the evening time period (19:00-22:00) when the user terminal traffic volume is large, the third data acquisition request may include: the coarse-grainedness of "collected once every 1 minute", the data type of "CPU usage", and the time period of "19:00-22:00". In this way, the NMS can collect the CPU usage of the time period of "19:00-22:00" from the original full amount of network performance data of the managed communication network with a collection frequency of 1 minute to obtain the corresponding coarse-grained network performance data.

[0174] S502: The NMS obtains time series characteristic data of network performance data.

[0175] In one practicable manner, the EMS can determine the timing characteristic data based on the original full amount of network performance data of each network device in the communication network, and send the timing characteristic data of the network performance data to the NMS. Accordingly, the NMS can receive the timing characteristic data of the network performance data from the EMS. For example, the NMS can periodically obtain the timing characteristic data of the network performance data from the EMS through a subscription mode.

[0176] In another achievable manner, the NMS may send a first data acquisition request to the EMS to request the EMS to send the timing characteristic data of the network performance data to the NMS. After receiving the first data acquisition request sent by the NMS, the EMS may generate the timing characteristic data based on the original full amount of network performance data, and send the generated timing characteristic data to the NMS. Accordingly, the NMS may receive the timing characteristic data from the EMS. For example, the NMS may obtain the timing characteristic data from the EMS through a polling mode.

[0177] In this implementation, the first data acquisition request may include various types of information of the required time series feature data. In this way, the NMS can obtain accurate time series feature data and improve the accuracy of determining the target data.

[0178] Optionally, the first data acquisition request may include: a data type of the network performance data corresponding to the required timing characteristic data, and / or a time period of the network performance data corresponding to the required timing characteristic data.

[0179] Exemplarily, referring to the description of the above network performance data, taking the network performance data as the CPU usage rate of the network equipment in the communication network as an example, when the time series characteristic data that the NMS needs to obtain is: the time series characteristic data of the CPU usage rate in the evening time period (19:00-22:00) when the user terminal traffic volume is large, the first data acquisition request may include: the data type of "CPU usage rate" and the time period of "19:00-22:00". In this way, the NMS can generate the time series characteristic data of the CPU usage rate in the time period of "19:00-22:00" in the original full amount of network performance data of the managed communication network.

[0180] In some embodiments, referring to the description of the above time series feature data, taking the time series feature data including a feature vector of network performance data as an example, the method for EMS to generate time series feature data based on the original full amount of network performance data specifically includes:

[0181] A trained feature extraction model can be pre-deployed in EMS. The feature extraction model has the ability to extract feature vectors of network performance data based on network performance data. When it is necessary to send feature vectors of network performance data to NMS, EMS can input the original full amount of network performance data into the feature extraction model to obtain the feature vector of network performance data and use it as the time series feature data of network performance data. For example, continuing with the above example, EMS can input the original full amount of CPU usage in the time period of "19:00-22:00" into the feature extraction model to obtain the feature vector of the CPU usage.

[0182] In some embodiments, referring to the description of the above time series feature data, taking the time series feature data including statistical data of network performance data as an example, the method for EMS to generate time series feature data according to the original full amount of network performance data specifically includes:

[0183] When it is necessary to send statistical data of network performance data to NMS, EMS can perform statistical processing on the original full amount of network performance data to obtain statistical data of network performance data and use it as time series feature data of network performance data.

[0184] Optionally, the above statistical processing can be variance calculation of network performance data, standard deviation calculation of network performance data, or other statistical processing methods, which are not limited in the embodiments of the present application. For example, continuing with the above example, the EMS can calculate the variance of the CPU usage rate in the time period of "19:00-22:00" of the original full amount to obtain the statistical data of the CPU usage rate.

[0185] In some embodiments, referring to the description of the above time series feature data, taking the time series feature data including fine-grained network performance data less than a preset amount as an example, the method for EMS to generate time series feature data based on the original full amount of network performance data specifically includes:

[0186] When it is necessary to send fine-grained network performance data less than a preset amount to the NMS, the EMS can extract fine-grained network performance data less than a preset amount from the original full amount of network performance data as time series feature data of the network performance data.

[0187] Exemplarily, assuming that the original full amount of network performance data includes the CPU usage from 00:00 to 23:59, the EMS can extract the CPU usage from 20:00 to 21:00 from the CPU usage from 00:00 to 23:59 as the time series feature data of the network performance data.

[0188] It can be understood that in the above S501 and S502, the coarse-grained network performance data of the communication network and the timing characteristic data of the network performance data are obtained from the EMS. Therefore, the NMS can obtain the coarse-grained network performance data and the timing characteristic data of the network performance data by sending a data acquisition request to the EMS.

[0189] For example, the NMS may send a data acquisition request to the EMS for obtaining coarse-grained network performance data and time series characteristic data of the network performance data. After receiving the data acquisition request sent by the NMS, the EMS may collect the coarse-grained network performance data of the communication network, and process the original full amount of network performance data to obtain the time series characteristic data of the network performance data. Afterwards, the EMS may send the coarse-grained network performance data and the time series characteristic data of the network performance data to the NMS. Correspondingly, the NMS may receive the coarse-grained network performance data and the time series characteristic data of the network performance data from the EMS.

[0190] Of course, the NMS may also send two data acquisition requests to the EMS (such as the third data acquisition request and the first data acquisition request mentioned above), respectively for requesting coarse-grained network performance data and time series characteristic data of the network performance data. The embodiment of the present application does not specifically limit the implementation method of acquiring coarse-grained network performance data and time series characteristic data of the network performance data from the EMS.

[0191] It should be noted that when the NMS sends two data acquisition requests to the EMS, the embodiment of the present application does not limit the order in which the NMS obtains the timing characteristic data of the network performance data and the coarse-grained network performance data of the communication network.

[0192] S503 . The NMS performs super-resolution processing on the coarse-grained network performance data according to the time series feature data to obtain target data.

[0193] The target data includes fine-grained network performance data greater than a preset amount.

[0194] After obtaining the time series feature data and the coarse-grained network performance data, since the time series feature data of the network performance data can supplement the missing data of the coarse-grained network performance data, the NMS can perform super-resolution processing on the coarse-grained network performance data according to the time series feature data to obtain fine-grained network performance data greater than a preset amount, i.e., the target data. In this way, the technical solution provided by the embodiment of the present application does not need to obtain a large amount of fine-grained network performance data, reducing the pressure of data collection. Secondly, the time series feature data of the network performance data can supplement the missing data of the coarse-grained network performance data, thereby improving the accuracy of the target data.

[0195] In one achievable manner, the NMS performs super-resolution processing on the coarse-grained network performance data according to the time series feature data to obtain the target data, specifically including:

[0196] a1. NMS obtains super-resolution parameters.

[0197] Specifically, in order to improve the accuracy of the super-resolution model, in this embodiment, different super-resolution parameters correspond to different super-resolution models. Based on this, in order to generate more accurate large amounts of fine-grained data, the NMS can obtain the corresponding super-resolution parameters.

[0198] The super-resolution parameters include one or more of the following parameters: the type of data to be processed with super-resolution, the time period of the data to be processed with super-resolution, and the super-resolution ratio.

[0199] Optionally, the time period of the data to be processed by super-resolution may be a time period corresponding to a communication network failure. The super-resolution ratio may be a super-resolution ratio required by downstream services, such as 4 times or 16 times.

[0200] Exemplarily, when the target data that the NMS needs to obtain is a large amount of CPU usage, the type of data to be super-resolution processed in the super-resolution parameter may be the CPU usage.

[0201] a2. NMS obtains the pre-trained super-resolution model corresponding to the super-resolution parameters.

[0202] Among them, the super-resolution model has the ability to obtain fine-grained network performance data based on time series feature data and coarse-grained network performance data.

[0203] Specifically, the NMS may pre-deploy multiple super-resolution models, and different super-resolution models correspond to different super-resolution parameters.

[0204] Exemplarily, the NMS may be pre-deployed with a super-resolution model 1 and a super-resolution model 2. The super-resolution model 1 is used to determine the target data corresponding to the CPU usage rate, and the super-resolution model 2 is used to determine the target data corresponding to the memory occupancy rate.

[0205] When it is determined that the type of data to be processed by super-resolution in the super-resolution parameters is CPU usage, the NMS may determine the super-resolution model 1 as a pre-trained super-resolution model corresponding to the super-resolution parameters.

[0206] a3. NMS inputs the time series feature data and coarse-grained network performance data into the super-resolution model to obtain the target data.

[0207] Specifically, NMS can perform super-resolution processing on coarse-grained network performance data according to the pre-trained super-resolution model corresponding to the super-resolution parameters based on the time series feature data to obtain target data. In this way, NMS can perform super-resolution processing on coarse-grained network performance data according to the pre-trained super-resolution model corresponding to the acquired super-resolution parameters and the time series feature data to obtain target data corresponding to the super-resolution parameters, thereby improving the flexibility and accuracy of generating target data.

[0208] In another achievable manner, the NMS performs super-resolution processing on the coarse-grained network performance data according to the time series feature data to obtain the target data, specifically including:

[0209] b1. NMS inputs the time series feature data and coarse-grained network performance data into the pre-trained super-resolution model to obtain the target data.

[0210] The NMS may also deploy a general super-resolution model that has the ability to obtain fine-grained network performance data based on time series feature data and coarse-grained network performance data.

[0211] After acquiring the time series feature data and the coarse-grained network performance data, the NMS may also input the time series feature data and the coarse-grained network performance data into a pre-trained general super-resolution model to obtain target data.

[0212] For example, the coarse-grained network performance data obtained by the NMS is: 10 CPU usage rates collected once every 1 minute; the time series feature data of the network performance data obtained by the NMS is: the feature vectors of the above 10 CPU usage rates. Then, the NMS can perform super-resolution processing on the 10 CPU usage rates based on the feature vectors of the 10 CPU usage rates to obtain 100 CPU usage rates, i.e., the target data.

[0213] In some embodiments, after obtaining the target data, the NMS may also verify the accuracy of the target data and determine whether to update the super-resolution model and the feature extraction model based on the verification result. Figure 6 A flowchart of another data processing method provided in an embodiment of the present application. Figure 6 As shown, the data processing method includes:

[0214] S601: NMS obtains verification data of target data.

[0215] In one practicable manner, the EMS may acquire verification data of the target data based on the original full amount of network performance data of each network device in the management communication network, and send the collected verification data to the NMS.

[0216] The verification data includes fine-grained network performance data corresponding to the coarse-grained network performance data and greater than a preset amount.

[0217] For example, when the target data is the CPU usage rate in the time period of 15:00-23:00 on the current day, the verification data may be the CPU usage rate in the time period of 15:00-23:00 on the previous day. In this way, the target data may be accurately verified by the fine-grained network performance data corresponding to the coarse-grained network performance data and greater than a preset amount.

[0218] Accordingly, the NMS may receive the verification data of the target data from the EMS. For example, the NMS may periodically obtain the verification data of the target data from the EMS through a subscription mode.

[0219] In another achievable manner, the NMS may send a second data acquisition request to the EMS to request the EMS to send verification data to the NMS. After receiving the second data acquisition request sent by the NMS, the EMS may determine the verification data based on the original full amount of network performance data and send the verification data to the NMS. Accordingly, the NMS may receive the verification data from the EMS. For example, the NMS may obtain the verification data from the EMS in a polling mode.

[0220] In this implementation, the second data acquisition request may include various types of information of the required verification data. In this way, the NMS can obtain accurate verification data, thereby improving the accuracy of the verification target data.

[0221] Optionally, since the target data is fine-grained network performance data, the verification data also needs to be fine-grained network performance data. In this case, the second data acquisition request may include fine granularity (eg, may be a specific required acquisition frequency).

[0222] Optionally, since the verification data verifies the target data, the data type of the verification data may be the same as the data type of the target data. In this case, the second data acquisition request further includes: the data type of the target data.

[0223] Optionally, in order to improve the accuracy of the verification result, the second data acquisition request further includes: the number of required verification data, wherein the number of required verification data can be set according to the requirements of the actual application scenario.

[0224] Optionally, in order to verify the accuracy of the result, the amount of verification data required may be greater than or equal to the amount of target data. In this way, the accuracy of the verification result can be improved by a large amount of verification data.

[0225] S602: When the verification data and the target data meet preset conditions, the NMS updates the super-resolution model.

[0226] Among them, the preset conditions include: the matching degree between the verification data and the target data is less than the preset matching degree, and / or the result difference between the data application result corresponding to the verification data and the data application result corresponding to the timing feature data is less than the preset difference.

[0227] Specifically, NMS can verify the target data from multiple aspects such as accuracy, consistency, and stability.

[0228] When the accuracy and / or consistency of the target data and the verification data need to be verified, the NMS may determine whether the matching degree between the verification data and the target data is less than a preset matching degree.

[0229] When the matching degree between the verification data and the target data is less than the preset matching degree, it means that the accuracy of the target data is low. In this case, the NMS can update the super-resolution model.

[0230] Correspondingly, when the matching degree between the verification data and the target data is greater than or equal to the preset matching degree, it indicates that the accuracy of the target data is high. In this case, the NMS does not need to update the super-resolution model.

[0231] In one possible implementation, NMS can determine the match between the validation data and the target data based on the Mean Absolute Percentage Error (MAPE) in the Train on Synthetic Test on Real (TSTR) evaluation metric.

[0232] When the stability of the target data needs to be verified, the NMS can determine whether the result difference between the data application result corresponding to the verification data and the data application result corresponding to the time series feature data is less than a preset difference.

[0233] Specifically, the NMS can send target data and verification data to the downstream business device, so that the downstream business device can determine and send two data application results to the NMS based on the target data and verification data (for example, building a first network digital twin based on the verification data, and building a second network digital twin based on the target data). Then, the NMS can compare the two data application results to obtain the result difference.

[0234] When the result difference is greater than or equal to the preset difference, it means that the stability of the target data is low. In this case, NMS can update the super-resolution model.

[0235] When the result difference is smaller than the preset difference, it indicates that the target data is more stable. In this case, NMS does not need to update the super-resolution model.

[0236] In one achievable manner, the NMS may update the super-resolution model according to a preset loss function.

[0237] In another possible implementation, the staff may perform an update operation on the NMS based on historical experience. Accordingly, the NMS may update the super-resolution model in response to the staff's update operation.

[0238] In some embodiments, the EMS is usually a gateway for simple data processing and data forwarding, so the NMS can also obtain the feature extraction model deployed in the EMS and determine whether to update the feature extraction model according to the verification result. In this case, the data processing method also includes:

[0239] S603: The NMS obtains a feature extraction model and updates the feature extraction model.

[0240] As can be seen from the above, when the verification data and the target data meet the preset conditions, it means that the accuracy of the feature extraction model may be relatively low. In the case that the EMS cannot update the feature extraction model, the NMS can obtain the feature extraction model and update the feature extraction model.

[0241] In one practicable manner, the NMS may update the feature extraction model according to a preset loss function.

[0242] In another possible implementation, the staff may perform an update operation on the NMS based on historical experience. Accordingly, the NMS may update the feature extraction model in response to the staff's update operation.

[0243] S604: The NMS sends the updated feature extraction model to the EMS. Correspondingly, the EMS receives the updated feature extraction model from the NMS.

[0244] In this way, EMS can determine the feature vector of the network performance data according to the updated feature extraction model, thereby improving the accuracy of the feature vector.

[0245] In some embodiments, in the above data processing method, the super-resolution model used by the NMS may be pre-trained by a model training device and sent to the NMS. In this case, Figure 7 A flow chart of a model training method provided in an embodiment of the present application. Figure 7 As shown, the model training method includes:

[0246] S701. The model training device obtains first sample data.

[0247] The first sample data may include sample network performance data of the first network and time series feature data of the sample network performance data. The sample network performance data of the first network may include coarse-grained sample network performance data and fine-grained sample network performance data.

[0248] In one possible implementation, the first sample data obtained by the model training device is Figure 5 The network performance data and timing characteristic data obtained in the data processing flow shown in the figure may be data in the same communication network. In this case, the first network may be Figure 5 The communication network in the data processing flow shown is the same network. In this way, the model training device can train the first sample data under the communication network to obtain a super-resolution model suitable for the communication network.

[0249] In one possible implementation, the first sample data acquired by the model training device may also be data in a communication network specifically used for model training. In this case, the first network may be Figure 5 The communication networks in the data processing flow shown are different networks. In this way, the model training device does not need to obtain the first sample data of the current communication network in each communication network.

[0250] The model training device may obtain the sample network performance data of the first network from the EMS. For example, the model training device may send a data acquisition request to the EMS to request the EMS to send the sample network performance data of the first network to the model training device.

[0251] Regarding the time series characteristic data of the sample network performance data, in one achievable manner, the model training device may process the sample network performance data obtained from the EMS to obtain the time series characteristic data of the sample network performance data.

[0252] In another possible implementation, the model training device may obtain the time series characteristic data of the sample network performance data from the EMS. In this implementation, the model training device may send a data acquisition request to the EMS to request the EMS to send the time series characteristic data of the sample network performance data to the model training device.

[0253] It can be understood that in this implementation, the timing characteristic data and sample network performance data of the sample network performance data are both obtained from the EMS. Therefore, the model training device can obtain the timing characteristic data and sample network performance data of the sample network performance data by sending a data acquisition request to the EMS.

[0254] For example, the model training device may send a fourth data acquisition request to the EMS. The fourth data acquisition request is used to request the acquisition of the above-mentioned first sample data. After receiving the fourth data acquisition request sent by the model training device, the EMS may process the sample network performance data to obtain the timing characteristic data of the sample network performance data. Afterwards, the EMS may send the first sample data including the sample network performance data and the timing characteristic data of the sample network performance data to the model training device. Accordingly, the model training device may receive the first sample data from the EMS.

[0255] In this implementation, the fourth data acquisition request may include various types of information of the required sample data. In this way, the model training device can obtain accurate first sample data, thereby improving the accuracy of the sample data for training the super-resolution model.

[0256] Optionally, the fourth data acquisition request includes one or more of the following information: data type of the required first sample data, quantity of the required first sample data, and time period of the required first sample data.

[0257] Of course, the model training device can also send two data acquisition requests to the EMS, respectively requesting the EMS to send sample network performance data and timing characteristic data of the sample network performance data to the model training device. The embodiment of the present application does not specifically limit the implementation method of obtaining sample network performance data and timing characteristic data of the sample network performance data from the EMS.

[0258] It should be noted that the specific introduction of the sample network performance data can refer to the introduction of the network performance data in the embodiments of the present application, which will not be repeated here.

[0259] S702: The model training device obtains a super-resolution model based on the first sample data.

[0260] After obtaining the first sample data, the model training device can train a super-resolution model based on the first sample data. Since the first sample data includes coarse-grained sample network performance data, fine-grained sample network performance data, and time series feature data of the sample network performance data, the trained super-resolution model has the ability to obtain fine-grained network performance data based on time series feature data and coarse-grained network performance data.

[0261] Optionally, the model training device can obtain a super-resolution model based on a convolutional neural network (CNN) algorithm, or based on a recurrent neural network (RNN) algorithm, or based on other artificial intelligence (AI) algorithms, but the embodiments of the present application do not limit this.

[0262] S703: The model training device sends the super-resolution model to the NMS.

[0263] After the super-resolution model is trained, the model training device can send the super-resolution model to the NMS. In this way, the NMS only needs to collect coarse-grained network performance data and time series feature data with a small amount of data, and can generate a large amount of fine-grained network performance data through the super-resolution model. The NMS does not need to obtain a large amount of fine-grained network performance data, which reduces the pressure of data collection on the NMS.

[0264] In some embodiments, during the above data processing method, the feature extraction model used by the EMS may be pre-trained by a model training device and sent to the EMS. In this case, Figure 8 A flow chart of a model training method provided in an embodiment of the present application. Figure 8 As shown, the model training method includes:

[0265] S801. The model training device obtains second sample data.

[0266] The second sample data includes network performance data of the second network.

[0267] Similar to the first sample data, the second network can be Figure 5 The communication networks in the illustrated embodiments may be different or the same.

[0268] In addition, the model training device can also obtain the first sample data and the second sample data under the same communication network to train the super-resolution model and the feature extraction model. In this case, the first network and the second network can be the same communication network.

[0269] The model training device may obtain the second sample data from the EMS. For example, the model training device may send a data acquisition request to the EMS to request the EMS to send the second sample data to the model training device.

[0270] For example, the model training device may send a fifth data acquisition request to the EMS. The fifth data acquisition request is used to request the acquisition of the second sample data. After receiving the fifth data acquisition request sent by the model training device, the EMS may collect the second sample data. Afterwards, the EMS may send the second sample data to the model training device. Accordingly, the model training device may receive the second sample data from the EMS.

[0271] In this implementation, the fifth data acquisition request may include various types of information of the required sample data. In this way, the model training device can obtain accurate second sample data, thereby improving the accuracy of the sample data for training the feature extraction model.

[0272] Optionally, the fifth data acquisition request includes one or more of the following information: data type of the required second sample data, quantity of the required second sample data, and time period of the required second sample data.

[0273] S802: The model training device obtains a feature extraction model based on the second sample data training.

[0274] Optionally, the model training device may extract a feature vector of the second sample data based on a feature extraction algorithm. Then, the model training device may obtain a feature extraction model based on the second sample data and the feature vector of the second sample data.

[0275] S803: The model training device sends the feature extraction model to the EMS.

[0276] Accordingly, the EMS receives the feature extraction model from the model training device. After receiving the feature extraction model from the model training device, the EMS can determine the feature vector of the network performance data based on the feature extraction model.

[0277] In actual applications, EMS is usually only used for data processing and forwarding, and does not have the ability to update models. Therefore, after training the feature extraction model, the model training device can also send the feature extraction model to the NMS for the NMS to update the feature extraction model. After updating the feature extraction model, the NMS can send the updated feature extraction model to the EMS for the EMS to extract more accurate feature vectors based on the updated feature extraction model.

[0278] Of course, the model training device can also send only the feature extraction model to the EMS after training the feature extraction model. When the NMS determines that the feature extraction model needs to be updated, it can send a model acquisition request to the model training device. In this case, the model training device can send the feature extraction model to the NMS in response to the model acquisition request, so that the NMS can update the feature extraction model.

[0279] The above mainly introduces the scheme of the embodiment of the present application from the perspective of each specific embodiment. The following describes in detail the overall process of the data processing method provided by the embodiment of the present application by taking the different modules used to implement the different functions of the above-mentioned specific embodiments as an example. Fig. 9 As shown, NMS901 may include: a control module 9011, a super-resolution processing module 9012, a model updating module 9013, a data verification module 9014 and a network digital twin construction module 9015.

[0280] The EMS 902 may include: a coarse-grained data acquisition module 9021 , a time series feature data processing module 9022 , and a fine-grained data acquisition module 9023 .

[0281] In one achievable method, during super-resolution processing of coarse-grained network performance data (i.e., the data super-resolution stage), NMS 901, such as the control module 9011 of NMS 901, sends a first data acquisition request to EMS 902, such as the timing feature data processing module 9022 of EMS902, to request to obtain timing feature data of the network performance data.

[0282] The NMS 901 , such as the control module 9011 of the NMS 901 , sends a third data acquisition request to the EMS 902 , such as the coarse-grained data acquisition module 9021 of the EMS 902 , to request to acquire the coarse-grained network performance data of the communication network.

[0283] EMS 902, such as the timing characteristic data processing module 9022 of EMS902, generates timing characteristic data according to the collected original full network performance data in response to the first data acquisition request, and sends the timing characteristic data to NMS 901, such as to the super-resolution processing module 9012 of NMS 901.

[0284] EMS 902 , such as coarse-grained data collection module 9021 of EMS 902 , collects coarse-grained network performance data in response to the third data acquisition request, and sends the coarse-grained network performance data to NMS 901 , such as to super-resolution processing module 9012 of NMS 901 .

[0285] NMS 901 , such as the super-resolution processing module 9012 of NMS 901 , performs super-resolution processing on the coarse-grained network performance data according to the time series feature data to obtain target data, and sends the target data to the network digital twin construction module 9015 of NMS 901 .

[0286] NMS 901, such as the network digital twin construction module 9015 of NMS 901, constructs a network digital twin according to the received target data.

[0287] In another achievable manner, the NMS 901 may also update the super-resolution model and the feature extraction model in the super-resolution process (i.e., the model update phase). In this case, the NMS 901, such as the control module 9011 of the NMS 901, sends a second data acquisition request to the EMS 902, such as the fine-grained data acquisition module 9023 of the EMS 902, to request to obtain verification data of the target data, and sends a data verification request to the data verification module 9014 of the NMS 901 to request to verify the accuracy of the target data.

[0288] EMS 902 , such as fine-grained data collection module 9023 of EMS 902 , generates verification data of target data according to collected network performance data in response to the second data acquisition request, and sends the verification data to NMS 901 , such as to data verification module 9014 of NMS 901 .

[0289] NMS 901, such as data verification module 9014 of NMS 901, obtains verification data from EMS 902, such as fine-grained data acquisition module 9023 of EMS 902, and target data from super-resolution processing module 9012 of NMS 901 in response to the data verification request, and verifies the target data according to the verification data.

[0290] The NMS 901 , such as the data verification module 9014 of the NMS 901 , sends a model update strategy to the control module 9011 of the NMS 901 according to the verification result.

[0291] When the model update policy is not to update the model, the NMS 901, such as the control module 9011 of the NMS 901, does not perform any operation.

[0292] When the model update strategy is to update the model, the NMS 901 , such as the control module 9011 of the NMS 901 , sends a model update request to the model update module 9013 of the NMS 901 .

[0293] Optionally, a super-resolution model is deployed in the NMS 901, such as the super-resolution processing module 9012 of the NMS 901. When the time series feature data includes feature vectors of network performance data, a feature extraction model is deployed in the EMS 902, such as the time series feature data processing module 9022 of the EMS 902.

[0294] The NMS 901, such as the model updating module 9013 of the NMS 901, updates the super-resolution model in the super-resolution processing module 9012 of the NMS 901 in response to the received model updating request.

[0295] NMS 901, such as model updating module 9013 of NMS 901, obtains a feature extraction model and updates the feature extraction model in response to the received model update request, and sends the updated feature extraction model to EMS 902, such as to temporal feature data processing module 9022 of EMS 902.

[0296] The above mainly introduces the scheme of the embodiment of the present application from the perspective of method. It is understandable that, in order to realize the above functions, the data processing device includes a hardware structure and / or software module corresponding to the execution of each function. Those skilled in the art should easily realize that, in combination with the units and algorithm steps of each example described in the embodiments disclosed herein, the embodiments of the present application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a function is executed in the form of hardware or computer software driving hardware depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the embodiments of the present application.

[0297] The embodiment of the present application can divide the data processing device into functional units according to the above method example. For example, each functional unit can be divided according to each function, or two or more functions can be integrated into one processing unit. The above integrated unit can be implemented in the form of hardware or in the form of software functional units. It should be noted that the division of units in the embodiment of the present application is schematic and is only a logical function division. There may be other division methods in actual implementation.

[0298] The present application embodiment provides a data processing device (referred to as data processing device 100), which can be applied to the above-mentioned NMS. Fig.10 As shown, the data processing device 100 includes: a communication unit 1001 and a processing unit 1002. Optionally, the data processing device 100 also includes a storage unit 1003. The storage unit 1003 is used to store program codes and data of the data processing device 100.

[0299] The communication unit 1001 is used to obtain network performance data of a first granularity of the communication network.

[0300] The communication unit 1001 is also used to obtain time series characteristic data of the network performance data, and the time series characteristic data includes one or more of the following data: a characteristic vector of the network performance data, statistical data of the network performance data, and network performance data of a second granularity less than a preset number, the characteristic vector is used to represent the data content of the network performance data from multiple dimensions, and the data collection frequency corresponding to the second granularity is greater than the data collection frequency corresponding to the first granularity.

[0301] The processing unit 1002 is used to perform super-resolution processing on the network performance data of the first granularity according to the time series feature data to obtain target data, where the target data includes network performance data of the second granularity greater than a preset number.

[0302] In one achievable manner, the communication unit 1001 is specifically configured to:

[0303] Receive timing characteristic data from the equipment management system EMS.

[0304] In one achievable manner, the communication unit 1001 is further used to send a first data acquisition request to the EMS, where the first data acquisition request is used to request to acquire time series feature data.

[0305] In one achievable manner, the first data acquisition request includes: the data type of the required time series characteristic data, and / or the time period of the required time series characteristic data.

[0306] In one achievable manner, the processing unit 1002 is specifically configured to:

[0307] Obtain super-resolution parameters, which include one or more of the following parameters: the type of data to be super-resolution processed, the time period of the data to be super-resolution processed, and the super-resolution ratio; perform super-resolution processing on the network performance data of the first granularity according to the super-resolution parameters based on the time series feature data to obtain the target data.

[0308] In one achievable manner, the processing unit 1002 is specifically configured to:

[0309] A pre-trained super-resolution model corresponding to super-resolution parameters is obtained, where the super-resolution model has the ability to obtain fine-grained network performance data based on time series feature data and coarse-grained network performance data; the time series feature data and the first-grained network performance data are input into the super-resolution model to obtain target data.

[0310] In one practicable manner, the network performance data includes one or more of the following data: central processing unit CPU utilization, memory occupancy, query rate per second (QPS), network request latency, network alarm information, network traffic, number of terminal connections, and data transmission rate.

[0311] In one achievable manner, the communication unit 1001 is further used to obtain verification data of the target data, where the verification data includes network performance data of a second granularity corresponding to the target data and greater than a preset number.

[0312] Processing unit 1002 is also used to update the super-resolution model when the verification data and the target data meet preset conditions, and the preset conditions include: the matching degree between the verification data and the target data is less than the preset matching degree, and / or the difference between the data application results corresponding to the verification data and the data application results corresponding to the timing feature data is less than the preset difference.

[0313] In one achievable manner, the communication unit 1001 is specifically configured to:

[0314] Receive verification data from EMS.

[0315] In one achievable manner, the communication unit 1001 is further used to send a second data acquisition request to the EMS, where the second data acquisition request is used to request verification data, and the second data acquisition request includes one or more of the following information: a second granularity, a data type of the target data, and a quantity of required verification data.

[0316] In one practicable manner, the communication unit 1001 is further used to obtain a feature extraction model, and the feature extraction model has the ability to extract feature vectors of network performance data based on the network performance data.

[0317] The processing unit 1002 is further configured to update the feature extraction model when the verification data and the target data meet a preset condition.

[0318] The communication unit 1001 is also used to send the updated feature extraction model to the EMS.

[0319] The present application also provides a data processing device (referred to as data processing device 110), which can be applied to the above-mentioned EMS. Fig.11 As shown, the data processing device 110 includes: a communication unit 1101 and a processing unit 1102. Optionally, the data processing device 110 further includes a storage unit 1103. The storage unit 1103 is used to store program codes and data of the data processing device 110.

[0320] The communication unit 1101 is used to send the network performance data of the first granularity of the communication network to the NMS.

[0321] The communication unit 1101 is further used to send time series characteristic data of the network performance data to the NMS, where the time series characteristic data includes one or more of the following data: a characteristic vector of the network performance data, statistical data of the network performance data, and network performance data of a second granularity less than a preset number, where the characteristic vector is used to represent data content of the network performance data from multiple dimensions, and a data collection frequency corresponding to the second granularity is greater than a data collection frequency corresponding to the first granularity;

[0322] The time series feature data is used by the NMS to perform super-resolution processing on the received network performance data of the first granularity to obtain target data, and the target data includes network performance data of the second granularity greater than a preset number.

[0323] In one practicable manner, the communication unit 1101 is further configured to receive a first data acquisition request from the NMS, where the first data acquisition request is used to request acquisition of time series feature data.

[0324] In one achievable manner, the first data acquisition request includes: the data type of the required time series characteristic data, and / or the time period of the required time series characteristic data.

[0325] In one achievable manner, the processing unit 1102 is used to input the network performance data into a feature extraction model to obtain a feature vector of the network performance data; the feature extraction model has the ability to extract the feature vector of the network performance data based on the network performance data.

[0326] In one achievable method, the communication unit 1101 is also used to send verification data of the target data to the NMS, the verification data is used by the NMS to determine whether to update the super-resolution model and the feature extraction model, the verification data includes network performance data of a second granularity corresponding to the target data and greater than a preset number.

[0327] In one achievable manner, the communication unit 1101 is further used to receive a second data acquisition request from the NMS, where the second data acquisition request is used to request verification data, and the second data acquisition request includes one or more of the following information: a second granularity, a data type of the target data, and a quantity of required verification data.

[0328] In one practicable manner, the communication unit 1101 is further configured to receive an updated feature extraction model sent from the NMS.

[0329] Fig.10 and Fig.11 The units in the system may also be referred to as modules. For example, a processing unit may be referred to as a processing module. Fig.10 and Fig.11 In the illustrated embodiment, the names of the various units may not be the names shown in the figure. For example, the data processing unit may also be called a transceiver unit.

[0330] Fig.10 and Fig.11 If each unit in the embodiment is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor (processor) to perform all or part of the steps of the method described in each embodiment of the present application. The storage medium for storing computer software products includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program codes.

[0331] The present application also provides a hardware structure diagram of a data processing device, see Fig.12 The data processing device includes a processor 1201 and a transceiver 1202 , and optionally, also includes a memory 1203 connected to the processor 1201 .

[0332] Processor 1201 may be a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of the program of the present application. Processor 1201 may also include multiple CPUs, and processor 1201 may be a single-core (single-CPU) processor or a multi-core (multi-CPU) processor. The processor here may refer to one or more devices, circuits, or processing cores for processing data (such as computer program instructions).

[0333] The processor 1201, the memory 1203 and the transceiver 1202 are connected via a bus. The transceiver 1202 is used to communicate with other data processing devices. Optionally, the transceiver 1202 may include a transmitter and a receiver. The device used to implement the receiving function in the transceiver 1202 may be regarded as a receiver, and the receiver is used to perform the receiving step in the embodiment of the present application. The device used to implement the sending function in the transceiver 1202 may be regarded as a transmitter, and the transmitter is used to perform the sending step in the embodiment of the present application.

[0334] In the first possible implementation, see Fig.12, the data processing device also includes a transceiver 1202. The memory 1203 can be a ROM or other types of static storage devices that can store static information and instructions, a RAM or other types of dynamic storage devices that can store information and instructions, or an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical disc, laser disc, optical disc, digital versatile disc, Blu-ray disc, etc.), a disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of an instruction or data structure and can be accessed by a computer, and the embodiment of the present application does not impose any restrictions on this. The memory 1203 can exist independently or be integrated with the processor 1201. Among them, the memory 1203 can contain computer program code. The processor 1201 is used to execute the computer program code stored in the memory 1203, so as to realize the method provided in the embodiment of the present application.

[0335] An embodiment of the present application also provides a computer-readable storage medium, including instructions, which, when executed on a computer, enable the computer to execute any of the above methods.

[0336] The embodiment of the present application also provides a computer program product including instructions, which, when executed on a computer, enables the computer to execute any of the above methods.

[0337] An embodiment of the present application also provides a chip, including: a processor and an interface, wherein the processor is coupled to a memory via the interface, and when the processor executes a computer program or instruction in the memory, any one of the methods provided in the above embodiments is executed.

[0338] An embodiment of the present application also provides a data processing system, including: the terminal device and access network device in the above embodiment.

[0339] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using a software program, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. that contains one or more servers that can be integrated with the medium. The available medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a DVD), or a semiconductor medium (eg, a solid state disk (SSD)).

[0340] Although the present application is described herein in conjunction with various embodiments, in the process of implementing the claimed application, those skilled in the art may understand and implement other changes to the disclosed embodiments by viewing the drawings, the disclosure, and the appended claims. In the claims, the word "comprising" does not exclude other components or steps, and "one" or "an" does not exclude multiple situations. A single processor or other unit may implement several functions listed in the claims. Certain measures are recorded in different dependent claims, but this does not mean that these measures cannot be combined to produce good results.

[0341] Although the present application has been described in conjunction with features and embodiments thereof, it is apparent that various modifications and combinations may be made thereto without departing from the spirit and scope of the present application. Accordingly, this specification and the drawings are merely exemplary illustrations of the present application as defined by the appended claims, and are deemed to have covered any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, those skilled in the art may make various modifications and variations to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

[0342] The above is only an implementation method of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions within the technical scope disclosed in the present application shall be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A data processing method, characterized in that: Applied to a network management system NMS, the method comprises: Acquire first-granularity network performance data of the communication network; Acquire time series characteristic data of the network performance data, the time series characteristic data including one or more of the following data: a characteristic vector of the network performance data, statistical data of the network performance data, and network performance data of a second granularity less than a preset number, the characteristic vector being used to represent data content of the network performance data from multiple dimensions, and a data collection frequency corresponding to the second granularity being greater than a data collection frequency corresponding to the first granularity; Based on the time series feature data, super-resolution processing is performed on the network performance data of the first granularity to obtain target data, and the target data includes network performance data of a second granularity greater than the preset number.

2. The data processing method according to claim 1, characterized in that: The acquiring of the timing characteristic data of the network performance data comprises: The timing characteristic data is received from an equipment management system EMS.

3. The data processing method according to claim 2, characterized in that: Before receiving the timing characteristic data from the equipment management system EMS, the method further includes: A first data acquisition request is sent to the EMS, where the first data acquisition request is used to request to acquire the timing characteristic data.

4. The data processing method according to claim 3, characterized in that: The first data acquisition request includes: the data type of the network performance data corresponding to the required timing characteristic data, and / or the time period of the network performance data corresponding to the required timing characteristic data.

5. The data processing method according to any one of claims 1 to 4, characterized in that: The performing super-resolution processing on the network performance data of the first granularity according to the time series feature data to obtain target data includes: Acquire super-resolution parameters, wherein the super-resolution parameters include one or more of the following parameters: the type of data to be processed by super-resolution, the time period of the data to be processed by super-resolution, and the super-resolution ratio; According to the time series feature data, super-resolution processing is performed on the network performance data of the first granularity according to the super-resolution parameters to obtain the target data.

6. The data processing method according to claim 5, characterized in that: The performing super-resolution processing on the network performance data of the first granularity according to the super-resolution parameter to obtain the target data includes: Obtaining a pre-trained super-resolution model corresponding to the super-resolution parameters, wherein the super-resolution model has the ability to obtain fine-grained network performance data based on time series feature data and coarse-grained network performance data; The time series feature data and the network performance data of the first granularity are input into the super-resolution model to obtain the target data.

7. The data processing method according to any one of claims 1 to 6, characterized in that: The network performance data includes one or more of the following data: central processing unit CPU usage, memory occupancy, query rate per second (QPS), network request latency, network alarm information, network traffic, number of terminal connections, and data transmission rate.

8. The data processing method according to claim 6, characterized in that: After performing super-resolution processing on the network performance data of the first granularity according to the time series feature data to obtain target data, the method further includes: Acquire verification data of the target data, the verification data including network performance data of a second granularity corresponding to the target data and greater than the preset number; When the verification data and the target data meet preset conditions, the super-resolution model is updated, and the preset conditions include: the matching degree between the verification data and the target data is less than a preset matching degree, and / or the difference between the data application result corresponding to the verification data and the data application result corresponding to the timing feature data is less than a preset difference.

9. The data processing method according to claim 8, characterized in that: The obtaining verification data of the target data includes: The verification data is received from the EMS.

10. The data processing method according to claim 9, characterized in that: Prior to receiving the verification data from the EMS, the method further includes: A second data acquisition request is sent to the EMS, where the second data acquisition request is used to request the verification data, and the second data acquisition request includes one or more of the following information: the second granularity, the data type of the target data, and the amount of required verification data.

11. The data processing method according to any one of claims 8 to 10, characterized in that: The method further comprises: Acquire a feature extraction model, wherein the feature extraction model has the ability to extract a feature vector of the network performance data based on the network performance data; When the verification data and the target data meet the preset condition, updating the feature extraction model; Send the updated feature extraction model to EMS.

12. A data processing method, characterized in that: Applied to an equipment management system (EMS), the method comprises: Sending first-granularity network performance data of the communication network to a network management system NMS; Sending time series characteristic data of the network performance data to the NMS, the time series characteristic data including one or more of the following data: a feature vector of the network performance data, statistical data of the network performance data, and network performance data of a second granularity less than a preset number, the feature vector being used to represent data content of the network performance data from multiple dimensions, and a data collection frequency corresponding to the second granularity being greater than a data collection frequency corresponding to the first granularity; The time series feature data is used by the NMS to perform super-resolution processing on the received network performance data of the first granularity to obtain target data, and the target data includes network performance data of a second granularity greater than the preset number.

13. The data processing method according to claim 12, characterized in that: Before sending the timing characteristic data of the network performance data to the NMS, the method further includes: A first data acquisition request is received from the NMS, where the first data acquisition request is used to request to acquire the time series feature data.

14. The data processing method according to claim 13, characterized in that: The first data acquisition request includes: the data type of the network performance data corresponding to the required timing characteristic data, and / or the time period of the network performance data corresponding to the required timing characteristic data.

15. The data processing method according to any one of claims 12 to 14, characterized in that: In the case where the time series feature data includes a feature vector of the network performance data, the method further includes: The network performance data is input into a feature extraction model to obtain a feature vector of the network performance data; the feature extraction model has the ability to extract the feature vector of the network performance data based on the network performance data.

16. The data processing method according to any one of claims 12 to 15, characterized in that: The method further comprises: Verification data of the target data is sent to the NMS, where the verification data is used by the NMS to determine whether to update the super-resolution model and the feature extraction model, and the verification data includes network performance data of a second granularity corresponding to the target data and greater than the preset number.

17. The data processing method according to claim 16, characterized in that: Before sending the verification data to the NMS, the method further includes: A second data acquisition request is received from the NMS, where the second data acquisition request is used to request the verification data, and the second data acquisition request includes one or more of the following information: the second granularity, the data type of the target data, and the amount of required verification data.

18. The data processing method according to claim 16 or 17, characterized in that: The method further comprises: The updated feature extraction model is received from the NMS.

19. A data processing device, characterized in that: include: A functional unit for executing the method according to any one of claims 1 to 11, or a functional unit for executing the method according to any one of claims 12 to 18; The actions performed by the functional units are implemented through hardware or through hardware executing corresponding software implementations.

20. A network management system NMS, characterized in that: include: processor; The processor is connected to a memory, the memory is used to store computer-executable instructions, and the processor executes the computer-executable instructions stored in the memory so that the data processing device implements the method according to any one of claims 1 to 11.

21. An equipment management system EMS, characterized in that: include: processor; The processor is connected to a memory, the memory is used to store computer-executable instructions, and the processor executes the computer-executable instructions stored in the memory so that the data processing device implements the method according to any one of claims 12 to 18.

22. A data processing system, characterized in that: include: A network management system NMS as claimed in claim 20 and an equipment management system EMS as claimed in claim 21.

23. A computer-readable storage medium, characterized in that: The method comprises instructions, which, when executed on a computer, enable the computer to execute the method according to any one of claims 1 to 11, or enable the computer to execute the method according to any one of claims 12 to 18.

Images