A network slice validity determination method, a computer device and a storage medium

Abstract

The application discloses a network slice effectiveness determination method, a computer device and a storage medium, which comprises the following steps: acquiring a first effectiveness time sequence of a first network slice; inputting the first effectiveness time sequence into a LightGBM model; and determining an effectiveness prediction result of the first network slice in a future time period according to a processing result output by the LightGBM model. The application uses the LightGBM model to quickly respond to a time-varying communication environment, adaptively analyzes and learns, can predict the effectiveness of the first network slice in the future time period, and can configure the validity period of the first network slice based on the prediction result, so that the validity period of the first network slice can be kept at a level suitable for the task to be carried, the task can be normally executed without a too-short validity period, or the network is not wasted and the security risk of network attack is not faced with a too-long validity period. The application is widely applied in the technical field of communication.

Description

Technical Field

[0001] This invention relates to the field of communication technology, and in particular to a method for determining the validity of network slicing, a computer device, and a storage medium. Background Technology

[0002] Network slicing technology is applied in 5G and more advanced communication technologies. Taking 5G communication networks as an example, network slicing technology can be applied not only to the three major application scenarios of eMBB, uRLLC, and mMTC, but also to application scenarios such as virtual reality, augmented reality, autonomous driving, and smart cities. The basic principle of network slicing technology is to divide the physical network into multiple virtual networks, adapting to the network connectivity characteristics of different fields, thus greatly improving the utilization rate of communication resources, providing more possibilities for operators and users, facilitating the convergence and interconnection of various industries, and providing a more powerful network service quality.

[0003] Network slicing technology features isolation, achieving network isolation between slices. Each slice has a Network Slice Selection Auxiliary Information (S-NSSAI), which identifies a network slice. When opening an account, a user subscribes to one or more S-NSSAIs on the UDM. According to relevant 3GPP standards, the AMF in the core network should support the function of indicating the validity period of the S-NSSAI to the user terminal.

[0004] In communication networks, a common scenario is that the network side needs to provide slice services for a limited time. This can be achieved by setting the validity period of the S-NSSAI (Slice Selection Auxiliary Information). However, existing technologies cannot accurately determine the validity period of the S-NSSAI. As a result, the validity period of the network slice does not match the time of the slice service. The network slice may expire before the slice service ends, resulting in the inability to provide normal slice services. Alternatively, the network slice may expire a long time after the slice service ends, making the network slice, which no longer carries slice services, vulnerable to network attacks and other risks, posing potential security vulnerabilities. Moreover, maintaining the network slice after the slice task has been completed will waste network resources.

[0005] Terminology Explanation:

[0006] AMF: Access and Mobility Management Function;

[0007] NSSF: Network Slice Selection Function.

[0008] 3GPP: 3rd Generation Partnership Project;

[0009] UE: User Equipment;

[0010] UDM: Unified Data Management;

[0011] SUPI: Subscription Permanent Identifier

[0012] SUCI: Subscription Concealed Identifier;

[0013] S-NSSAI: Single Network Slice Selection Assistance Information;

[0014] GPSI: Generic Public Subscription Identifier;

[0015] Configured NSSAI: Configured Network Slice Selection Assistance Information; Configured NSSAI eMBB: Enhanced Mobile Broadband;

[0016] uRLLC: Ultra-Reliable Low-Latency Communications;

[0017] mMTC: Massive Machine Type Communication;

[0018] LightGBM: Light Gradient Boosting Machine;

[0019] Leaf-wise: grows according to leaves. Summary of the Invention

[0020] In view of the technical problems in current communication technologies, such as the network slice validity period being set too long or too short, the purpose of this invention is to provide a method for determining the validity of network slices, a computer device, and a storage medium.

[0021] On one hand, embodiments of the present invention include a method for determining the validity of network slices, comprising:

[0022] Load the LightGBM model;

[0023] Obtain the first validity time series of the first network slice;

[0024] The first effective time series is input into the LightGBM model;

[0025] Obtain the processing results output by the LightGBM model;

[0026] Based on the processing results, the effectiveness prediction result of the first network slice in the future time period is determined.

[0027] Furthermore, the method for determining the validity of network slices also includes:

[0028] Based on the validity prediction results, the validity period prediction results of the first network slice in the future time period are determined.

[0029] Further, determining the validity period prediction result of the first network slice in a future time period based on the validity prediction result includes:

[0030] Based on the time segments in the validity prediction results, the validity period prediction results are obtained by weighted summation according to the corresponding valid values.

[0031] Furthermore, the method for determining the validity of network slices also includes:

[0032] The LightGBM model is trained.

[0033] Furthermore, training the LightGBM model includes:

[0034] Obtain the second validity time series of several second network slices; the second validity time series includes valid values ​​arranged in chronological order, and the valid values ​​are used to represent the validity of the second network slice within the corresponding time segment;

[0035] For any second-validity time series, the second-validity time series is decomposed into a first time series and a second time series; the time segment corresponding to the valid value in the first time series precedes the time segment corresponding to the valid value in the second time series;

[0036] The LightGBM model is trained by using the first time series as input and the second time series as the expected output.

[0037] Further, obtaining the second validity time series of each of the several second network slices includes:

[0038] Obtain a second network slice set; the second network slice set includes multiple second network slices;

[0039] Obtain the attribute information of each of the second network slices in the second network slice set;

[0040] Determine the correlation between attribute variables and validity variables; the attribute variables are variables formed by the presence or absence of attribute information, and the validity variables are variables formed by the changes in the validity values ​​in each of the second validity time series.

[0041] The corresponding attribute information with a relevance greater than a threshold is detected;

[0042] The second network slice having the attribute information is deleted from the second network slice set; or, the second network slice not having the attribute information is deleted from the second network slice set.

[0043] Obtain the second validity time series of each of the second network slices in the second network slice set.

[0044] Furthermore, determining the correlation between the attribute variable and the validity variable includes:

[0045] A chi-square test was performed on the attribute variables and the validity variables to obtain the chi-square values;

[0046] The chi-square value is used as the correlation.

[0047] Furthermore, training the LightGBM model further includes:

[0048] Construct a retraining dataset; the retraining dataset includes several first effective time series and several second effective time series;

[0049] The LightGBM model is retrained using the retraining dataset.

[0050] On the other hand, embodiments of the present invention also include a computer device, including a memory and a processor, the memory being used to store at least one program, and the processor being used to load the at least one program to execute a network slice validity determination method according to the embodiments.

[0051] On the other hand, embodiments of the present invention also include a storage medium storing a processor-executable program, which, when executed by a processor, is used to perform a network slice validity determination method in the embodiments.

[0052] The beneficial effects of the present invention are as follows: The network slice validity determination method in the embodiments uses the LightGBM model, which is suitable for feature extraction and time-series prediction, to quickly react to and adaptively analyze and learn from the time-varying communication environment. It can predict the validity of the first network slice in the future time period. Based on the prediction results, the validity period of the first network slice can be configured, so that the validity period of the first network slice can be kept at a level that is adapted to the task it is to carry. This avoids the situation where the validity period is set too short, resulting in the inability to perform the task normally, or the validity period is set too long, resulting in security risks such as network attacks and waste of network resources. Attached Figure Description

[0053] Figure 1 This is a flowchart illustrating the steps of the method for determining the validity of network slices in the embodiment;

[0054] Figure 2 This is a schematic diagram illustrating the principle of the network slice validity determination method in the embodiment;

[0055] Figure 3 This is a flowchart illustrating the method for determining the validity of network slices using the 5G core network in this embodiment. Detailed Implementation

[0056] In this embodiment, refer to Figure 1 A method for determining the validity of network slices includes the following steps:

[0057] S1. Load the LightGBM model;

[0058] S2. Obtain the first validity time series of the first network slice;

[0059] S3. Input the first effective time series into the LightGBM model;

[0060] S4. Obtain the processing results output by the LightGBM model;

[0061] S5. Based on the processing results, determine the effectiveness prediction results of the first network slice in the future time period.

[0062] The principle of steps S1-S5 is as follows: Figure 2 As shown.

[0063] Reference Figure 2 The LightGBM model loaded in step S1 works by providing a function to predict the future validity period of S-NSSAI, including a histogram-based decision tree algorithm, a depth-limited leaf-wise algorithm, a one-sided gradient sampling algorithm, and a mutually exclusive feature binding algorithm.

[0064] Reference Figure 2 In the LightGBM model, a histogram-based decision tree algorithm is used to traverse and find the optimal split point for continuous features. The histogram algorithm discretizes the continuous floating-point feature values ​​into k integers and constructs a histogram of width k. During data traversal, the discretized values ​​are used as indices to accumulate statistics in the histogram. After one data traversal, the histogram has accumulated the necessary statistics. Then, based on the discrete values ​​in the histogram, the optimal split point is found. The input is continuous floating-point feature value data. The output is the optimal split point.

[0065] Reference Figure 2 In the LightGBM model, a depth-constrained leaf-wise algorithm is used to reduce computational overhead. Each time the decision tree splits, it selects the leaf that yields the maximum gain for splitting. A one-sided gradient sampling algorithm is used to filter samples using the calculated gradient. Mutually exclusive feature bundling is used to losslessly merge features using sparsity. The leaf-wise algorithm optimizes the model by repeatedly finding the leaf with the maximum splitting gain and then splitting it. The input is the decision tree model, and the output is the optimized decision tree model.

[0066] Reference Figure 2 In the LightGBM model, the one-sided gradient sampling algorithm is a method of sampling the training dataset to reduce the complexity of calculating the objective function gain. First, the first a% of training samples with large gradient values ​​are selected, and then b% of the remaining (1-a%) training samples with small gradient values ​​are randomly selected. The input is the training data, and the output is the sample data after one-sided gradient sampling of the training data.

[0067] Reference Figure 2 In the LightGBM model, the mutually exclusive feature bundling algorithm is used to solve the problem of sparse data. It merges data from different dimensions together to transform a sparse matrix into a dense matrix. First, the features that are best suited for fusion are identified, and then these features are merged together. Input: Features F, maximum conflict measure K. Output: Merged features.

[0068] After obtaining the LightGBM model through step S1, it can be trained. The steps for training the LightGBM model include:

[0069] S101. Obtain the second validity time series of each of the several second network slices;

[0070] S102. For any second-validity time series, decompose the second-validity time series into a first time series and a second time series;

[0071] S103. Train the LightGBM model by using the first time series as input and the second time series as the expected output of the LightGBM model.

[0072] In step S101, the second network slice can be the same as the first network slice or a different network slice. For each second network slice, one or more corresponding second valid time series are found. In this embodiment, the form of the second valid time series is shown in Table 1.

[0073] Table 1

[0074]

[0075] Referring to Table 1, S-NSSAI is used to label the second network slice, meaning that each second network slice has a corresponding S-NSSAI. In Table 1, each S-NSSAI has a corresponding second validity time series with valid values ​​for time segments T1, T2, etc., where a valid value of "1" indicates that the second network slice is valid in the corresponding time segment, and a valid value of "0" indicates that the second network slice is invalid in the corresponding time segment.

[0076] For example, the second network slice with S-NSSAI of 10101 has a corresponding second validity time series of 1, 1, 1, 0, 0, 0, 0, 0, which means that through actual testing, the second network slice was recorded as valid in time segment T1, valid in time segment T2, valid in time segment T3, invalid in time segment T4, invalid in time segment T5, invalid in time segment T6, invalid in time segment T7, invalid in time segment T8, and invalid in time segment T9.

[0077] In step S102, for any second valid time series, the second valid time series is decomposed into a first time series and a second time series. For example, the portion of any second valid time series corresponding to time segment T1-T7 can be divided into the first time series, and the portion corresponding to time segment T8-T9 can be divided into the second time series.

[0078] For example, for the second network slice with S-NSSAI of 10106, its corresponding second valid time series is 0, 0, 0, 1, 1, 0, 0, 0, and the corresponding first time series is 0, 0, 0, 1, 1, 0, 0, and the second time series is 0, 0; for the second network slice with S-NSSAI of 10108, its corresponding second valid time series is 1, 1, 0, 0, 0, 0, 0, 0, and the corresponding first time series is 1, 1, 0, 0, 0, 0, 0, and the second time series is 0, 0.

[0079] In step S103, the first time series is used as the input of the LightGBM model, and the corresponding second time series is used as the expected output of the LightGBM model to train the LightGBM model.

[0080] For example, the first time series 0, 0, 0, 1, 1, 0, 0 obtained by dividing the same second effective time series is input into the LightGBM model. The LightGBM model processes the first time series using the current parameters and outputs the actual output results. The error function is calculated based on the actual output results and the second time series 0, 0 corresponding to the first time series 0, 0, 1, 1, 0, 0. If the error function does not converge, the parameters of the LightGBM model are adjusted, and the next set of first and second time series is retrieved to continue training. Otherwise, the training of the LightGBM model is terminated.

[0081] By using the second effective time series obtained from the actual recording of the second network slice, a first time series and a second time series, which are also obtained from the actual recording, can be obtained. These can be used as input data and expected output data for training the LightGBM model, respectively. Thus, the LightGBM model can be trained based on the actual results, enabling the LightGBM model to process a time series composed of effective values ​​of a historical period and output a time series representing effective values ​​of a future period. This gives the model the ability to predict the effectiveness of a network slice in the future.

[0082] In this embodiment, when performing step S101, which is to obtain the second effective time series of each of the several second network slices, the following steps can be specifically performed:

[0083] S10101. Obtain the second network slice set;

[0084] S10102. Obtain the attribute information of each second network slice in the second network slice set;

[0085] S10103. Determine the correlation between attribute variables and validity variables;

[0086] S10104. Detect the corresponding attribute information whose relevance is greater than the threshold;

[0087] S10105. Delete the second network slice with attribute information from the second network slice set, or delete the second network slice without attribute information from the second network slice set;

[0088] S10106. Obtain the second validity time series of each second network slice in the second network slice set.

[0089] In step S10101, a second network slice set is formed by searching multiple second network slices that can record the second valid time series. That is, the second network slice set includes multiple second network slices.

[0090] In step S10102, each second network slice in the second network slice set has its own attribute information. This attribute information can describe attributes such as latency, throughput, security, bandwidth characteristics, and the type of service carried by the second network slice. In this embodiment, "service type" is used as an example of attribute information for illustration.

[0091] In step S10102, the attribute information in the form of service type specifically includes conversation services (e.g., voice calls, video conferencing), streaming media services (e.g., online music, online video), interactive services (e.g., web browsing, online games), and background services (e.g., email, SMS). That is, "attribute information" corresponds to a variable, namely the attribute variable, which can specifically take the values ​​of conversation services, streaming media services, interactive services, and background services.

[0092] In this embodiment, the valid value can be "0" (indicating that the second network slice is invalid in the corresponding time segment) or "1" (indicating that the second network slice is valid in the corresponding time segment). That is, the valid value corresponds to a variable, namely the validity variable, which can specifically take the value of 0, 1, etc.

[0093] In step S10103, the correlation between the attribute variable and the validity variable is determined. The correlation indicates the degree of independence between the attribute variable and the validity variable; a higher correlation indicates a greater influence of the attribute variable's value on the validity variable's value.

[0094] Specifically, when performing step S10103, a chi-square test can be performed on the attribute variable and the validity variable, and the obtained chi-square value can be used as the correlation between the attribute variable and the validity variable.

[0095] In this embodiment, the validity variable can take two values, "0" or "1", and is itself a binary variable; while the attribute variable has values ​​such as session service, streaming media service, interactive service, and background service. The attribute variable can be represented as "session service" and "non-session service", thus representing the attribute variable as a binary variable. For the validity variable and attribute variable, which are both binary variables, the chi-square test of the 2×2 contingency table shown in Table 2 can be used.

[0096] Table 2

[0097] <![CDATA[T n The effective value within the time segment is 1. <![CDATA[T n The valid value within the time segment is 0. total Session services a b a+b Non-session services c d c+d total a+c b+d a+b+c+d

[0098] The significance of Table 2 is: In the second network slice set, "T" n The valid value within the time segment is 1, and the number of second network slices carrying session services is a; "T n The valid value within the time segment is 0, and the number of second network slices carrying session services is b; n The number of second network slices whose valid values ​​are 1 within a time segment and which do not carry session services is c; n The number of second network slices whose valid values ​​within a time segment are 0 and which do not carry session services is d.

[0099] For the data shown in Table 2, the formula can be used. Calculate the chi-square value χ 2 , with chi-square value χ 2 As a correlation between validity variables and attribute variables.

[0100] In step S10103, based on the principle of the chi-square test and other variable correlation algorithm, the correlation (chi-square value χ²) is calculated. 2 The larger the relevance (chi-square value χ²), the greater the influence of the attribute variable on the validity variable of the same second network slice. 2 When the correlation is less than the threshold, it can be considered that the correlation is small enough and the validity variables and attribute variables are close to being independent of each other.

[0101] In step S10104, taking Table 2 as an example, if the calculated chi-square value χ 2If the value is greater than the threshold, it indicates that the validity variable of the second network slice (the validity of the second network slice within a certain time segment) is greatly affected by the attribute information of the second network slice (whether it carries session services or not). Therefore, it can be inferred that if a set of second network slices contains both second network slices that carry session services and second network slices that do not carry session services, then the validity of the second network slices in this set is actually affected by the factor of "whether it carries session services or not". Therefore, the attribute information of "session services" is marked.

[0102] In step S10105, taking Table 2 as an example, the calculated chi-square value χ has been determined. 2 If the value is greater than the threshold, then you can perform either operation A or B:

[0103] A. Remove all second network slices carrying session services from the second network slice set;

[0104] B. Remove all second network slices that do not carry session services from the second network slice set.

[0105] By executing operation A, all second network slices remaining in the second network slice set are second network slices that do not carry session services; by executing operation B, all second network slices remaining in the second network slice set are second network slices that carry session services. Whether operation A or operation B is executed, the execution of session services by the second network slices in the second network slice set is unified. After operation A or operation B, the validity of each second network slice in the second network slice set is no longer affected by the factor of "whether or not it carries session services," thus highlighting the independence of network slices. When executing step S10106, a second validity time series conforming to relevant communication standards can be obtained. This data set can be used to train the LightGBM model, which is beneficial for obtaining a LightGBM model that can predict validity according to communication standards.

[0106] In this embodiment, when describing steps S10101-S10106, the "attribute variable" is explained using the session service as an example. Alternatively, other values ​​can be used for the "attribute variable" to execute steps S10101-S10106.

[0107] After completing steps S101-S103 and training the LightGBM model, step S2 can be executed.

[0108] In step S2, for the first network slice, its corresponding first validity time series is obtained. Before inputting the first validity time series into the LightGBM model for processing, preprocessing can be performed using the data preprocessing module. The data preprocessing module provides data preprocessing functions, including data normalization, feature encoding and selection, and missing value handling. Based on the historical validity data of S-NSSAI obtained from a third party (including the valid values ​​of S-NSSAI and S-NSSAI by time series, 1 for valid and 0 for invalid), missing value handling is performed to obtain complete dimensional data, and then the data is normalized.

[0109] In step S2, the first effective time series can have a similar format to the second effective time series used for training, representing T1-T obtained from the measured records, respectively. m Within a given time segment, the validity of the first network slice is determined. The trained LightGBM model has the ability to process a time series of valid values ​​from a historical period and output a time series representing valid values ​​for a future period. This allows it to predict the validity of a network slice in the future. Therefore, in steps S3-S5, the first valid time series is input into the LightGBM model, and the processing result output by the LightGBM model represents the predicted validity of the first network slice in the future time segment.

[0110] In this embodiment, the processing result output by the LightGBM model has the same format as the second time series, for example, it can represent a future time period T. m+1 -T m+2 The validity of the first network slice within the same time segment.

[0111] In this embodiment, refer to Figure 2 It can set a result output module to provide a prediction result output display function, outputting the prediction result (1 or 0) of the effective value of the next time period corresponding to the S-NSSAI of the first network slice.

[0112] After completing steps S1-S5, the following steps can also be performed:

[0113] S6. Based on the validity prediction results, determine the validity prediction results of the first network slice in the future time period.

[0114] When executing step S6, since the processing result output by the LightGBM model can represent T m+1 -T m+2 The validity of the first network slice within the same time segment, i.e., the validity of the future time segment T. m+1 and T m+2 Each has its own valid value. Taking the future time segment T as an example...m+1 The effective value is 1, and the future time segment T m+2 Taking an effective value of 0 as an example, it can be calculated using the formula T = T m+1 ×1+T m+2 ×0, and the validity period prediction result T is obtained. The validity period prediction result T can represent the predicted validity period of the first network slice in the future as T.

[0115] After obtaining the validity prediction results of the first network slice in the future time period, the validity period of the Default Configured S-NSSAI can be configured based on the prediction results. This ensures that the validity period of the first network slice is kept at a level that is appropriate for the tasks it is to carry out, avoiding the situation where the validity period is set too short, which would prevent the task from being executed normally, or the validity period is set too long, which would lead to security risks such as network attacks.

[0116] In this embodiment, refer to Figure 2 A model management module can be set up to provide model update and iteration functions for model retraining, update iteration, or prediction of multiple future time units (predicting multiple effective values ​​through iterative iteration). Specifically, after obtaining the first effective time series, if the format of the first effective time series is the same as that of the second effective time series, several first effective time series and several second effective time series can be extracted to form a retraining dataset. Referring to steps S101-S103, the LightGBM model is retrained using the retraining dataset, thereby enabling the LightGBM model to obtain iterative updates and maintain good predictive ability.

[0117] In this embodiment, the network slice validity determination method can be applied to the 5G core network, which includes network elements such as AMF, UDM, NSSF, and NWDAF. When executing steps S1-S5 of the network slice validity determination method, the processes performed by each network element in the core network are as follows: Figure 3 As shown, it specifically includes:

[0118] 1. The trained LightGBM prediction model is deployed on the NWDAF network element. First, the AMF sends a request to the UDM to obtain the validity period of S-NSSAI: Nudm_SDM_Get Request(SUPI, S-NSSAI).

[0119] 2. UDM sends a request to NSSF to obtain the validity period of S-NSSAI: Nnssf_NSSAIAvailability_GetRequest(S-NSSAI);

[0120] 3. NSSF sends prediction requests to the prediction models deployed on NWDAF;

[0121] 4. After the LightGBM model completes its predictions, it sends the results to NSSF;

[0122] 5. NSSF configures the validity period of S-NSSAI based on the prediction results and replies to UDM with the obtained parameter values;

[0123] 6. UDM replies to AMF with the obtained parameter values.

[0124] Through processes 1-6, the 5G core network can execute the network slice validity determination method in this embodiment, thereby setting an appropriate network slice validity period. This ensures that the validity period of the network slice is kept at a level that is appropriate for the tasks it is to carry out, avoiding situations where the validity period is set too short, resulting in the inability to perform tasks normally, or the validity period is set too long, leading to security risks such as network attacks and waste of network resources.

[0125] A computer program that executes a network slice validity determination method according to this embodiment can be written into a computer device or storage medium. When the computer program is read out and run, the network slice validity determination method according to this embodiment is executed, thereby achieving the same technical effect as the network slice validity determination method in the embodiment.

[0126] It should be noted that, unless otherwise specified, when a feature is referred to as "fixed" or "connected" to another feature, it can be directly fixed or connected to the other feature, or indirectly fixed or connected to the other feature. Furthermore, the descriptions of "upper," "lower," "left," and "right" used in this disclosure are only relative to the relative positional relationships of the various components of this disclosure in the accompanying drawings. The singular forms "a," "described," and "the" used in this disclosure are also intended to include the plural forms, unless the context clearly indicates otherwise. Moreover, unless otherwise defined, all technical and scientific terms used in this embodiment have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in this embodiment specification is only for describing particular embodiments and is not intended to limit the invention. The term "and / or" as used in this embodiment includes any combination of one or more of the associated listed items.

[0127] It should be understood that although the terms first, second, third, etc., may be used to describe various elements in this disclosure, these elements should not be limited to these terms. These terms are only used to distinguish elements of the same type from each other. For example, a first element may also be referred to as a second element without departing from the scope of this disclosure, and similarly, a second element may also be referred to as a first element. The use of any and all instances or exemplary language (“e.g.,” “such as,” etc.) provided in this embodiment is intended only to better illustrate embodiments of the invention and, unless otherwise required, does not impose a limitation on the scope of the invention.

[0128] It should be recognized that embodiments of the present invention can be implemented or carried out by computer hardware, a combination of hardware and software, or by computer instructions stored in a non-transitory computer-readable storage medium. The method can be implemented using standard programming techniques—including a non-transitory computer-readable storage medium configured with a computer program, wherein such a storage medium causes the computer to operate in a specific and predefined manner—according to the methods and drawings described in the specific embodiments. Each program can be implemented in a high-level procedural or object-oriented programming language to communicate with the computer system. However, if desired, the program can be implemented in assembly or machine language. In any case, the language can be a compiled or interpreted language. Furthermore, for this purpose, the program can run on a programmed application-specific integrated circuit (ASIC).

[0129] Furthermore, the procedures described in this embodiment can be performed in any suitable order unless otherwise indicated by this embodiment or clearly contradicted by the context. The procedures (or variations and / or combinations thereof) described in this embodiment can be executed under the control of one or more computer systems configured with executable instructions, and can be implemented by hardware or a combination thereof as code (e.g., executable instructions, one or more computer programs, or one or more applications) that commonly executes on one or more processors. The computer program includes a plurality of instructions executable by one or more processors.

[0130] Furthermore, the method can be implemented in any suitable type of computing platform, including but not limited to personal computers, minicomputers, mainframes, workstations, networked or distributed computing environments, standalone or integrated computer platforms, or in communication with charged particle tools or other imaging devices. Aspects of the invention can be implemented as machine-readable code stored on a non-transitory storage medium or device, whether removable or integrated into a computing platform, such as a hard disk, optical read and / or write storage medium, RAM, ROM, etc., such that it is readable by a programmable computer, and when the storage medium or device is read by the computer, it can be used to configure and operate the computer to perform the processes described herein. Furthermore, the machine-readable code, or portions thereof, can be transmitted via wired or wireless networks. The invention described in this embodiment includes these and other different types of non-transitory computer-readable storage media when such media comprises instructions or programs that implement the steps described above in conjunction with a microprocessor or other data processor. When programmed according to the methods and techniques described in the invention, the invention also includes the computer itself.

[0131] A computer program can be applied to input data to perform the functions described in this embodiment, thereby transforming the input data to generate output data stored in non-volatile memory. The output information can also be applied to one or more output devices, such as a display. In a preferred embodiment of the invention, the transformed data represents physical and tangible objects, including specific visual depictions of physical and tangible objects generated on the display.

[0132] The above description is merely a preferred embodiment of the present invention. The present invention is not limited to the above-described embodiments. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention, as long as they achieve the technical effects of the present invention by the same means, should be included within the scope of protection of the present invention. Within the scope of protection of the present invention, the technical solutions and / or implementation methods can have various modifications and variations.

Claims

1. A method for determining the validity of network slices, characterized in that, The method for determining the validity of network slices includes: Load the LightGBM model; The LightGBM model is trained. Obtain the first validity time series of the first network slice; The first effective time series is input into the LightGBM model; Obtain the processing results output by the LightGBM model; Based on the processing results, the effectiveness prediction result of the first network slice in the future time period is determined; The training of the LightGBM model includes: Obtain a second network slice set; the second network slice set includes multiple second network slices; Obtain the attribute information of each of the second network slices in the second network slice set; Determine the correlation between attribute variables and validity variables; the attribute variables are variables formed by the presence or absence of attribute information, and the validity variables are variables formed by changes in valid values ​​in each second validity time series. The corresponding attribute information with a relevance greater than a threshold is detected; The second network slice having the attribute information is deleted from the second network slice set; or, the second network slice not having the attribute information is deleted from the second network slice set. Obtain the second validity time series of each of the second network slices in the second network slice set; the second validity time series includes valid values ​​arranged in chronological order, the valid values ​​being used to represent the validity of the second network slice within the corresponding time segment; For any second-validity time series, the second-validity time series is decomposed into a first time series and a second time series; the time segment corresponding to the valid value in the first time series precedes the time segment corresponding to the valid value in the second time series; The LightGBM model is trained by using the first time series as input and the second time series as the expected output.

2. The method for determining the validity of network slices according to claim 1, characterized in that, The method for determining the validity of network slices also includes: Based on the validity prediction results, the validity period prediction results of the first network slice in the future time period are determined.

3. The method for determining the validity of network slices according to claim 2, characterized in that, The step of determining the validity period prediction result of the first network slice in a future time period based on the validity prediction result includes: Based on the time segments in the validity prediction results, the validity period prediction results are obtained by weighted summation according to the corresponding valid values.

4. The method for determining the validity of network slices according to claim 1, characterized in that, Determining the correlation between attribute variables and validity variables includes: Perform a chi-square test on the attribute variables and the validity variables to obtain the chi-square values; The chi-square value is used as the correlation.

5. The method for determining the validity of network slices according to claim 4, characterized in that, The training of the LightGBM model also includes: Construct a retraining dataset; the retraining dataset includes several first-effective time series and several second-effective time series; The LightGBM model is retrained using the retraining dataset.

6. A computer device, characterized in that, The device includes a memory and a processor, the memory being used to store at least one program, and the processor being used to load the at least one program to execute the network slice validity determination method according to any one of claims 1-5.

7. A computer-readable storage medium storing a processor-executable program, characterized in that, The processor-executable program, when executed by the processor, is used to perform the network slice validity determination method according to any one of claims 1-5.

Images