Network slicing methods, apparatus, electronic devices and storage media

Abstract

This invention provides a network slicing method, apparatus, electronic device, and storage medium. The method includes: acquiring a slice service work order sent by a slice management function network element; acquiring target requirement information based on the slice service work order; inputting the target requirement information into a classification model to acquire a target slice parameter template output by the classification model; generating a target instruction based on the target slice parameter template and the slice service work order; and sending the target instruction to a target base station to enable the target base station to execute the target instruction and create a network slice instance. The classification model is obtained by training on the requirement information of each sample and the slice parameter template corresponding to the sample requirement information. The network slicing method, apparatus, electronic device, and storage medium provided by this invention, by interfacing with NSMF (Network Service Management Network) to transfer slice requirements, can realize the automated and intelligent production and implementation of network slicing, improving the efficiency and timeliness of network slicing.

Description

Technical Field

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

[0002] Network slicing is a significant network architecture innovation in mobile communication technology, enabling the coexistence of multiple isolated and independent virtual networks on the same physical network infrastructure. Network slicing offers multiple performance advantages: it reduces capital expenditure in network deployment and operation; it enables service differentiation and guarantees service level agreements (SSAs) for each service type; and it increases the flexibility and adaptability of network management.

[0003] Currently, network slicing mainly relies on NSMF (Network Slice Management Function) network elements for automatic management and orchestration. Based on the orchestration results, staff configure base stations to generate network slices. Existing technologies suffer from cumbersome online and offline coordination processes, lack of intelligence, and uncontrollable flow of slice parameters, resulting in low slicing efficiency and the risk of mismatches or omissions. Summary of the Invention

[0004] This invention provides a network slicing method, apparatus, electronic device, and storage medium to address the shortcomings of low slicing efficiency in the prior art and achieve efficient automated network slicing.

[0005] In a first aspect, the present invention provides a network slicing method, comprising:

[0006] Obtain the slice service work order sent by the slice management function network element;

[0007] Based on the sliced ​​service work order, obtain the target requirement information;

[0008] Input the target requirement information into the classification model and obtain the target slice parameter template output by the classification model;

[0009] Based on the target slice parameter template and the slice business work order, generate the target instruction;

[0010] The target instruction is sent to the target base station so that the target base station executes the target instruction and creates a network slice instance;

[0011] The classification model is obtained by training based on the demand information of each sample and the slice parameter template corresponding to the demand information of the sample.

[0012] In one embodiment, before inputting the target demand information into the classification model and obtaining the target slice parameter template output by the classification model, the method further includes:

[0013] The classification model is obtained by training based on the sample requirement information and the slice parameter template corresponding to the sample requirement information.

[0014] In one embodiment, the classification model is trained based on the sample requirement information and the slice parameter template corresponding to the sample requirement information, specifically including:

[0015] Obtain the information gain of each attribute of the training sample dataset;

[0016] Select the attribute with the highest information gain;

[0017] Group the sample demand information with the same feature value of the attribute with the largest information gain into the same subset of training sample data;

[0018] The information gain of each attribute in each training sample data subset is obtained sequentially. The attribute with the largest information gain is selected, and the sample demand information with the same feature value of the attribute with the largest information gain is classified into the same training sample data subset. The decision tree is constructed by recursively splitting the subset, which serves as the classification model.

[0019] The sample dataset is a collection of the requirement information of each sample.

[0020] In one embodiment, before training the classification model based on the sample requirement information and the slice parameter template corresponding to the sample requirement information, the method further includes:

[0021] Obtain the slice parameter templates of the base stations deployed by each manufacturer to obtain the slice parameter template library;

[0022] The slice parameter template corresponding to the sample requirement information is a slice parameter template from the slice parameter template library.

[0023] In one embodiment, obtaining the target requirement information based on the slice service work order specifically includes:

[0024] Based on the sliced ​​service work order, obtain the original requirement information;

[0025] The original requirement information is processed by word segmentation and text recognition to obtain the target requirement information.

[0026] In one embodiment, generating the target instruction based on the target slice parameter template and the slice service work order specifically includes:

[0027] Based on the slice service work order, obtain the target information of the target base station;

[0028] Based on the target slice parameter template and the target information of the target base station, a target execution script is generated;

[0029] Generate the target instruction carrying the target execution script.

[0030] In one embodiment, sending the target instruction to the target base station specifically includes:

[0031] The target instruction is sent to the operation and maintenance center of the target base station manufacturer, so that the operation and maintenance center can issue the target instruction to the target base station.

[0032] In a second aspect, the present invention provides a network slicing apparatus, comprising:

[0033] The work order acquisition module is used to acquire slice service work orders sent by the slice management function network element;

[0034] The requirement acquisition module is used to acquire target requirement information based on the sliced ​​business work order;

[0035] The template matching module is used to input the target requirement information into the classification model and obtain the target slice parameter template output by the classification model.

[0036] The instruction generation module is used to generate target instructions based on the target slice parameter template and the slice business work order;

[0037] The slice execution module is used to send the target instruction to the target base station so that the target base station executes the target instruction and creates a network slice instance;

[0038] The classification model is obtained by training based on the demand information of each sample and the slice parameter template corresponding to the demand information of the sample.

[0039] Thirdly, the present invention provides an electronic device, including a processor and a memory storing a computer program, wherein the processor executes the computer program to implement the steps of any of the above-described network slicing methods.

[0040] Fourthly, the present invention provides a processor-readable storage medium storing a computer program for causing the processor to perform the steps of any of the above-described network slicing methods.

[0041] The network slicing method, apparatus, electronic device, and storage medium provided by this invention acquire slice service work orders sent by slice management function network elements, obtain target requirement information based on the slice service work orders, acquire target slice parameter templates corresponding to the target requirement information based on a pre-trained classification model, generate target instructions based on the target slice parameter templates and slice service work orders, and send the target instructions to the target base station so that the target base station executes the target instructions to create a network slice instance. By establishing an interface with NSMF, it seamlessly connects with NSMF to transfer slice requirements, automatically acquires the target slice parameter templates corresponding to the target requirement information, and enables the base station to execute the target instructions corresponding to the target slice parameter templates to create network slice instances. This enables automated and intelligent production and implementation of network slicing, improving the efficiency and timeliness of network slicing. Furthermore, it enables operators to achieve configurable and controllable slice parameter IT-based full-process management, adapting to the development needs of subsequent industry applications. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0043] Figure 1 This is a flowchart illustrating the network slicing method provided by the present invention;

[0044] Figure 2 This is a schematic diagram of the decision tree in the network slicing method provided by the present invention;

[0045] Figure 3 This is a schematic diagram illustrating the construction process of the slice parameter template library in the network slicing method provided by the present invention;

[0046] Figure 4 This is a schematic diagram of the application environment of the network slicing method provided by the present invention;

[0047] Figure 5 This is a flowchart illustrating the network slicing method provided by the present invention;

[0048] Figure 6 This is a schematic diagram of the network slicing device provided by the present invention;

[0049] Figure 7 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0051] The following is combined with Figures 1-6 The present invention describes a network slicing method, apparatus, electronic device, and storage medium.

[0052] Figure 1 This is a flowchart illustrating the network slicing method provided by the present invention. The following is a summary of the process. Figure 1 This application describes a network slicing method provided in its embodiments. For example... Figure 1 As shown, the method includes: Step 101, obtaining the slice service work order sent by the slice management function network element.

[0053] Specifically, the network slicing method provided in this embodiment of the invention is executed by the network slicing device provided in this invention.

[0054] The network slicing device provided in this embodiment of the invention connects to the slice management function (NSMF) network element through the NSMF interface and receives slice service work orders sent by the NSMF network element.

[0055] Optionally, NSMF network elements can proactively initiate slicing service work orders, or they can return slicing service work orders to the network slicing device based on query requests sent by the network slicing device.

[0056] For example, the NSMF network element can send new slice service work orders received in the current period to the network slicing device at preset intervals.

[0057] For example, after receiving the query request, the NSMF network element can, in response to the query request, send the new slice service work order received between the time of the last return of the slice service work order and the time of receiving the current query request to the network slicing device.

[0058] The slicing service order carries information about ToB (enterprise-oriented) slicing requirements. Based on the slicing service order, network slicing can be performed to obtain wireless sub-slices for ToB services. Therefore, this network slicing device can serve as a ToB parameter management platform.

[0059] Step 102: Obtain target requirement information based on slice business work orders.

[0060] Optionally, key information from the slice requirement information carried in the slice business work order can be extracted by performing text recognition on the slice business work order, and used as the target requirement information.

[0061] Optionally, the target requirement information may include the required values ​​for one or more business metrics.

[0062] Step 103: Input the target requirement information into the classification model and obtain the target slice parameter template output by the classification model.

[0063] The classification model is obtained by training based on the sample requirement information and the slice parameter template corresponding to the sample requirement information.

[0064] Specifically, a classification model is used to map the relationship between demand information and slice parameter templates.

[0065] Based on the classification model, the target requirement information can be matched with each slice parameter template in the slice parameter template, and the slice parameter template that matches the target requirement information can be determined as the target slice parameter template.

[0066] Understandably, before step 103, a classification algorithm can be used to train the algorithm, with each sample's demand information as the sample data and the slice parameter template corresponding to that sample's demand information as the classification label corresponding to that sample data.

[0067] A classification algorithm is based on a set of sample data with known classification labels. It trains a classifier to classify samples with unknown classification labels. Optionally, the classification algorithm can be a supervised machine learning algorithm.

[0068] The slice parameter template includes the values ​​of each slice parameter. For any two slice parameter templates, they may contain different slice parameters, or they may contain the same slice parameters but at least one slice parameter has a different value.

[0069] For example, a slice parameter template that only needs to meet QoS (Quality of Service) requirements may include the following slice parameters: RLC (Radio Link Control) transmission mode, maximum RLC retransmission count, logical channel group ID, MAC scheduling logical channel priority, 5QI value, slice home PLMN (Public Land Mobile Network), base station number, cell number, and the number of the Virtual Local Area Network (VLAN) mapped to the cell. For a slice parameter template that needs to meet QoS and low latency requirements, the following slice parameters may include: RLC transmission mode, maximum RLC retransmission count, logical channel group ID, MAC scheduling logical channel priority, SR (Scheduling Request) parameter mapping strategy, SR transmission period, slice identifier, slice amateur type, and slice division. For a slice parameter template that needs to meet QoS, low latency, and PRB (Physical Relay) requirements... ResourceBlock (Physical Resource Block) is a template for reserved slice parameters. The slice parameters may include RLC transmission mode, maximum RLC retransmission count, logical channel group ID, MAC scheduling logical channel priority, pre-scheduling policy, pre-scheduling data volume, minimum proportion of downlink resource allocation for slice group, uplink resource type of network slice group, maximum guaranteed proportion of uplink resource allocation for slice group, and minimum guaranteed proportion of uplink resource allocation for slice group.

[0070] Step 104: Generate target instructions based on the target slice parameter template and slice business work order.

[0071] Specifically, the slice service work order also carries target information for the target base station. This target information may include coverage area, TAC (Tracking Area Code), and VLAN ID, among other things.

[0072] The target instruction can carry target information of the target base station and the values ​​of each slice parameter included in the target slice parameter template.

[0073] Step 105: Send the target instruction to the target base station so that the target base station executes the target instruction and creates a network slice instance.

[0074] Specifically, the network slicing device can send target instructions to the target base station.

[0075] After receiving the target instruction, the target base station executes the instruction to complete the creation of the network slice instance.

[0076] Preferably, the network slice instance is a wireless slice instance.

[0077] The 3GPP protocol defines three types of network slices: eMBB, uRLLC, and mMTC. Each network slice type is designed for a specific type of service. For example, eMBB slices are for high-data-rate, high-mobility services; uRLLC slices can be used to handle high-reliability and low-latency communication scenarios; and mMTC slices can serve a large number of services with small data volumes, tolerable latency, and infrequent access (such as sensor and wearable device services).

[0078] This invention, through its embodiments, acquires slice service work orders sent by slice management function network elements. Based on these work orders, it obtains target requirement information and, using a pre-trained classification model, acquires the target slice parameter template corresponding to the target requirement information. Based on the target slice parameter template and the slice service work order, it generates target instructions and sends these instructions to the target base station, enabling the base station to execute the instructions and create a network slice instance. By establishing an interface with NSMF (Network Slice Management Function), it seamlessly integrates with NSMF to handle slice requests, automatically acquiring the target slice parameter template corresponding to the target requirement information. This allows the base station to execute the target instructions corresponding to the template to create a network slice instance, achieving automated and intelligent network slicing production and implementation, improving efficiency and response timeliness. Furthermore, it enables operators to manage configurable and controllable slice parameters across the entire IT process, adapting to the evolving needs of future industry applications.

[0079] Based on any of the above embodiments, before inputting the target demand information into the classification model and obtaining the target slice parameter template output by the classification model, the method further includes: training the classification model based on the demand information of each sample and the slice parameter template corresponding to the sample demand information to obtain the classification model.

[0080] Specifically, it can be obtained by training based on classification algorithms, such as artificial neural networks, decision trees, Taylor formulas, Naive Bayes, logistic regression, random forests, and support vector machines, using the demand information of each sample as the sample data and the slice parameter template corresponding to the demand information of that sample as the classification label of that sample data.

[0081] The embodiments of the present invention are based on a classification algorithm to train a classification model, which can then determine the target slice parameter template that best matches the target requirement information. This enables the base station to create network slice instances that better meet the target requirement information based on the target slice parameter template, resulting in higher accuracy of network slicing.

[0082] Based on any of the above embodiments, a classification model is obtained by training according to the sample demand information and the slice parameter template corresponding to the sample demand information. Specifically, this includes: obtaining the information gain of each attribute in the training sample dataset; selecting the attribute with the largest information gain; classifying the sample demand information with the same feature value of the attribute with the largest information gain into the same training sample data subset; sequentially obtaining the information gain of each attribute in each training sample data subset, selecting the attribute with the largest information gain, classifying the sample demand information with the same feature value of the attribute with the largest information gain into the same training sample data subset, and recursively splitting to construct a decision tree as the classification model; wherein, the sample dataset is a set composed of the sample demand information.

[0083] Specifically, the classification algorithm can be a decision tree classification algorithm, and the resulting classification model is a decision tree.

[0084] Decision tree learning typically involves three steps: feature selection, decision tree generation, and decision tree pruning. A decision tree consists of nodes and directed edges. There are two types of nodes: internal nodes and leaf nodes. Internal nodes represent test conditions for a feature or attribute (used to separate records with different characteristics), while leaf nodes represent a classification. Specifically, starting from the root node, a feature of an instance is tested. Based on the test result, the instance is assigned to its child node (i.e., an appropriate branch is selected). If a branch leads to a leaf node or another internal node, a new test condition is applied recursively until a leaf node is reached. Upon reaching a leaf node, the final classification result is obtained.

[0085] The branching criterion of a decision tree determines which attribute a current tree node should use as the branch attribute for the current training data. Generally, the information gain principle is used as the branching criterion. Information gain measures how much information a feature brings to the classification system; the more information it brings, the more important that attribute is.

[0086] In this embodiment of the invention, the attribute can be a business metric.

[0087] Information entropy is used to measure the uncertainty of things. The more uncertain something is, the greater its entropy. Specifically, the expression for the entropy of a random variable X is as follows:

[0088]

[0089] Where n represents the n different discrete values ​​of X, pi represents the probability that X takes the value i, and log is the logarithm to the base 2 or e.

[0090] The conditional entropy H(Y|X) of a random variable Y represents the uncertainty of a random variable Y given that the random variable X is known.

[0091]

[0092] Where, p i This indicates that X takes the value x. i The probability of.

[0093] Information gain: For a given attribute, the difference between having it and not having it in a classification system is the amount of information the system gains. The metric is how much information the attribute brings to the classification system; the more information it brings, the more important the gain.

[0094] The information gain of attribute A on training set D is g(D,A), which represents the difference between the entropy H(D) of set D and the conditional entropy H(D|A) of D given attribute A.

[0095] g(D,A)=H(D)-H(D|A)

[0096] Where H(D) represents the uncertainty of classifying dataset D; H(D|A) represents the uncertainty of D given gain A; H(D)-H(D|A) represents the degree to which gain A reduces the uncertainty of D.

[0097] The calculation process for information gain is as follows:

[0098] a) Calculate the information entropy H(D) of dataset D.

[0099]

[0100] Where K is the number of categories in the final classification. C k Let be the number of samples in the k-th class, and D be the total number of samples in the set.

[0101] b) Calculate the conditional entropy of dataset D.

[0102]

[0103] Where n represents the n possible values ​​of a certain feature A, and Di is the number of samples for the i-th value under feature A. ik D i The number of samples in the k-th category.

[0104] c) Calculate information gain

[0105] g(D,A)=H(D)-H(D|A)

[0106] The target demand information is generated as a list of elements, such as ['downlink speed 80Mbps', 'uplink speed 20Mbps', etc.]. Each element in the target demand information is processed using a decision tree, and template matching is performed through a decision tree-based classification model to finally obtain the target slice parameter template.

[0107] Figure 2 A decision tree is shown. Figure 2 The decision tree generation process is shown below. Table 1 shows the sample requirement information and the corresponding slice parameter template.

[0108] Table 1. Sample requirement information and corresponding slice parameter template table

[0109]

[0110] 1. Calculate the information entropy of the sample dataset before splitting.

[0111] The unsplit sample dataset contained 6 samples (i.e., sample requirement information). The calculation results of information entropy are shown in Table 2.

[0112] Table 2 Information Entropy Table of Sample Dataset Before Splitting

[0113] Template 1 Template 2 Template 3 Summation Sample size 2 3 1 6 p 0.3333 0.5 0.1667 1 log(p,2) -1.585 -1 -2.585 -- p*log(p,2) -0.528 -0.5 -0.431 1.459

[0114] 2. Calculate the information entropy after dividing the dataset according to a certain business indicator.

[0115] Based on whether the uplink speed is greater than or equal to 20Mbps, the data is divided into two subsets, left and right (by branching at a node of the decision tree).

[0116] The subset of sample data with an uplink rate less than 20 (hereinafter referred to as the "subset") contains 2 samples, accounting for 0.333%. The calculation results of the information entropy of this subset are shown in Table 3.

[0117] Table 3 Information Entropy Table after Sample Dataset Splitting

[0118] Template 1 Template 2 Template 3 Summation Sample size 0 1 1 2 p 0 0.5 0.5 1 log(p,2) 0 -1 -1 -- p*log(p,2) 0.000 -0.5 -0.5 1

[0119] The sample data subset with an uplink rate greater than or equal to 20 Mbps consists of 4 samples, accounting for 0.667%. The calculation results of the information entropy of this subset are shown in Table 4.

[0120] Table 4. Information Entropy Table 2 after Sample Dataset Splitting

[0121] Template 1 Template 2 Template 3 Summation Sample size 2 2 0 4 p 0.5 0.5 0 1 log(p,2) -1 -1 0 -- p*log(p,2) -0.5 -0.5 0.000 1

[0122] Therefore, the conditional entropy of the sample dataset after dividing it according to whether the uplink rate is greater than or equal to 20Mbps is 1, and the information gain is 0.459.

[0123] 3. Repeat the above process for the sample subset from step 2, selecting the business indicators and values ​​for the partitioning. The following example, using a sample data subset with an uplink speed greater than or equal to 20Mbps, illustrates the execution process of step 3.

[0124] The sample data subset with a downlink rate of less than 80 Mbps consists of 3 samples. The calculation results of the information entropy of this subset are shown in Table 5.

[0125] Table 5 Information Entropy after Sample Dataset Splitting (Table 3)

[0126] Template 1 Template 2 Template 3 Summation Sample size 2 1 0 3 p 0.6667 0.3333 0 1 log(p,2) -0.585 -1.585 0 -- p*log(p,2) -0.39 -0.528 0.000 0.918

[0127] The sample data subset with a downlink rate greater than or equal to 80 Mbps consists of 3 samples. The calculation results of the information entropy of this subset are shown in Table 6.

[0128] Table 6 Information Entropy Table 4 after Sample Dataset Splitting

[0129] Template 1 Template 2 Template 3 Summation Sample size 0 1 0 1 p 0 1 0 1 log(p,2) 0 0 0 -- p*log(p,2) 0.000 0.000 0.000 0

[0130] Therefore, the conditional entropy of the sample dataset after dividing it according to whether the downlink rate is greater than or equal to 80Mbps is 0.6887, and the information gain is 0.3113.

[0131] 4. The subset obtained by dividing the sample dataset according to the uplink rate >= 20Mbps and the downlink rate >= 80Mbps has only 1 sample. The decision tree stops splitting and uses template 2 as the output. The subset obtained by dividing the sample dataset according to the uplink rate >= 20Mbps and the downlink rate < 80Mbps has the largest number of samples, so template 1 is used as the output.

[0132] 5. Training complete, the classification model includes rules:

[0133] IF(uplink speed >= 20Mbps and downlink speed >= 80Mbps) THEN template 2;

[0134] IF(uplink speed >= 20Mbps and downlink speed < 80Mbps) THEN Template 1.

[0135] 6. When new requirements emerge, the requirements are judged according to a series of rules of the classification model, and can be assigned to the corresponding template.

[0136] Referring to the example above, calculate the information gain of all business indicator partition values, and select the business indicator machine partition value with the largest information gain to partition the sample dataset; repeat the above process until the number of samples in the sample data subset is less than the first threshold or the information gain is less than the second threshold, or the depth of the tree (the depth of the tree increases by one layer with each partition) reaches the specified threshold.

[0137] Once the model training is complete, the slice parameter template with the most occurrences in the leaf node dataset is the target slice parameter template. The trained model contains a series of discrimination rules and corresponding slice parameter templates, such as "if the downlink bandwidth is greater than 20Mbps and the uplink bandwidth is less than 80Mbps, then match template 2".

[0138] When a new business requirement arises (including several business indicators and their value requirements), it is input into the classification model for prediction. Starting from the root node, the required value of the business indicator is compared with the split value of the current node in the decision tree model to determine which branch to enter, until a leaf node is reached. The template corresponding to the leaf node is used as the output template for the new business requirement. The new business requirement is the target requirement information, and the output template corresponding to the new business requirement is the target slice parameter template.

[0139] This invention categorizes sample data by selecting the attribute with the greatest information gain and constructs a decision tree. This simplifies calculations, improves classification accuracy, and enables the determination of the target slice parameter template that best matches the target requirement information based on the classification model. This allows the base station to create network slice instances that better meet the target requirement information based on the target slice parameter template, resulting in higher accuracy of network slicing.

[0140] Based on the content of any of the above embodiments, before training according to the sample requirement information and the slice parameter template corresponding to the sample requirement information to obtain the classification model, the method further includes: obtaining the slice parameter templates deployed by the base stations of each manufacturer to obtain a slice parameter template library; the slice parameter template corresponding to the sample requirement information is a slice parameter template in the slice parameter template library.

[0141] Specifically, the slice parameter template corresponding to the sample requirement information is a slice parameter template in the slice parameter template library, and each slice parameter template in the slice parameter template library has corresponding sample requirement information.

[0142] The slice parameter template library includes slice parameter templates that have been deployed in base stations from various manufacturers.

[0143] The slice parameter template library can obtain the slice parameter template for each network slice instance by retrieving the values ​​of each slice parameter from the base stations created by each manufacturer. It can be understood that this slice parameter template is a deployed slice parameter template.

[0144] like Figure 3 As shown, the slice parameter templates already deployed by base stations of various manufacturers can be used as live network cases and added to the slice parameter template library (which can be referred to as the "template library"); the slice parameter templates obtained after performing the deployment, implementation, optimization and template entry steps can also be added to the template library.

[0145] It should be noted that after step 105, the step of deploying, implementing, optimizing, and adding the template to the library can be performed. This step specifically includes: the base station can evaluate the key performance indicators (KPIs) of the created network slice instance; based on the evaluated key performance indicators, self-learning is performed to optimize the slice parameters, resulting in a new slice parameter template and an optimized network slice instance, which are then added to the slice parameter template library (which can be referred to as the "template library").

[0146] It should be noted that after step 105, the base station can evaluate the key performance indicators of the created network slice instance; based on the evaluated key performance indicators, it can perform self-learning, and through self-learning, it can optimize the slice parameters to obtain a new slice parameter template and an optimized network slice instance, so that the optimized network slice instance can better meet the target requirements.

[0147] Optionally, after the slice parameters in the target slice parameter template are deployed and implemented, the performance indicators of the relevant slices can be collected through the northbound interface. Based on the performance indicators, the slice parameters are self-optimized and adjusted to adapt to changes in the wireless environment and achieve optimal performance, thus maximizing the guarantee of slice service requirements.

[0148] It should be noted that, based on the parameter differences between different base station manufacturers, the slice parameter templates already deployed on the base stations produced by the first manufacturer can be mapped to obtain slice parameter templates that can be used on base stations produced by the second manufacturer. This breaks down the internal technical barriers of equipment manufacturers and enables the generation and configuration of slice parameter templates across equipment manufacturers.

[0149] This invention provides a slice parameter template library by acquiring the slice parameter templates deployed by base stations from various manufacturers. As the ToB parameter platform is promoted and a large number of slice services are deployed and applied, the template library continues to learn and grow, and the output of the template library tends to be more accurate, thereby reducing or even eliminating the need for manual intervention and improving network slicing efficiency.

[0150] Based on the content of any of the above embodiments, the target requirement information is obtained based on the slice business work order, specifically including: obtaining the original requirement information based on the slice business work order.

[0151] Specifically, the original requirement information can be obtained by performing text recognition on slice business work orders.

[0152] The original demand information may include uplink and downlink rates, uplink and downlink bandwidth, guaranteed rate, guaranteed bandwidth, eMBB (Enhanced Mobile Broadband), dedicated bandwidth, and latency.

[0153] The original requirement information is processed by word segmentation and text recognition to obtain the target requirement information.

[0154] Specifically, the original demand information is processed through word segmentation, stop word removal, and text recognition, such as generating word segmentation for rate information, bandwidth information, latency information, and eMBB information, to obtain the target demand information.

[0155] For example, the original requirement information may include a rate of 70 Mbps, a bandwidth of 50 Mbps, a latency of 5 ms, reliability, jitter, traffic density, QoS level, uplink reserved resources, downlink reserved resources, slice optimization level, rate guarantee, PRB guarantee, number of connections, mobility range, VPN, transmission isolation, and IP tunnel, etc. After performing word segmentation and text recognition on the original requirement information, the target requirement information may include an uplink rate of 70 Mbps, a downlink rate of 100 Mbps, an uplink bandwidth of 30 Mbps, a downlink bandwidth of 50 Mbps, a latency of 10 ms, and a minimum guaranteed rate of 10 Mbps, etc.

[0156] This application extracts more important requirement information as target requirement information by performing word segmentation and text recognition on the original requirement information in the sliced ​​business work order. This reduces the amount of data while retaining the key requirement information, reduces the dimension of the input vector of the classification model, improves the efficiency of template matching, and improves the efficiency and timeliness of network slicing while ensuring the accuracy of network slicing.

[0157] Based on any of the above embodiments, a target instruction is generated based on the target slice parameter template and the slice service work order, specifically including: obtaining the target information of the target base station based on the slice service work order.

[0158] Specifically, the slice service work order also carries target information of the target base station, and the target information such as the coverage area, TAC and VLAN number of the target base station can be extracted from the slice service work order.

[0159] Based on the target slice parameter template and the target information of the target base station, a target execution script is generated.

[0160] Specifically, the slice parameters and their values ​​from the target slice parameter template, along with the target information of the target base station, can be filled into the corresponding positions in the pre-obtained script template to generate the target execution script.

[0161] Generate target instructions carrying the target execution script.

[0162] Specifically, after generating the target execution script, a target instruction carrying the target execution script can be generated based on the target execution script, so that the target base station can directly execute the target execution script and complete the creation of the network slice instance.

[0163] This invention generates a target execution script based on a target slice parameter template and target information of a target base station, and generates target instructions carrying the target execution script, so that the target base station can directly execute the target execution script to complete the creation of a network slice instance, thereby improving the efficiency and timeliness of network slicing.

[0164] Based on any of the above embodiments, sending the target instruction to the target base station specifically includes: sending the target instruction to the operation and maintenance center of the target base station manufacturer, so that the operation and maintenance center can issue the target instruction to the target base station.

[0165] Specifically, such as Figure 4 As shown, the network slicing device 401 communicates with the slice management function network element 402 and the operation and maintenance center (OMC) 403 respectively to implement the network slicing method of this embodiment of the invention.

[0166] Depending on the manufacturer's implementation, the operation and maintenance center can be divided into the wireless subsystem operation and maintenance center (OMC-R) and the switching subsystem operation and maintenance center (OMC-S). In this embodiment of the invention, the operation and maintenance center 403 can be an OMC-R.

[0167] Operation and Maintenance Center 403 can be the Operation and Maintenance Center (OMC-R) for the Wireless Subsystem.

[0168] The network slicing device 401 can receive slice service work orders sent by the slice management function network element 402. The network slicing device 401 can also send target instructions to the operation and maintenance center 403 of the target base station manufacturer. The operation and maintenance center 403 of the target base station manufacturer then issues the target instructions to the target base station.

[0169] In this embodiment of the invention, the target instructions are forwarded to the target base station through the operation and maintenance center of the target base station manufacturer, which enables the generation and configuration of slice parameter templates across equipment manufacturers.

[0170] To facilitate understanding of the above embodiments of the present invention, the process of the network slicing method is described below through examples.

[0171] like Figure 5 As shown, the network slicing method can include two stages: intelligent identification and production deployment. Requirement 1 to Requirement 5 are five different original requirement information, representing five different business requirements; parameter template 1 to parameter template 5 are the target slice parameter templates corresponding to Requirement 1 to Requirement 5, respectively.

[0172] The network slicing apparatus provided by the present invention is described below. The network slicing apparatus described below can be referred to in correspondence with the network slicing method described above.

[0173] Figure 6 This is a schematic diagram of the network slicing device provided by the present invention. Based on any of the above embodiments, such as… Figure 6 As shown, the network slicing device includes a work order acquisition module 601, a demand acquisition module 602, a template matching module 603, an instruction generation module 604, and a slice execution module 605, wherein:

[0174] The work order acquisition module 601 is used to acquire slice service work orders sent by the slice management function network element;

[0175] The requirement acquisition module 602 is used to acquire target requirement information based on sliced ​​business work orders;

[0176] The template matching module 603 is used to input target requirement information into the classification model and obtain the target slice parameter template output by the classification model.

[0177] The instruction generation module 604 is used to generate target instructions based on the target slice parameter template and the slice business work order;

[0178] The slice execution module 605 is used to send the target instruction to the target base station so that the target base station executes the target instruction and creates a network slice instance;

[0179] The classification model is obtained by training based on the sample requirement information and the slice parameter template corresponding to the sample requirement information.

[0180] Specifically, the work order acquisition module 601, the demand acquisition module 602, the template matching module 603, the instruction generation module 604, and the slice execution module 605 are electrically connected in sequence.

[0181] The work order acquisition module 601 connects to the NSMF network element through the NSMF interface and receives slice service work orders sent by the NSMF network element.

[0182] The requirement acquisition module 602 can extract key information from the slice requirement information carried in the slice business work order by performing text recognition on the slice business work order, and use it as target requirement information.

[0183] The template matching module 603 can match the target requirement information with each slice parameter template in the slice parameter template, and determine the slice parameter template that matches the target requirement information as the target slice parameter template.

[0184] The instruction generation module 604 can generate a target instruction carrying target information of the target base station and the values ​​of each slice parameter included in the target slice parameter template.

[0185] The slice execution module 605 can send the target command to the target base station.

[0186] After receiving the target instruction, the target base station executes the instruction to complete the creation of the network slice instance.

[0187] Optionally, the network slicing device may also include:

[0188] The model training module is used to train the classification model based on the sample requirement information and the corresponding slice parameter template.

[0189] Optionally, the model training module can be specifically used for:

[0190] Obtain the information gain of each attribute in the training sample dataset;

[0191] Select the attribute with the highest information gain;

[0192] Group samples with the same feature value of the attribute with the largest information gain into the same subset of training sample data;

[0193] The information gain of each attribute in each training sample data subset is obtained in turn. The attribute with the largest information gain is selected. Samples with the same feature value of the attribute with the largest information gain are classified into the same training sample data subset. The decision tree is constructed by recursively splitting the subset and used as a classification model.

[0194] The sample dataset is a collection of requirement information for each sample.

[0195] Optionally, the network slicing device may also include:

[0196] The template acquisition module is used to acquire the slice parameter templates that have been deployed in the base stations of various manufacturers, and to obtain the slice parameter template library;

[0197] The slice parameter template corresponding to the sample requirement information is a slice parameter template from the slice parameter template library.

[0198] Optionally, the demand acquisition module 602 may include:

[0199] The first acquisition unit is used to acquire original requirement information based on the sliced ​​business work order;

[0200] The processing unit is used to perform word segmentation and text recognition on the original requirement information to obtain the target requirement information.

[0201] Optionally, the instruction generation module 604 may include:

[0202] The second acquisition unit is used to acquire target information of the target base station based on the slice service work order;

[0203] The script generation unit is used to generate a target execution script based on the target slice parameter template and the target information of the target base station;

[0204] The instruction generation unit is used to generate target instructions that carry the target execution script.

[0205] Optionally, the slice execution module 605 may be specifically used for:

[0206] The target command is sent to the operation and maintenance center of the target base station manufacturer, so that the operation and maintenance center can then issue the target command to the target base station.

[0207] The network slicing apparatus provided in this embodiment of the invention is used to execute the network slicing method described above. Its implementation method is consistent with that of the network slicing method provided by this invention, and it can achieve the same beneficial effects. Therefore, it will not be described again here.

[0208] This network slicing device is used in the network slicing methods of the foregoing embodiments. Therefore, the descriptions and definitions in the network slicing methods of the foregoing embodiments can be used to understand the execution modules in the embodiments of the present invention.

[0209] This invention, through its embodiments, acquires slice service work orders sent by slice management function network elements. Based on these work orders, it obtains target requirement information and, using a pre-trained classification model, acquires the target slice parameter template corresponding to the target requirement information. Based on the target slice parameter template and the slice service work order, it generates target instructions and sends these instructions to the target base station, enabling the base station to execute the instructions and create a network slice instance. By establishing an interface with NSMF (Network Slice Management Function), it seamlessly integrates with NSMF to handle slice requests, automatically acquiring the target slice parameter template corresponding to the target requirement information. This allows the base station to execute the target instructions corresponding to the template to create a network slice instance, achieving automated and intelligent network slicing production and implementation, improving efficiency and response timeliness. Furthermore, it enables operators to manage configurable and controllable slice parameters across the entire IT process, adapting to the evolving needs of future industry applications.

[0210] The electronic device and storage medium provided by the present invention are described below. The electronic device and storage medium described below can be referred to in correspondence with the network slicing method described above.

[0211] Figure 7 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 7 As shown, the electronic device may include a processor 710, a communication interface 720, a memory 730, and a communication bus 740. The processor 710, communication interface 720, and memory 730 communicate with each other via the communication bus 740. The processor 710 can call a computer program in the memory 730 to execute the steps of the network slicing method, such as: obtaining a slice service work order sent by a slice management function network element; obtaining target demand information based on the slice service work order; inputting the target demand information into a classification model to obtain a target slice parameter template output by the classification model; generating a target instruction based on the target slice parameter template and the slice service work order; and sending the target instruction to a target base station so that the target base station executes the target instruction and creates a network slice instance. The classification model is obtained by training based on the demand information of each sample and the slice parameter template corresponding to the sample demand information.

[0212] Furthermore, the logical instructions in the aforementioned memory 730 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0213] On the other hand, the present invention also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions, and when the program instructions are executed by a computer, the computer can execute the steps of the network slicing method provided by the above methods, such as: obtaining a slice service work order sent by a slice management function network element; obtaining target demand information based on the slice service work order; inputting the target demand information into a classification model to obtain a target slice parameter template output by the classification model; generating a target instruction based on the target slice parameter template and the slice service work order; and sending the target instruction to a target base station so that the target base station executes the target instruction and creates a network slice instance; wherein the classification model is obtained by training based on the demand information of each sample and the slice parameter template corresponding to the demand information of the sample.

[0214] On the other hand, embodiments of this application also provide a processor-readable storage medium storing a computer program. This computer program is used to cause the processor to execute the steps of the methods provided in the above embodiments, including, for example: obtaining a slice service work order sent by a slice management function network element; obtaining target requirement information based on the slice service work order; inputting the target requirement information into a classification model to obtain a target slice parameter template output by the classification model; generating a target instruction based on the target slice parameter template and the slice service work order; and sending the target instruction to a target base station so that the target base station executes the target instruction and creates a network slice instance. The classification model is obtained by training based on the requirement information of each sample and the slice parameter template corresponding to the sample requirement information.

[0215] The processor-readable storage medium can be any available medium or data storage device that the processor can access, including but not limited to magnetic memory (e.g., floppy disk, hard disk, magnetic tape, magneto-optical disk (MO)), optical memory (e.g., CD, DVD, BD, HVD), and semiconductor memory (e.g., ROM, EPROM, EEPROM, non-volatile memory (NAND FLASH), solid-state drive (SSD)).

[0216] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0217] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0218] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A network slicing method, characterized in that, include: Obtain the slice service work order sent by the slice management function network element; Based on the sliced ​​service work order, obtain the target requirement information; Input the target requirement information into the classification model and obtain the target slice parameter template output by the classification model; Based on the target slice parameter template and the slice business work order, generate the target instruction; The target instruction is sent to the target base station so that the target base station executes the target instruction and creates a network slice instance; The classification model is obtained by training based on the demand information of each sample and the slice parameter template corresponding to the demand information of the sample. Before inputting the target requirement information into the classification model and obtaining the target slice parameter template output by the classification model, the method further includes: The classification model is obtained by training based on the sample requirement information and the slice parameter template corresponding to the sample requirement information; The step of training the classification model based on the sample requirement information and the corresponding slice parameter template to obtain the sample requirement information specifically includes: Obtain the information gain of each attribute in the training sample dataset; Select the attribute with the highest information gain; Group the sample demand information with the same feature value of the attribute with the largest information gain into the same subset of training sample data; The information gain of each attribute in each training sample data subset is obtained sequentially. The attribute with the largest information gain is selected, and the sample demand information with the same feature value of the attribute with the largest information gain is classified into the same training sample data subset. The decision tree is constructed by recursively splitting the subset, which serves as the classification model. The sample dataset is a collection of the requirement information of each sample.

2. The network slicing method according to claim 1, characterized in that, Before training the classification model based on the sample requirement information and the corresponding slice parameter template, the method further includes: Obtain the slice parameter templates of the base stations deployed by each manufacturer to obtain the slice parameter template library; The slice parameter template corresponding to the sample requirement information is a slice parameter template from the slice parameter template library.

3. The network slicing method according to claim 1, characterized in that, The process of obtaining target requirement information based on the sliced ​​service work order specifically includes: Based on the sliced ​​service work order, obtain the original requirement information; The original requirement information is processed by word segmentation and text recognition to obtain the target requirement information.

4. The network slicing method according to claim 1, characterized in that, The step of generating target instructions based on the target slice parameter template and the slice service work order specifically includes: Based on the slice service work order, obtain the target information of the target base station; Based on the target slice parameter template and the target information of the target base station, a target execution script is generated; Generate the target instruction carrying the target execution script.

5. The network slicing method according to any one of claims 1 to 4, characterized in that, Sending the target instruction to the target base station specifically includes: The target instruction is sent to the operation and maintenance center of the target base station manufacturer, so that the operation and maintenance center can issue the target instruction to the target base station.

6. A network slicing device, characterized in that, include: The work order acquisition module is used to acquire slice service work orders sent by the slice management function network element; The requirement acquisition module is used to acquire target requirement information based on the sliced ​​business work order; The template matching module is used to input the target requirement information into the classification model and obtain the target slice parameter template output by the classification model. The instruction generation module is used to generate target instructions based on the target slice parameter template and the slice business work order; The slice execution module is used to send the target instruction to the target base station so that the target base station executes the target instruction and creates a network slice instance; The classification model is obtained by training based on the demand information of each sample and the slice parameter template corresponding to the demand information of the sample. Before inputting the target requirement information into the classification model and obtaining the target slice parameter template output by the classification model, the device is further configured to: The classification model is obtained by training based on the sample requirement information and the slice parameter template corresponding to the sample requirement information; The device is specifically used for: Obtain the information gain of each attribute in the training sample dataset; Select the attribute with the highest information gain; Group the sample demand information with the same feature value of the attribute with the largest information gain into the same subset of training sample data; The information gain of each attribute in each training sample data subset is obtained sequentially. The attribute with the largest information gain is selected, and the sample demand information with the same feature value of the attribute with the largest information gain is classified into the same training sample data subset. The decision tree is constructed by recursively splitting the subset, which serves as the classification model. The sample dataset is a collection of the requirement information of each sample.

7. An electronic device comprising a processor and a memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the network slicing method according to any one of claims 1 to 5.

8. A processor-readable storage medium, characterized in that, The processor-readable storage medium stores a computer program for causing the processor to perform the steps of the network slicing method according to any one of claims 1 to 5.

Images