CN115955700B

CN115955700B - Method for enhancing continuity of network slice service and computer readable storage medium

Info

Publication number: CN115955700B
Application number: CN202211644913.8A
Authority: CN
Inventors: 谢涵; 王丹
Original assignee: Guangzhou Aipu Road Network Technology Co Ltd
Current assignee: Guangzhou Aipu Road Network Technology Co Ltd
Priority date: 2022-12-16
Filing date: 2022-12-16
Publication date: 2023-07-25
Anticipated expiration: 2042-12-16
Also published as: CN115955700A

Abstract

The application discloses a method and a computer readable storage medium for enhancing network slice service continuity, the method comprising: AMF subscribes network slice load degree analysis to NWDAF; the UE establishes PDU session according to S-NSSAI 1 in S-NSSAI list in the request message; the NWAF collects data of each feature at preset time intervals, wherein each feature is a load-related feature of NSI corresponding to each S-NSSAI in the S-NSSAI list; the NWDAF predicts according to the data of each characteristic, and when the prediction result shows that the load degree of NSI corresponding to S-NSSAI 1 in the future within the preset time period reaches a given threshold value of the load degree, the NWDAF sends a notification message to the AMF; the AMF initiates a request for selecting network slices to NSSF, wherein the request message comprises each S-NSSAI information allowed by UE and a load degree prediction message of NSI corresponding to each S-NSSAI in a preset time length in the future; NSSF selects S-NSSAI2 and responds the selection result to AMF; the S-NSSAI switching operation is performed according to S-NSSAI 1 as a source S-NSSAI and S-NSSAI2 as a target S-NSSAI. The method and the device are beneficial to avoiding abnormal communication caused by overload of the network slice.

Description

Method for enhancing continuity of network slice service and computer readable storage medium

Technical Field

The application relates to the technical field of 5GS, in particular to a method for enhancing network slice service continuity and a computer readable storage medium.

Background

With the advent of the 5G era, an era of everything interconnection is started, three scenes of eMBB, mMTC and uRLLC are supported by 5G, and various diversified and differentiated services and applications are contained in the three scenes, so that in order to adapt to the different types of services and application requirements, the 5G network virtually forms a plurality of end-to-end networks on the basis of general hardware by dividing the resources and functions of an actual network and utilizing a slicing technology to form different network slices, each network has different network functions, and the services and applications of different industries can be borne by using different slicing networks, so that the service and application requirements of different industries are met.

The network slices consist of a RAN part and a 5GC part, each network slice is uniquely identified by an S-nsai, which may be associated with one or more NSIs, which may be associated with one or more S-nsais, which is a specific end-to-end logical network, containing a series of network elements and combinations of resources (computing, storage, etc.), depending on the operator' S operational or deployment requirements. For any S-nsai, the network may serve the UE with only one NSI associated with that S-nsai at any time, since only one NSI associated with a given S-nsai is used to serve the UE, once the NSSF selects an NSI to serve the UE based on that S-nsai, the AMF instance serving the UE may limit discovery of a series of instances such as SMFs that use the same network slice, such that the configuration of PDU sessions is only in the currently selected NSI unless that NSI is no longer valid in the given registration area, or the nsai allowed by the UE changes, etc. However, during actual service of the network, with continuous change of network slice resources, overload may occur in the network slice selected by the UE, so that a part of PDU session is interrupted, communication of the UE is abnormal, and for some critical services, the abnormal interruption may have serious consequences.

Therefore, a more intelligent and optimized manner is needed for the 5G network to avoid the situation of abnormal communication caused by overload of network slices.

Disclosure of Invention

The present application is directed to a method and a computer-readable storage medium for enhancing continuity of network slice services, which are beneficial to avoiding abnormal communication caused by overload of a network slice.

To achieve the above object, the present application provides a method for enhancing continuity of network slice services, including:

AMF subscribes network slice load degree analysis to NWDAF during network service;

the UE establishes PDU session according to S-NSSAI 1 in S-NSSAI list in the request message;

the NWDAF collects data of each characteristic at preset time intervals, each characteristic is a load-related characteristic of NSI corresponding to each S-NSSAI in the S-NSSAI list, and the data of each characteristic is used for analyzing the load degree of the corresponding NSI respectively;

the NWDAF predicts according to the collected historical data of each characteristic, and when a prediction result shows that the load degree of NSI corresponding to S-NSSAI 1 in the preset time length in the future reaches a given threshold value of the load degree, the NWDAF sends a notification message to the AMF, wherein the notification message comprises a load degree prediction message of NSI corresponding to each S-NSSAI in the preset time length in the future;

the AMF initiates a request for selecting network slices to NSSF according to the notification message, wherein the request message comprises each S-NSSAI information allowed by the UE and a load degree prediction message of NSI corresponding to each S-NSSAI in the future preset duration;

the NSSF selects the S-NSSAI2 according to the S-NSSAI information allowed by the UE and the load degree prediction message of the NSI corresponding to each S-NSSAI in the future preset duration, and responds the selection result to the AMF, wherein the S-NSSAI2 and the S-NSSAI 1 are related to the same NSSAI;

the S-NSSAI switching operation is performed according to S-NSSAI 1 as a source S-NSSAI and S-NSSAI2 as a target S-NSSAI.

Optionally, the request message of the AMF subscription network slice load degree analysis includes an analysis ID: load degree prediction message, analysis and filtration information: S-NSSAI and NSI ID, target of analysis report: all IDs, given threshold for load level, predicted future preset duration.

Optionally, the NWDAF collects data of each of the features to a corresponding data source at the preset time interval, where the data sources include AMF, OAM, and SMF.

Optionally, each of the features includes a total number of UEs served by the NSI at the current time, an average number of UEs registered/deregistered on the NSI within the preset time interval, an average number of PDU sessions established/released, and an average use condition of NF instance virtual resources.

Optionally, the selecting, by the NSSF, the S-nsai 2 according to the S-nsai information allowed by the UE and the load degree prediction message of the NSI corresponding to the S-nsai within the future preset duration includes:

NSSF subscribes the number of UE registered and the number of established PDU sessions on each S-NSSAI allowed by the UE to NSACF;

the NSACF responds the number of the UE registered on each S-NSSAI allowed by the UE and the number of the established PDU session to NSSF;

the NSSF selects the S-NSSAI2 according to the S-NSSAI information allowed by the UE, the number of the received UE registered on each S-NSSAI and the number of established PDU sessions, and the load degree prediction message of the NSI corresponding to each S-NSSAI in the future preset time length, and responds the selection result to the AMF.

Optionally, the performing the S-nsai switching operation according to the S-nsai 1 as the source S-nsai and the S-nsai 2 as the target S-nsai includes:

AMF initiates SM context update request to corresponding SMF1, wherein the message of the update request comprises PDU session switching indication, S-NSSAI 1 as source S-NSSAI and S-NSSAI2 as target S-NSSAI;

SMF1 sends an N1N2 message to AMF, wherein the N1N2 message comprises newly selected S-NSSAI2;

the AMF forwards the N1N2 message to the RAN, and the RAN forwards the N1 message to the UE, wherein the N1 message contains newly selected S-NSSAI2;

the UE establishes PDU session to the network based on S-NSSAI2;

the UE releases the PDU session on the NSI associated with S-nsai 1.

Optionally, after performing the S-nsai switching operation according to the S-nsai 1 as the source S-nsai and the S-nsai 2 as the target S-nsai, the method further includes:

the NWDAF continuously predicts the loading degree of NSI corresponding to the S-NSSAI 1;

and performing an operation of switching from the S-NSSAI2 to the S-NSSAI 1 until the returned prediction result indicates that the load degree in the future preset time period does not reach the given threshold value of the load degree.

Optionally, the NWDAF predicting according to the collected historical data of each of the features includes:

the NWDAF establishes a data set according to the collected historical data of each feature, wherein the data set also comprises a tag column formed by tag data, each tag data corresponds to the data of each feature collected at each moment, the tag data takes a first numerical value or a second numerical value, the first numerical value represents that the corresponding NSI does not reach a given threshold value of the load degree, and the second numerical value represents that the corresponding NSI reaches a given threshold value of the load degree;

dividing the data set into a training set and a testing set;

training a prediction model by utilizing the training set;

verifying the prediction model after training by using the test set;

and predicting by using the prediction model after verification to obtain a load degree prediction message.

Alternatively, the process may be carried out in a single-stage,

the prediction model is a weighted random forest model;

the training of the prediction model by using the training set comprises:

sampling k sub-data sets from the training set with a place back, wherein the corresponding data quantity of each sub-data set is q, and q is smaller than the data quantity of the training set;

training a CART decision tree for each of the sub-data sets, respectively:

randomly selecting p features from the features, respectively performing binary division, and respectively performing the following calculation on the p features:

(1) Calculating a base index:

(2) Calculating information gain:

wherein D is a node to be selected, N _L And N _R Left and right nodes of the node D; n is n _i For the number of samples of the i-th type, W _i Class weights representing class i, the class weights being inversely proportional to the number of samples of both classes;

(3) Selecting the characteristic with the maximum information gain as the classification characteristic of the root node;

(4) Determining all nodes under the root node based on the modes of the steps (1) to (3), and finally outputting a first category or a second category by the CART decision tree;

the predicting with the prediction model of completion verification includes:

inputting the newly collected characteristic data into k CART decision trees respectively to obtain the number of output first categories and the number of output second categories;

using the highest class obtained in the k CART decision trees for the final classification result by a voting method:

Result＝argmax(I _i W _i )

wherein W is _i Class weights representing class I, I _i Accumulated votes for category i.

To achieve the above object, the present application further provides a computer-readable storage medium having stored thereon a program which, when executed by a processor, implements a method of enhancing network slice service continuity as described above.

The present application also provides a computer program product or computer program comprising computer instructions stored in a computer readable storage medium. The computer instructions are read from a computer-readable storage medium by a processor of an electronic device, and executed by the processor, cause the electronic device to perform a method of enhancing network slice service continuity as described above.

According to the method, the device and the system, the NWAF can predict by utilizing the collected historical data of the load related characteristics of NSIs corresponding to each S-NSSAI in the S-NSSAI list, when a prediction result shows that the load degree of the NSI corresponding to the S-NSSAI 1 in the future preset time reaches a given threshold value of the load degree, the NWAF sends a notification message to the AMF, the AMF further initiates a request for selecting a network slice to the NSSF according to the notification message, then the NSSF can select S-NSSAI2 according to the information of each S-NSSAI allowed by UE from the AMF and the load degree prediction message of the NSI corresponding to each S-NSSAI in the future preset time and respond the selection result to the AMF, and then S-NSSAI switching operation can be executed, so that the situation of communication abnormality caused by network slice overload can be avoided.

Drawings

Fig. 1 is a flow chart of a method of enhancing network slice service continuity in accordance with an embodiment of the present application.

Fig. 2 is a flow chart of a handoff to a new network slice in an embodiment of the present application.

Fig. 3 is a schematic diagram of a CART decision tree according to an embodiment of the present application.

Fig. 4 is a flow chart of a method of enhancing network slice service continuity in accordance with another embodiment of the present application.

Detailed Description

In order to describe the technical content, constructional features, achieved objects and effects of the present application in detail, the following description is made in connection with the embodiments and the accompanying drawings.

For ease of understanding the present application, the relevant terms presented herein are explained as follows:

ebb: enhanced Mobile Broadband enhanced mobile broadband

mctc: massive Machine Type Communication mass machine type communication

uRLLC: ultra Reliable Low Latency Communication ultra-high reliability low latency communications

UE: user Equipment

RAN: radio Access Network radio access network

5GC:5G Core network

NF: network Function: english abbreviation for core network element in 5G network

AMF: access and Mobility Management Function access and mobility management functions

SMF: session Management Function session management functionality

UPF: user Plane Function user plane functionality

S-NSSAI: single Network Slice Selection Assistance Information single network slice selection assistance information

NSI: network Slice instance network slice example

NSSF: network Slice Selection Function network slice selection function

PDU: protocol Data Unit protocol data unit

DNN: data Network Name data network name

PLMN: public Land Mobile Network public land mobile network

SSC: session and Service Continuity session and service continuity

OAM: operation Administration and Maintenance operation and maintenance management

Example 1

Referring to fig. 1 and 2, the present application discloses a method for enhancing network slice service continuity, which includes:

101. the AMF subscribes to the NWDAF network slice load level analysis during network service. After the AMF subscribes to the network slice load level analysis, the NWDAF may make analysis predictions as described later on the UE-related network slice load level to which the AMF relates.

Specifically, the request message of the AMF subscription network slice load degree analysis includes an analysis ID: load degree prediction message, analysis and filtration information: S-NSSAI and NSI ID, target of analysis report: and all the UEs, a given threshold value of the load degree and predicted future preset duration. For example, the given threshold for the load level may be set to 80%, i.e. the AMF is informed of the analysis result when 80% of the current NSI available resources are reached; the predicted future preset time period may be set to 10 minutes, i.e., the time period for which prediction is requested is 10 minutes. The NWDAF may respond to the AMF after receiving the request, and may include a time interval for periodically reporting data in the response, for example: two minutes, i.e., the AMF reports data to the NWDAF every two minutes.

102. The UE establishes a PDU session according to S-NSSAI 1 in the S-NSSAI list in the request message.

Specifically, the request message for establishing the PDU session further includes: DNN, PDU session ID, PDU session type, SSC mode, PLMN ID, etc.

103. The NWDAF collects data of each feature at a preset time interval (for example, two minutes), where each feature is a load-related feature of an NSI corresponding to each S-nsai in the S-nsai list, and the data of each feature is used to analyze the load degree of the corresponding NSI.

Specifically, the NWDAF collects data of each feature from a corresponding data source at preset time intervals, where the data source includes AMF, OAM, SMF, and the like.

Specifically, each feature includes the total number of UEs served by the NSI at the current time, the average number of UEs registered/deregistered on the NSI in a preset time interval, the average number of PDU sessions established/released, and the average usage of NF instance virtual resources (CPU, memory, disk, etc.). Of course, each feature is not limited to the above feature, and for example, other features may be included.

104. The NWAF predicts according to the collected historical data of each characteristic, and when the prediction result shows that the load degree of the NSI corresponding to the S-NSSAI 1 in the preset time period (such as 10 minutes) in the future reaches a given threshold value of the load degree, the NWAF sends a notification message to the AMF, wherein the notification message comprises a load degree prediction message of the NSI corresponding to each S-NSSAI in the preset time period in the future. In a specific example, the prediction result is 1 or 0, which respectively represents that the load degree reaches a given threshold value of the load degree and that the load degree does not reach a given threshold value of the load degree.

In some implementations, the NWDAF predicting from the collected historical data for each feature includes:

firstly, the NWDAF establishes a data set according to the collected historical data of each feature, the data set further comprises a tag column composed of tag data, each tag data corresponds to the data of each feature collected at each moment, the tag data takes a value of a first value (for example, 0) or a second value (for example, 1), the first value represents that the corresponding NSI does not reach a given threshold of the load degree, and the second value represents that the corresponding NSI reaches a given threshold of the load degree.

Specifically, the historical data of each feature collected by NWDAF is:

where m is the data amount and n is the feature number.

Next, the data set is divided into a training set and a testing set. For example, 70% of the data set is divided into training sets and 30% of the data set is divided into test sets.

Next, the predictive model is trained using the training set.

Next, the test set is used to validate the trained predictive model.

And then, predicting by using a prediction model with verification completed to obtain a load degree prediction message.

Further, the prediction model is a weighted random forest model, and training the prediction model by using the training set includes:

obtaining k sub-data sets from a training set (m observation samples with n characteristics) by sampling back (sampling by using Bootstrap), wherein the corresponding data quantity of each sub-data set is q, and q is smaller than the data quantity m of the training set;

training a CART decision tree for each sub-dataset (let each CART decision tree grow as fully as possible, i.e. each CART decision tree does not need pruning):

randomly selecting p features (p < n) from n features, respectively performing binary division, and respectively performing the following calculation on the p features:

(1) Calculating a base index:

(2) Calculating information gain:

(3) Selecting the characteristic with the maximum information gain as the classification characteristic of the root node; the larger the obtained information gain is, the higher the purity of the representative node is, and the more important the characteristic is, so that when the characteristic is screened by using the base index, the characteristic with the largest information gain is preferentially selected;

(4) Determining all nodes under the root node based on the modes of the steps (1) to (3), namely after determining the root node, selecting the characteristic with the maximum information gain of other nodes under the root node, wherein the modes are known to a person skilled in the art and are not described in detail herein; the CART decision tree ultimately outputs either the first class (0) or the second class (1).

Predicting using the prediction model to complete the verification includes:

inputting the newly collected characteristic data into k CART decision trees respectively to obtain the number of output first categories (0) and the number of output second categories (1);

Result＝argmax(I _i W _i )

Specifically, the class weight of the first class is obtained from the ratio of the number of samples exceeding a given threshold of the degree of loading to the number of samples not exceeding the given threshold of the degree of loading in the total training data m, and the class weight of the second class is obtained from the ratio of the number of samples not exceeding 80% of the slice loading threshold to the number of samples exceeding 80% of the slice loading threshold in the total training data m.

Referring to fig. 3, for example, four features including the total number of UEs served by NSI at the current time, the average number of UEs registered on NSI in a given time interval, the average number of de-registered UEs, and the average usage of disk are selected to construct a CART decision tree.

Taking the characteristic of total number of UE served by NSI at the current moment as an example, firstly, binary division is carried out, the number of samples with the total number of UE exceeding 10000 (which can be obtained by multiple adjustment) is n1, the number of samples with the total number of UE not exceeding 10000 is n2, the class weight W1 corresponding to n1 is the ratio of n2 to n1, and the class weight W2 corresponding to n2 is the ratio of n1 to n 2; the base index of the node to be selected is obtained by summing the square of the product of the number of each category and the corresponding category weight and dividing the sum of the product of the number of each category and the corresponding category weight, and the base index of the left node is obtained by dividing the square of the product of the number of samples with the total number of UE not exceeding 10000 and the corresponding category weight by the product of the number of samples with the total number of UE not exceeding 10000 and the corresponding category weight; similarly, the base index of the right node is obtained by dividing the square of the product of the number of samples with the total number of UEs exceeding 10000 and the corresponding class weight by the product of the number of samples with the total number of UEs exceeding 10000 and the corresponding class weight.

And then, subtracting the base index of the left node from the base index of the node to be selected, and subtracting the base index of the right node from the base index of the node to be selected to obtain the corresponding information gain. The three features of the average number of UEs registered on NSI, the average number of de-registered UEs and the average usage of disk in a given time interval are also used to obtain the corresponding information gain in the above manner. Among the four features, the information gain obtained by the feature of the total number of UEs served by NSI at the current time is the largest, so that the feature is taken as the classification feature of the root node.

Next, other nodes under the root node are determined in the same manner (the feature of the maximum information gain is selected as the current node). Finally, the CART decision tree shown in figure 3 is obtained. After verifying the CART decision tree, the CART decision tree can be used for prediction.

It is to be understood that the present application is not limited to the prediction method in the specific example described above when predicting based on the history data of each feature.

105. The AMF initiates a request for selecting network slices to NSSF according to the notification message, wherein the request message comprises each S-NSSAI information allowed by the UE and a load degree prediction message of NSI corresponding to each S-NSSAI in a preset time length in the future. Specifically, after the AMF receives the NWDAF notification message, the result indicates that the resource utilization of the current slice instance is too high, and an overload condition may occur, so that a network slice selection request is initiated to the NSSF.

106. The NSSF selects the S-NSSAI2 according to the S-NSSAI information allowed by the UE and the load degree prediction message of the NSI corresponding to each S-NSSAI in the future preset time period, and responds the selection result to the AMF, wherein the S-NSSAI2 and the S-NSSAI 1 are related to the same NSSAI.

In some embodiments, the selecting the S-nsai 2 by the NSSF according to the S-nsai information allowed by the UE and the load level prediction message of the NSI corresponding to the S-nsai within the future preset time period includes:

Specifically, referring to fig. 2, when selecting a new S-nsai, the NSSF first determines whether the S-nsai to be screened and the S-nsai 1 are related to the same nsai, if yes, then the second step of screening is performed, that is, whether the maximum registration number of the UE is 80% on the S-nsai is determined, if no, the NSSF starts to screen the next S-nsai. And if the result of the second step of screening is negative, entering a third step of screening, namely judging whether 80% of the maximum number of PDU sessions is reached on the S-NSSAI, otherwise, starting to screen the next S-NSSAI by NSSF. And if the result of the third step of screening is negative, the screening is successful, the selection result is returned to the AMF, and otherwise, NSSF starts to screen the next S-NSSAI.

107. The S-NSSAI switching operation is performed according to S-NSSAI 1 as a source S-NSSAI and S-NSSAI2 as a target S-NSSAI.

Specifically, performing the S-NSSAI switching operation based on S-NSSAI 1 as a source S-NSSAI and S-NSSAI2 as a target S-NSSAI includes:

the AMF forwards the N1N2 message to the RAN, the RAN forwards the N1 message to the UE, and the N1 message contains the newly selected S-NSSAI2;

the UE establishes PDU session to the network based on S-NSSAI2;

the UE releases the PDU session on the NSI associated with S-nsai 1.

Specifically, after performing the S-nsai switching operation according to the S-nsai 1 as the source S-nsai and the S-nsai 2 as the target S-nsai, the method further includes:

and performing the operation of switching from S-NSSAI2 to S-NSSAI 1 until the returned prediction result shows that the load degree in the future preset time period does not reach the given threshold value of the load degree, so as to avoid influencing the service on other slices. The operation of switching back to S-NSSAI 1 is similar to that of switching back to S-NSSAI2 and will not be described again here.

Example two

Referring to fig. 4, the present application discloses a method for enhancing network slice service continuity, which includes:

1. the AMF subscribes to the NWDAF network slice load level analysis during network service.

2. The UE establishes a PDU session according to S-NSSAI 1 in the S-NSSAI list in the request message.

3. The NWDAF collects data of each feature at preset time intervals, each feature is a load-related feature of NSI corresponding to each S-NSSAI in the S-NSSAI list, and the data of each feature is used for analyzing the load degree of the corresponding NSI respectively.

4. The NWAF predicts according to the collected historical data of each feature, and when the prediction result shows that the load degree of the NSI corresponding to the S-NSSAI 1 in the preset time period in the future reaches a given threshold value of the load degree, the NWAF sends a notification message to the AMF, wherein the notification message comprises the load degree prediction message of the NSI corresponding to each S-NSSAI in the preset time period in the future.

5. The AMF initiates a request for selecting network slices to NSSF according to the notification message, wherein the request message comprises each S-NSSAI information allowed by the UE and a load degree prediction message of NSI corresponding to each S-NSSAI in a preset time length in the future;

6. NSSF subscribes the number of UE registered and the number of established PDU sessions on each S-NSSAI allowed by the UE to NSACF;

7. the NSACF responds the number of the UE registered on each S-NSSAI allowed by the UE and the number of the established PDU session to NSSF;

8. the NSSF selects the S-NSSAI2 according to the S-NSSAI information allowed by the UE, the number of the received UE registered on each S-NSSAI and the number of established PDU sessions and the load degree prediction message of the NSI corresponding to each S-NSSAI in the future preset time period, and responds the selection result to the AMF, wherein the S-NSSAI2 and the S-NSSAI 1 are related to the same NSSAI.

9. AMF initiates SM context update request to corresponding SMF1, wherein the message of the update request comprises PDU session switching indication, S-NSSAI 1 as source S-NSSAI and S-NSSAI2 as target S-NSSAI;

10. SMF1 sends an N1N2 message to AMF, wherein the N1N2 message comprises newly selected S-NSSAI2;

11. the AMF forwards the N1N2 message to the RAN, the RAN forwards the N1 message to the UE, and the N1 message contains the newly selected S-NSSAI2;

12. the UE establishes PDU session to the network based on S-NSSAI2;

13. the UE releases the PDU session on the NSI associated with S-nsai 1.

Example III

The application discloses a computer readable storage medium having a program stored thereon, which when executed by a processor implements a method for enhancing network slice service continuity as in embodiment one or embodiment two.

Example IV

Embodiments of the present application disclose a computer program product or computer program comprising computer instructions stored in a computer readable storage medium. The processor of the electronic device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the electronic device performs the method of enhancing the network slice service continuity as in the first or second embodiment.

It should be appreciated that in embodiments of the present application, the processor may be a central processing module (CentralProcessing Unit, CPU), which may also be other general purpose processors, digital signal processors (DigitalSignal Processor, DSP), application specific integrated circuits (Application SpecificIntegrated Circuit, ASIC), off-the-shelf programmable gate arrays (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or the like. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.

Those skilled in the art will appreciate that the processes implementing all or part of the methods of the above embodiments may be implemented by hardware associated with computer program instructions, and the program may be stored in a computer readable storage medium, where the program when executed may include processes of embodiments of the methods described above. The storage medium may be a magnetic disk, an optical disk, a Read-only memory (ROM), a Random access memory (Random AccessMemory, RAM), or the like.

In the foregoing embodiments, the descriptions of the embodiments are emphasized, and for parts of one embodiment that are not described in detail, reference may be made to the related descriptions of other embodiments.

The foregoing disclosure is only illustrative of the preferred embodiments of the present application and is not intended to limit the scope of the claims hereof, as defined by the equivalents of the claims.

Claims

1. A method of enhancing network slice service continuity, comprising:

the access and mobility management function network element subscribes network slice load degree analysis to the network data analysis function network element during network service;

the terminal selects auxiliary information according to the single network slice in the request message to establish PDU session;

the network data analysis function network element collects data of each feature at preset time intervals, each feature is a load related feature of a network slice instance corresponding to each single network slice selection auxiliary information in the single network slice selection auxiliary information list, and the data of each feature is used for analyzing the load degree of the corresponding network slice instance;

the network data analysis function network element predicts according to the collected historical data of each feature, and when the prediction result shows that the load degree of the network slice instance corresponding to the first single network slice selection auxiliary information in the future preset time length reaches a given threshold value of the load degree, the network data analysis function network element sends a notification message to the access and mobility management function network element, wherein the notification message comprises a load degree prediction message of the network slice instance corresponding to each single network slice selection auxiliary information in the future preset time length;

the access and mobility management function network element initiates a request for selecting the network slice to the network slice selection function network element according to the notification message, wherein the request message comprises single network slice selection auxiliary information allowed by the terminal and a load degree prediction message of a network slice instance corresponding to the single network slice selection auxiliary information in the future preset duration;

the network slice selection function network element selects second single network slice selection auxiliary information according to the single network slice selection auxiliary information allowed by the terminal and the load degree prediction information of the network slice instance corresponding to the single network slice selection auxiliary information in the future preset time period, and responds the selection result to the access and mobility management function network element, wherein the second single network slice selection auxiliary information and the first single network slice selection auxiliary information are related to the same network slice selection auxiliary information;

the single network slice selection assistance information switching operation is performed based on the first single network slice selection assistance information as the source single network slice selection assistance information and the second single network slice selection assistance information as the target single network slice selection assistance information.

2. The method of enhancing network slice service continuity according to claim 1, wherein,

the request message for analyzing the network slice load degree of the access and mobility management function network element subscription comprises an analysis ID: load degree prediction message, analysis and filtration information: single network slice selection assistance information and network slice instance ID, target of analysis report: all IDs, given threshold for load level, predicted future preset duration.

3. The method of enhancing network slice service continuity according to claim 1, wherein a network data analysis function network element collects data of each of said features at said preset time intervals to corresponding data sources including an access and mobility management function network element, an operation maintenance management network element and a session management function network element.

4. The method of enhancing network slice service continuity according to claim 1, wherein,

the characteristics comprise the total number of the terminals served by the network slice instance at the current moment, the average number of the terminals registered/logged off on the network slice instance in the preset time interval, the average number of established/released PDU (protocol data unit) sessions and the average use condition of NF instance virtual resources.

5. The method of enhancing network slice service continuity according to claim 1, wherein,

the network element selecting the second single network slice selection auxiliary information according to the single network slice selection auxiliary information allowed by the terminal and the load degree prediction message of the network slice instance corresponding to the single network slice selection auxiliary information in the future preset duration comprises the following steps:

the network element of the network slice selection function subscribes the quantity of terminals registered on the auxiliary information and the quantity of established PDU session to each single network slice allowed by the terminal of the network element of the network slice admission control function;

the network slicing admission control function network element responds the number of the terminals registered on the single network slice selection auxiliary information allowed by the terminals and the number of established PDU session to the network slice selection function network element;

the network slice selection function network element selects second single network slice selection auxiliary information according to the single network slice selection auxiliary information allowed by the terminal, the number of the terminals registered on the received single network slice selection auxiliary information and the number of established PDU sessions, and the load degree prediction information of the network slice instance corresponding to the single network slice selection auxiliary information in the future preset time length, and responds the selection result to the access and mobility management function network element.

6. The method of enhancing network slice service continuity according to claim 1, wherein,

the performing a single network slice selection assistance information switching operation according to the first single network slice selection assistance information as the source single network slice selection assistance information and the second single network slice selection assistance information as the target single network slice selection assistance information includes:

the access and mobility management function network element initiates an SM context update request to a corresponding first session management function network element, wherein the message of the update request comprises PDU session switching indication, first single network slice selection auxiliary information serving as source single network slice selection auxiliary information and second single network slice selection auxiliary information serving as target single network slice selection auxiliary information;

the first session management function network element sends an N1N2 message to the access and mobility management function network element, wherein the N1N2 message comprises newly selected second single network slice selection auxiliary information;

the access and mobility management function network element forwards the N1N2 message to a wireless access network, the wireless access network forwards the N1 message to a terminal, and the N1 message contains newly selected second single network slice selection auxiliary information;

the terminal selects auxiliary information based on the second single network slice to establish PDU session to the network;

the terminal releases the PDU session on the network slice instance associated with the first single network slice selection assistance information.

7. The method of enhancing network slice service continuity according to claim 1, wherein,

the method further includes, after performing the single network slice selection assistance information switching operation based on the first single network slice selection assistance information as the source single network slice selection assistance information and the second single network slice selection assistance information as the target single network slice selection assistance information:

the network data analysis function network element continuously predicts the load degree of the network slice instance corresponding to the first single network slice selection auxiliary information;

and performing the operation of switching from the second single network slice selection auxiliary information to the first single network slice selection auxiliary information until the returned prediction result indicates that the load degree in the future preset time period does not reach the given threshold value of the load degree.

8. The method of enhancing network slice service continuity according to claim 1, wherein,

the network data analysis function network element predicts according to the collected historical data of each feature, and comprises:

the network data analysis function network element establishes a data set according to the collected historical data of each feature, the data set further comprises a tag column formed by tag data, each tag data corresponds to the data of each feature collected at each moment, the tag data takes a first value or a second value, the first value represents that the corresponding network slice instance does not reach a given threshold value of the load degree, and the second value represents that the corresponding network slice instance reaches a given threshold value of the load degree;

dividing the data set into a training set and a testing set;

training a prediction model by utilizing the training set;

verifying the prediction model after training by using the test set;

9. The method of enhancing network slice service continuity according to claim 8, wherein,

the prediction model is a weighted random forest model;

the training of the prediction model by using the training set comprises:

training a CART decision tree for each of the sub-data sets, respectively:

(1) Calculating a base index:

(2) Calculating information gain:

Gain(D)＝Gini(D)-Gini _NL (D)-Gini _NR (D)

the predicting with the prediction model of completion verification includes:

Result＝argmax(I _i W _i )

10. A computer readable storage medium having stored thereon a program, wherein the program when executed by a processor implements the method of enhancing network slice service continuity according to claim 9.