Service prediction method and device

Abstract

The invention provides a service prediction method, and relates to the technical field of wireless communication, and the method comprises the steps: a statistics module carries out the statistics of the number of concurrent sessions with the 5QI being i in the yth period under an edge cloud x; the concurrent session number candidate set calculation module calculates a prediction candidate set of the concurrent session number in the (y + 1) th period according to the concurrent session number counted in the recent N + 1 periods,..., m is a value of [y-N, y-1]; and the concurrent session quantity prediction module calculates the predicted concurrent session quantity of which the 5QI is i in the (y + 1) th period under the edge cloud x according to the predicted deviation value. According to the method, an interpolation coefficient set is established by taking a counted historical concurrent session number as prior information, a concurrent session number candidate set is generated based on the interpolation coefficient set, the candidate set is corrected according to a historical deviation empirical value, and a concurrent session number prediction result of each 5QI under each edge cloud is formed. According to the method, the historical concurrent session number and the deviation value prior information are effectively combined, telephone traffic prediction is realized, the prediction accuracy is greatly improved, and a key support is provided for cloud dynamic capacity expansion and contraction deployment of an access network function.

Description

Technical Field

[0001] This application relates to the field of wireless communication technology, and in particular to a service forecasting method and apparatus. Background Technology

[0002] With the rapid development of mobile communications, communication services have permeated all walks of life, becoming a key foundation for social development. In the foreseeable future, the development of all sectors of society will generate diverse demands for communications, forming uncertain new businesses and scenarios. To adapt to these diverse business needs, it is necessary to focus on improving the network's adaptability across all scenarios, flexibly combining necessary network functions as needed, and providing customized network service capabilities.

[0003] The end-to-end service-oriented architecture of cloud-native technologies is a necessary technical means to build the network's adaptability to all scenarios. Based on the service-oriented architecture, it enables flexible and dynamic scaling of network functions, maximizes the network's adaptability, and supports the diverse application needs of the network.

[0004] However, the dynamic scaling of access network protocol function processing in a service-oriented architecture requires accurate service prediction as support. Only by accurately predicting service trends can the access network protocol function processing capabilities be dynamically scaled up or down according to service changes. Otherwise, it cannot be guaranteed that the dynamic scaling deployment results will meet service requirements. Existing technologies have limited discussion on service prediction, mostly defining access network function service capabilities through static configuration during power-on initialization, lacking dynamic adaptability. Therefore, proposing an effective service prediction method is an unresolved issue in the industry and a key support for the cloud-based dynamic scaling deployment of access network functions. Summary of the Invention

[0005] This application aims to at least partially address one of the technical problems in the related art.

[0006] Therefore, the first objective of this application is to propose a service forecasting method to improve the accuracy of dynamic service forecasting and provide key support for the dynamic scaling up and down deployment of access network functions in the cloud.

[0007] The second objective of this application is to propose a business forecasting device.

[0008] To achieve the above objectives, a first aspect of this application proposes a business forecasting method, comprising:

[0009] Step 1: The statistics module counts the number of concurrent sessions with 5QI i in the y-th period under edge cloud x. ;

[0010] Step 2: The concurrent session candidate set calculation module calculates the number of concurrent sessions based on the statistics of the most recent N+1 periods. , … Calculate the prediction candidate set for the number of concurrent sessions in the (y+1)th cycle. Where m takes the value [yN, y-1];

[0011] Step 3: The concurrent session count prediction module predicts the number of concurrent sessions based on... And the prediction bias value, calculate the number of predicted concurrent sessions with 5QI of i in the (y+1)th period under edge cloud x. .

[0012] In step 1, the number of concurrent sessions with 5QI i in the y-th period under the statistical edge cloud x is... , refers to the maximum number of concurrent terminal PDU sessions with 5QI value i residing in the connected state under the cell site managed by edge cloud x during the corresponding time period of the y-th period;

[0013] In step 2, the prediction candidate set for calculating the number of concurrent sessions in the (y+1)th cycle is... The methods include:

[0014] Step 2.1: The concurrent session count candidate set calculation module calculates the number of concurrent sessions based on the service with 5QI i under the edge cloud x over the most recent N+1 periods. , … Construct N pairs of data {(m, ), (m+1, )}, where m takes the value [yN, y-1], and the interpolation coefficient set { is calculated according to equations (1) and (2). , }:

[0015] (1)

[0016] (2)

[0017] Step 2.2: The concurrent session count candidate set calculation module is based on the interpolation coefficient set { , }, according to equation (3), the candidate value of the number of concurrent sessions for the service with 5QI i under edge cloud x in the (y+1)th period is calculated. ;

[0018] (3)

[0019] In step 3, the predicted number of concurrent sessions with 5QI of i is calculated under the (y+1)th period of edge cloud x. The methods include:

[0020] Step 3.1a: The concurrent session number prediction module calculates the preset correction coefficient based on equations (4) and (5). , ;

[0021] (4)

[0022] (5)

[0023] Step 3.2a: Concurrent session count prediction module determination Is it greater than or equal to 0? If so, then... Assign to If not, then put Assign to ;

[0024] Step 3.3a: The concurrent session prediction module calculates the predicted concurrent session count for the service with 5QI i under edge cloud x in the (y+1)th period based on equation (6). :

[0025] (6)

[0026] or,

[0027] Step 3.1b: The concurrent session count prediction module calculates the correction coefficient based on equation (7). ;

[0028] (7)

[0029] Step 3.2b: The concurrent session count prediction module calculates the predicted concurrent session count for the service with 5QI i under edge cloud x in the (y+1)th period based on equation (8). :

[0030] (8)

[0031] To achieve the above objectives, a second aspect of this application provides a business forecasting apparatus, comprising:

[0032] The statistics module is used to count the number of concurrent sessions with 5QI i in the y-th period under edge cloud x. ;

[0033] The concurrent session count candidate set calculation module is used to calculate the number of concurrent sessions based on the statistics of the most recent N+1 periods. , … Calculate the prediction candidate set for the number of concurrent sessions in the (y+1)th cycle. Where m takes the value [yN, y-1];

[0034] The concurrent session count prediction module is used to predict the number of concurrent sessions based on... And the prediction bias value, calculate the number of predicted concurrent sessions with 5QI of i in the (y+1)th period under edge cloud x. .

[0035] This invention addresses the shortcomings of existing technologies where access network function service capabilities are defined through static configuration during power-on initialization. This approach fails to predict services, lacks dynamic adaptability, and struggles to support the dynamic scaling up and down deployment of access network functions in cloud-based scenarios that adapt to diverse future service changes. The invention proposes a service prediction method to improve the accuracy of service prediction and provide crucial support for the dynamic scaling up and down deployment of access network functions in cloud-based systems. Attached Figure Description

[0036] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0037] Figure 1 A flowchart of a business forecasting method provided in this application;

[0038] Figure 2 A schematic diagram of a business forecasting device provided in this application;

[0039] Figure 3 This is a schematic diagram of a concurrent session for an embodiment of a business forecasting method provided in this application. Detailed Implementation

[0040] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0041] The end-to-end service-oriented architecture of cloud-native technologies is a necessary technical means to build network adaptability across all scenarios. Based on the service-oriented architecture, it enables flexible and dynamic scaling of network functions, maximizing network adaptability and supporting diverse application needs. However, the dynamic scaling of access network protocol function processing in the service-oriented architecture requires accurate business prediction as support. Only by accurately predicting business trends can the access network protocol function processing capabilities be dynamically scaled up or down according to business changes; otherwise, it cannot be guaranteed whether the dynamic scaling deployment results meet business requirements. This application uses the statistically recorded number of historical concurrent sessions as prior information to establish an interpolation coefficient set, generates a candidate set of concurrent session numbers based on the interpolation coefficient set, and corrects the candidate set according to historical deviation empirical values ​​to form the concurrent session number prediction results for each edge cloud at 5QI. This invention effectively combines the prior information of historical concurrent session numbers and deviation values ​​to achieve traffic prediction, greatly improving prediction accuracy and providing key support for the cloud-based dynamic scaling deployment of access network functions.

[0042] A business forecasting method and apparatus according to an embodiment of this application are described below with reference to the accompanying drawings.

[0043] Figure 1 This is a flowchart illustrating a business forecasting method provided in an embodiment of this application. Figure 2 This is a schematic diagram of a business forecasting device provided in an embodiment of this application. Figure 3 This is a schematic diagram of a concurrent session for an embodiment of a business forecasting method provided in this application.

[0044] like Figure 1 As shown, this business forecasting method includes the following steps:

[0045] Step 1: The statistics module counts the number of concurrent sessions with 5QI i in the y-th period under edge cloud x. ;

[0046] Step 2: The concurrent session candidate set calculation module calculates the number of concurrent sessions based on the statistics of the most recent N+1 periods. , … Calculate the prediction candidate set for the number of concurrent sessions in the (y+1)th cycle. Where m takes the value [yN, y-1];

[0047] Step 3: The concurrent session count prediction module predicts the number of concurrent sessions based on... And the prediction bias value, calculate the number of predicted concurrent sessions with 5QI of i in the (y+1)th period under edge cloud x. .

[0048] This application uses the historical concurrent session count as prior information to establish an interpolation coefficient set, generates a candidate set of concurrent session counts based on the interpolation coefficient set, and corrects the candidate set according to historical deviation empirical values ​​to form the concurrent session count prediction results for each edge cloud's 5QI. This invention effectively combines the prior information of historical concurrent session counts and deviation values ​​to achieve traffic prediction, greatly improves prediction accuracy, and provides key support for the dynamic expansion and contraction deployment of access network functions in the cloud.

[0049] Furthermore, in this embodiment of the application, in step 1, the number of concurrent sessions with 5QI i in the y-th period under the statistical edge cloud x is... , refers to the maximum number of concurrent terminal PDU sessions with 5QI value i residing in the connected state under the cell site managed by edge cloud x during the corresponding time period of the y-th period;

[0050] In this embodiment, it is assumed that x takes the value of 1 and y takes the value of 10. Assume that the 10th period under edge cloud 1 corresponds to... Figure 3 During the time period [t0, t1], the business situation of Edge Cloud 1 in the 10th cycle is as follows: Figure 3 As shown, from Figure 3 As can be seen, during this time period, 12 terminals, including Terminal 1, Terminal 2, Terminal 3, Terminal 4, Terminal 5, Terminal 6, Terminal 7, Terminal 8, Terminal 9, Terminal 10, Terminal 11, and Terminal 12, conducted business. The session details for each terminal are as follows:

[0051] Terminal 1's PDU session 5QI is 5, and the duration of its connected state is the period [t0, t2].

[0052] Terminal 2's PDU session 5QI is 5, and its connection state duration is the period [t0, t2].

[0053] The PDU session 5QI of terminal 3 is 5, and the duration of its connected state is the time period [t0, t2].

[0054] The PDU session 5QI occurring on terminal 4 is 5, and the duration of its connected state is the period [t0, t2].

[0055] The PDU session 5QI of terminal 5 is 5, and the duration of its connected state is the period [t0, t2].

[0056] The PDU session 5QI of terminal 6 is 5, and the duration of its connected state is the period [t0, t2].

[0057] The PDU session 5QI of terminal 7 is 5, and the duration of its connected state is the period [t0, t2].

[0058] The PDU session 5QI of terminal 8 is 5, and the duration of its connected state is the period [t0, t2].

[0059] The PDU session 5QI of terminal 9 is 5, and the duration of its connected state is the time period [t0, t2].

[0060] The PDU session 5QI of terminal 10 is 5, and the duration of its connected state is the period [t1, t5].

[0061] The PDU session 5QI of terminal 11 is 5, and the duration of its connected state is the time period [t3, t6].

[0062] The PDU session 5QI of terminal 12 is 6, and the duration of its connected state is the period [t4, t7].

[0063] from Figure 3 As can be seen, the maximum concurrent number of PDU sessions with 5QI is 10, corresponding to the concurrent sessions of terminals 1, 2, 3, 4, 5, 6, 7, 8, 9, and 10 occurring in the time period [t1, t2]. The maximum concurrent number of PDU sessions with 5QI is 1, corresponding to terminal 12 occurring in the time period [t4, t7].

[0064] Therefore, in this embodiment, according to step 1, the number of concurrent sessions with 5QI of i in the y-th period under the statistical edge cloud x is... ,by Figure 3 For example, the statistical results are as follows:

[0065] Number of concurrent sessions with a 5QI of 5 in the 10th cycle of Edge Cloud 1 Equals 10,

[0066] Number of concurrent sessions with a 5QI of 6 in the 10th cycle of Edge Cloud 1 It equals 1.

[0067] Furthermore, in the embodiments of this application, in step 2, the prediction candidate set for calculating the number of concurrent sessions in the (y+1)th cycle is... The methods include:

[0068] Step 2.1: The concurrent session count candidate set calculation module calculates the number of concurrent sessions based on the service with 5QI i under the edge cloud x over the most recent N+1 periods. , … Construct N pairs of data {(m, ), (m+1, )}, where m takes the value [yN, y-1], and the interpolation coefficient set { is calculated according to equations (1) and (2). , }:

[0069] (1)

[0070] (2)

[0071] Step 2.2: The concurrent session count candidate set calculation module is based on the interpolation coefficient set { , }, according to equation (3), the candidate value of the number of concurrent sessions for the service with 5QI i under edge cloud x in the (y+1)th period is calculated. ;

[0072] (3)

[0073] In this embodiment, N is assumed to be 2. Based on the assumption in step 1 that x is 1 and y is 10, the value of m is [yN, y-1] = [10-2, 10-1], meaning m can be 8 or 9.

[0074] Then, under Edge Cloud 1, 5QI equals 5, and the actual concurrent session counts in periods y, y-1, and y-2 are respectively... = , = , = ,

[0075] assumed =10, =13, =15,

[0076] Then, according to step 2.1, the following formulas (1) and (2) are used to calculate:

[0077]

[0078]

[0079]

[0080]

[0081] Next, according to step 2.2, the following formula (3) is used to calculate:

[0082]

[0083]

[0084] Furthermore, in this embodiment of the application, in step 3, the number of predicted concurrent sessions with 5QI of i in the (y+1)th period under edge cloud x is calculated. The methods include:

[0085] Step 3.1a: The concurrent session number prediction module calculates the preset correction coefficient based on equations (4) and (5). , ;

[0086] (4)

[0087] (5)

[0088] Step 3.2a: Concurrent session count prediction module determination Is it greater than or equal to 0? If so, then... Assign to If not, then put Assign to ;

[0089] Step 3.3a: The concurrent session prediction module calculates the predicted concurrent session count for the service with 5QI i under edge cloud x in the (y+1)th period based on equation (6). :

[0090] (6)

[0091] or,

[0092] Step 3.1b: The concurrent session count prediction module calculates the correction coefficient based on equation (7). ;

[0093] (7)

[0094] Step 3.2b: The concurrent session count prediction module calculates the predicted concurrent session count for the service with 5QI i under edge cloud x in the (y+1)th period based on equation (8). :

[0095] (8)

[0096] Based on the implementation process of steps 2.1 and 2.2, it can be concluded that: =10, =13, =15, N takes the value of 2, and from the implementation process of step 1, we know that x takes the value of 1 and y takes the value of 10. In this embodiment, if =8, =10, =16.

[0097] This embodiment will first be described according to steps 3.1a to 3.3a:

[0098] According to step 3.1a, yN <= m <= y, so the value of m can be 8, 9, or 10.

[0099] When m=8

[0100]

[0101] When m=9

[0102]

[0103] When m=10

[0104]

[0105] According to equation (4), =3 / 2, according to equation (5), =1 / 2;

[0106] Next, according to step 3.2a, the concurrent session count prediction module determines... In this embodiment, whether it is greater than or equal to 0, =10-8=2, which is greater than or equal to 0. Therefore... Assign to ,now, The value is equal to 3 / 2;

[0107] Next, according to step 3.3a, the concurrent session count prediction module calculates the predicted concurrent session count for the 11th cycle of the service with 5QI i under edge cloud x based on equation (6). First calculate :

[0108]

[0109] Recalculate According to the implementation process in step 2.2,

[0110]

[0111]

[0112] but

[0113] therefore, Thus, based on the methods in steps 3.1a to 3.3a, the prediction of the number of concurrent sessions in the 11th cycle of Edge Cloud 1 is completed. The prediction results are as follows. It is 12.

[0114] Next, this embodiment will be described in accordance with steps 3.1b to 3.2b:

[0115] According to step 3.1b, yN <= n <= y, so the value of n is 8, 9, or 10. Therefore, according to step 3.1b, we know that...

[0116] When m=8

[0117]

[0118] When m=9

[0119]

[0120] When m=10

[0121]

[0122] therefore,

[0123] Next, according to step 3.2b, the concurrent session number prediction module calculates the predicted concurrent session number for the 11th cycle of the service with a 5QI of 5 under Edge Cloud 1 based on equation (8). ,

[0124]

[0125] Thus far, based on the methods in steps 3.1b to 3.2b, the prediction of the number of concurrent sessions in the 11th cycle of Edge Cloud 1 has been completed, and the prediction results are as follows. for .

[0126] This application uses the historical concurrent session count as prior information to establish an interpolation coefficient set, generates a candidate set of concurrent session counts based on the interpolation coefficient set, and corrects the candidate set according to historical deviation empirical values ​​to form the concurrent session count prediction results for each edge cloud's 5QI. This invention effectively combines the prior information of historical concurrent session counts and deviation values ​​to achieve traffic prediction, greatly improves prediction accuracy, and provides key support for the dynamic expansion and contraction deployment of access network functions in the cloud.

Claims

1. A business forecasting method, characterized in that, Includes the following steps: Step 1: The statistics module counts the number of concurrent sessions with 5QI i in the y-th period under edge cloud x. ; Step 2: The concurrent session candidate set calculation module calculates the number of concurrent sessions based on the statistics of the most recent N+1 periods. , … Calculate the prediction candidate set for the number of concurrent sessions in the (y+1)th cycle. Where m takes the value [yN, y-1]; Step 3: The concurrent session count prediction module predicts the number of concurrent sessions based on... And the prediction bias value, calculate the number of predicted concurrent sessions with 5QI of i in the (y+1)th period under edge cloud x. .

2. The method as described in claim 1, characterized in that, In step 1, the number of concurrent sessions with 5QI i in the y-th period under the statistical edge cloud x is... , refers to the maximum number of concurrent terminal PDU sessions with a 5QI value of i residing in the connected state under the cell site managed by edge cloud x during the corresponding time period of the y-th cycle.

3. The method as described in claim 1, characterized in that, In step 2, the prediction candidate set of the number of concurrent sessions in the (y+1)th cycle is calculated. The methods include: Step 2.1: The concurrent session count candidate set calculation module calculates the number of concurrent sessions based on the service with 5QI i under the edge cloud x over the most recent N+1 periods. , … Construct N pairs of data {(m, ), (m+1, )}, where m takes the value [yN, y-1], and the interpolation coefficient set { is calculated according to equations (1) and (2). , }: （1） （2） Step 2.2: The concurrent session count candidate set calculation module is based on the interpolation coefficient set { , }, according to equation (3), the candidate value of the number of concurrent sessions for the service with 5QI i under edge cloud x in the (y+1)th period is calculated. ; （3） 。 4. The method as described in claim 1, characterized in that, In step 3, the predicted number of concurrent sessions with 5QI of i is calculated under the (y+1)th period of edge cloud x. The methods include: Step 3.1a: The concurrent session number prediction module calculates the preset correction coefficient based on equations (4) and (5). , ; （4） （5） Step 3.2a: Concurrent session count prediction module determination Is it greater than or equal to 0? If so, then... Assign to If not, then put Assign to ; Step 3.3a: The concurrent session prediction module calculates the predicted concurrent session count for the service with 5QI i under edge cloud x in the (y+1)th period based on equation (6). : （6） or, Step 3.1b: The concurrent session number prediction module calculates the correction coefficient based on equation (7). ; （7） Step 3.2b: The concurrent session count prediction module calculates the predicted concurrent session count for the service with 5QI i under edge cloud x in the (y+1)th period based on equation (8). : （8） 。 5. A business forecasting device, characterized in that, include: The statistics module is used to count the number of concurrent sessions with 5QI i in the y-th period under edge cloud x. ; The concurrent session count candidate set calculation module is used to calculate the number of concurrent sessions based on the statistics of the most recent N+1 periods. , … Calculate the prediction candidate set for the number of concurrent sessions in the (y+1)th cycle. Where m takes the value [yN, y-1]; The concurrent session count prediction module is used to predict the number of concurrent sessions based on... And the prediction bias value, calculate the number of predicted concurrent sessions with 5QI of i in the (y+1)th period under edge cloud x. .

Images