CN109040744B

CN109040744B - Method, device and storage medium for predicting key quality index of video service

Info

Publication number: CN109040744B
Application number: CN201810841823.5A
Authority: CN
Inventors: 曹瑞; 刘建国; 许海明
Original assignee: Huawei Technologies Co Ltd
Current assignee: Huawei Technologies Co Ltd
Priority date: 2018-07-27
Filing date: 2018-07-27
Publication date: 2020-03-10
Anticipated expiration: 2038-07-27
Also published as: CN109040744A

Abstract

The application discloses a method, a device and a storage medium for predicting KQI of video service, and belongs to the technical field of communication. The method comprises the following steps: and acquiring a characteristic tag table of the target geographic region, searching a KQI model matched with the characteristic tag table of the target geographic region from the KQI model base according to the characteristic tag table of the target geographic region, and predicting the KQI of the video service of the target geographic region according to the searched KQI model. That is, in the present application, a KQI model library is constructed in advance, when a KQI of a video service in a target geographic region needs to be predicted, a KQI model matching a feature tag table of the target geographic region may be searched from the KQI model library, and then the KQI of the video service in the target geographic region is predicted according to the searched model, so that it is not necessary to train a KQI model for the target geographic region in advance, and the universality of the method for predicting the KQI of the video service provided in the present application is improved.

Description

Method, device and storage medium for predicting key quality index of video service

Technical Field

The present application relates to the field of communications technologies, and in particular, to a method, an apparatus, and a storage medium for predicting a Key Quality Indicator (KQI) of a video service.

Background

With the rapid growth of video traffic, the growth of mobile data traffic of operators will be mainly caused by video traffic. At this time, when a certain geographical area is planned with a station address, if only the signal coverage and capacity of the geographical area are considered, after the base station is arranged in the geographical area according to the station address planning, the network in the geographical area is likely to not meet the video service requirement of the user. Therefore, when the site planning is performed on the geographic area, the KQI of the video service of the geographic area after the site planning is also required to be predicted, and if the predicted KQI of the video service does not meet the video service requirement of the user, the relevant parameters in the site planning are required to be adjusted, so that the KQI of the video service of the network after the planning is performed according to the adjusted relevant parameters in the practical application can meet the video service requirement of the user. The signal coverage rate is a ratio of signal strengths of all users in the geographic area exceeding a signal strength threshold, and the capacity is a Physical Resource Block (PRB) that can be allocated to the users in the geographic area.

Disclosure of Invention

The application provides a method, a device and a storage medium for predicting a KQI of a video service, which can be used for predicting the KQI of the video service in a geographical area when the site planning is carried out on the geographical area so as to ensure that the KQI of the video service of a network after a base station is planned meets the video service requirement of a user, and the technical scheme is as follows:

in a first aspect, a method for predicting a KQI of a video service is provided, the method comprising: acquiring a feature tag table of a target geographic area, wherein the feature tag table comprises at least one feature tag and a tag value of each feature tag, and the at least one feature tag is used for describing video services in a network of the target geographic area; searching a KQI model matched with the characteristic tag table of the target geographic region from a KQI model library according to the characteristic tag table of the target geographic region, wherein the KQI model library comprises a plurality of KQI models, each KQI model corresponds to one characteristic tag table, and each KQI model is used for predicting the KQI of the video service; and predicting the KQI of the video service of the target geographic area according to the searched KQI model.

In the application, a characteristic tag table of a target geographic area is obtained, a KQI model matched with the characteristic tag table of the target geographic area is searched from a KQI model library according to the characteristic tag table of the target geographic area, and the KQI of the video service of the target geographic area is predicted according to the searched KQI model. That is, in the present application, a KQI model library is constructed in advance, when a KQI of a video service in a target geographic region needs to be predicted, a KQI model matching a feature tag table of the target geographic region may be searched from the KQI model library, and then the KQI of the video service in the target geographic region is predicted according to the searched model, so that it is not necessary to train a KQI model for the target geographic region in advance, and the universality of the method for predicting the KQI of the video service provided in the present application is improved.

Optionally, the searching, according to the feature tag table of the target geographic area, a KQI model matching the feature tag table of the target geographic area from a KQI model library includes: and searching a KQI model matched with all the feature tags in the feature tag table of the target geographic region from the KQI model library, and taking the searched KQI model as the KQI model matched with the feature tag table of the target geographic region.

Since the signature table usually includes a plurality of signatures, the KQI model matching the signature table of the target geographical area is searched, that is, the KQI model matching each signature in the signature table of the target geographical area is searched.

Optionally, the at least one feature tag includes a first type of feature tag and a second type of feature tag, each feature tag in the feature tag table of the target geographic region includes at least one tag value, and each feature tag in the feature tag table of each KQI model in the KQI model library includes one tag value; the searching, from the KQI model library, a KQI model that matches all the feature tags in the feature tag table of the target geographic region includes: for any feature tag A in the feature tag table of the target geographic area, if the feature tag A in the target geographic area belongs to a first class of feature tags, searching a KQI model with the same tag value of the feature tag A in the KQI model as any tag value in at least one tag value of the feature tag A in the target geographic area from the KQI model library; and if the characteristic label A in the target geographic region belongs to a second class of characteristic labels, searching a KQI model with the similarity between the label value of the characteristic label A in the KQI model and any label value of at least one label value of the characteristic label A in the target geographic region larger than a similarity threshold from the KQI model library.

In the method, the feature tags are divided into the first class feature tags and the second class feature tags, the first class feature tags can be matched in the accurate matching mode, and the second class feature tags can be matched in the fuzzy matching mode, so that the universality of the method for predicting the KQI of the video service provided by the method is further improved.

Optionally, the first type of feature tag includes an operator name, a network type, a data source for recording a video document of a video service, a video resolution, and a carrier aggregation CA characteristic, and the second type of feature tag includes a city name, a data source of a video service, a buffering threshold of a video service, and a scene.

For example, the administrator may manually set the above tags as the first type feature tag and the second type feature tag in advance.

Optionally, the target geographic area is pre-divided into a plurality of grids; the predicting the KQI of the video service of the target geographic area according to the searched KQI model comprises the following steps: determining simulation parameters for planning the station address of the target geographic area, wherein the simulation parameters at least comprise the number of base stations to be deployed, the address of each base station in the base stations to be deployed and the base station engineering parameters of each base station; performing site planning simulation on the target geographic area according to the simulation parameters, and determining a variable value of each characteristic variable in at least one characteristic variable of each grid after simulation, wherein the at least one characteristic variable is a variable influencing the KQI; and determining the KQI of the video service of each grid according to the variable value of each characteristic variable in the at least one characteristic variable of each grid after simulation and the searched KQI model.

In the application, the target geographic area may be divided into a plurality of grids in advance, and at this time, the KQI of the video service of the target geographic area is predicted, that is, the KQI of the video service of each grid is predicted, so that the accuracy of predicting the KQI of the video service of the target geographic area is improved.

Optionally, the searched KQI models include at least two KQI models: determining a KQI of the video service of each grid according to the variable value of each of the at least one characteristic variable of each grid after simulation and the searched KQI model, including: continuously searching a KQI model matched with a first grid from the searched KQI models, wherein the first grid is one of the grids; and determining the KQI of the video service of the first grid according to the variable value of each characteristic variable in the at least one characteristic variable of the first grid after simulation and the searched KQI model matched with the first grid.

Since there may be more than one KQI model matching with the tag list of the target geographic region, when predicting the KQI of the video service of each grid, the KQI models matching with each grid may be continuously searched from the searched KQI models.

Optionally, the method further comprises: the method comprises the steps of obtaining a video receipt data source and at least one associated data source, wherein the video receipt data source comprises a plurality of video receipts, each associated data source comprises a plurality of pieces of data, and the at least one associated data source is a data source related to video services; for any video document B in the multiple video documents, determining data associated with the video document B in each associated data source, and forming the determined data and the video document B into a piece of structured data to obtain multiple pieces of structured data; analyzing the KQI included in each piece of structured data and the variable value of each characteristic variable in at least one characteristic variable; determining at least one feature tag of each piece of structured data, classifying the plurality of pieces of structured data according to the at least one feature tag of each piece of structured data to obtain a plurality of data sets, wherein each data set comprises at least one piece of structured data, each data set corresponds to one feature tag table, and the feature tag table corresponding to each data set is determined according to the feature tags of the structured data included in the corresponding data set; and for any data set C in the plurality of data sets, taking the KQI included in each piece of structured data in the data set C as output, taking the variable value of each feature variable in at least one feature variable included in each piece of structured data as input, training an initialized algorithm model to obtain a KQI model corresponding to the data set C, taking a plurality of obtained KQI models corresponding to the data sets one by one as models in the KQI model library, and taking a feature tag table corresponding to each data set as a feature tag table of the corresponding KQI model.

In the application, each KQI model in the KQI model library can be determined in advance in the above manner, and each KQI model is distinguished according to the feature tag table, so that when the KQI of the video service in the target geographic region is predicted subsequently, the KQI model matched with the target geographic region can be searched from the KQI model library according to the feature tag table of the target geographic region.

Optionally, the at least one associated data source includes a speech system data source, and each piece of data in the speech system data source includes a cell identifier and a speech system statistics period, and each piece of data is used to describe a cell network characteristic of a cell corresponding to the cell identifier in the speech system statistics period; the determining data associated with the video document B in each associated data source comprises: and searching data which has the same cell identification as the cell identification in the video bill B and has the video starting time and the video ending time in the video bill B within a session statistics period from the session data source, and determining the searched data as the data associated with the video bill B.

Optionally, the at least one associated data source includes a measurement report data source, and each piece of data in the measurement report data source includes a user terminal identifier and a reporting time of a measurement report, and each piece of data is used to describe a characteristic of a user terminal corresponding to the user terminal identifier in a network connection process; the determining data associated with the video document B in each associated data source comprises: searching data, in which a user terminal identifier is the same as a user terminal identifier in the video document B and the reporting time of a measurement report is in a video period of a video service corresponding to the video document B, from the measurement report data source, and determining the searched data as data associated with the video document B, wherein the starting time of the video period is the video starting time in the video document B minus a duration threshold, and the ending time of the video period is the video ending time in the video document B plus a duration threshold.

Optionally, the at least one associated data source includes a cell parameter data source, and each piece of data in the cell parameter data source includes a cell identifier, and each piece of data is used to describe a physical feature of a cell corresponding to the cell identifier; the determining data associated with the video document B in each associated data source comprises: and searching data with the same cell identification as the cell identification in the video document B from the cell working parameter data source, and determining the searched data as the data associated with the video document B.

In the application, the data related to the video document in the speech system data source, the measurement report data source and the cell parameter data source can be determined respectively through the above modes.

Optionally, the at least one feature tag of each piece of structured data comprises a scene, the scene comprising in an urban, suburban or high-speed movement; after obtaining the plurality of pieces of structured data, the method further includes: determining address position information of a user terminal corresponding to a video call ticket in each piece of structured data; accordingly, the determining at least one feature tag for each piece of structured data includes: and determining the scene of each piece of structured data according to the address position information of the user terminal corresponding to the video call ticket in each piece of structured data.

And the characteristic label scene of each piece of structured data is determined by the address position information of the user terminal corresponding to the video call ticket in each piece of structured data.

Optionally, before determining the KQI in each piece of structured data and the value of each of the at least one characteristic variable, the method further includes: acquiring a plurality of variables related to KQI of the video service; and removing variables of which the correlation coefficient with the KQI of the video service is smaller than a first threshold value and/or variables of which the sample distribution variance is smaller than a second threshold value and/or variables of which the importance coefficient with the KQI of the video service is smaller than a third threshold value from the plurality of variables, and taking the removed variables as the at least one characteristic variable.

In the present application, at least one characteristic variable which influences the KQI can be determined in the manner described above.

In a second aspect, there is provided an apparatus for predicting a KQI of a video service, the apparatus for predicting a KQI of a video service having a function of implementing the method behavior for predicting a KQI of a video service in the first aspect. The apparatus for predicting a KQI of a video service comprises at least one module, where the at least one module is configured to implement the method for predicting a KQI of a video service provided in the above first aspect.

In a third aspect, there is provided an apparatus for predicting a KQI of a video service, where the apparatus for predicting a KQI of a video service structurally includes a processor and a memory, and the memory is configured to store a program for enabling the apparatus for predicting a KQI of a video service to perform the method for predicting a KQI of a video service provided in the first aspect, and store data related to implementing the method for predicting a KQI of a video service provided in the first aspect. The processor is configured to execute programs stored in the memory. The operating means of the memory device may further comprise a communication bus for establishing a connection between the processor and the memory.

In a fourth aspect, a computer-readable storage medium is provided, which has instructions stored therein, which when run on a computer, cause the computer to perform the method for predicting a KQI of a video service according to the first aspect.

In a fifth aspect, a computer program product is provided comprising instructions which, when run on a computer, cause the computer to perform the method of predicting a KQI of a video service according to the first aspect.

The technical effects obtained by the above second, third, fourth and fifth aspects are similar to the technical effects obtained by the corresponding technical means in the first aspect, and are not described herein again.

Drawings

FIG. 1 is a schematic structural diagram of a computer device according to an embodiment of the present disclosure;

fig. 2 is a flowchart of a method for creating a KQI model library according to an embodiment of the present application;

fig. 3 is a flowchart of a method for predicting a KQI of a video service according to an embodiment of the present disclosure;

fig. 4 is a schematic diagram of an apparatus for predicting a KQI of a video service according to an embodiment of the present disclosure;

fig. 5 is a schematic diagram of another apparatus for predicting a KQI of a video service according to an embodiment of the present disclosure.

Detailed Description

To make the objects, technical solutions and advantages of the present application more clear, embodiments of the present application will be described in further detail below with reference to the accompanying drawings.

Before the method for predicting the KQI of the video service provided in the embodiment of the present application, an application scenario in the embodiment of the present application is briefly introduced. The station address planning means: and planning the base stations in the target geographic area so that the network in the target geographic area after the base stations are planned can meet certain conditions, such as the target signal coverage rate and the target capacity. At present, with the increase of video services, the demands of users on the network mainly come from the demands on the video services, and therefore, when planning a site of a target geographic area, the KQI of the network video services after planning a base station needs to be predicted first, so as to ensure that the KQI of the network video services after actually planning the base station meets the demands of the users on the video services. That is, the method for predicting the KQI of the video service provided in the embodiment of the present application is applied to a scenario in which a site is planned for a target geographic area. Of course, the method for predicting the KQI of the video service provided in the embodiment of the present application may also be applied to other scenes that need network adjustment, for example, in cell planning of a target geographic area. The site Planning may be Accurate Site Planning (ASP), and the cell Planning may be Accurate Cell Planning (ACP).

Fig. 1 is a schematic structural diagram of a computer device according to an embodiment of the present application. Referring to fig. 1, the computer device comprises at least one processor 101, a communication bus 102, a memory 103 and at least one communication interface 104.

The processor 101 may be a Central Processing Unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of programs according to the present disclosure.

The communication bus 102 may include a path that conveys information between the aforementioned components.

The Memory 103 may be a Read-Only Memory (ROM) or other type of static storage device that can store static information and instructions, a Random Access Memory (RAM) or other type of dynamic storage device that can store information and instructions, an electrically erasable Programmable Read-Only Memory (EEPROM), a Compact Disc Read-Only Memory (CD-ROM) or other optical Disc storage, optical Disc storage (including Compact Disc, laser Disc, optical Disc, digital versatile Disc, blu-ray Disc, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer, but is not limited to these. The memory 103 may be self-contained and coupled to the processor 101 via the communication bus 102. The memory 103 may also be integrated with the processor 101.

The communication interface 104 may be any device, such as a transceiver, for communicating with other devices or communication Networks, such as ethernet, Radio Access Network (RAN), Wireless Local Area Network (WLAN), etc.

In particular implementations, a computer device may include multiple processors, such as processor 101 and processor 105 shown in FIG. 1, as one embodiment. Each of these processors may be a single-core (single-CPU) processor or a multi-core (multi-CPU) processor. A processor herein may refer to one or more devices, circuits, and/or processing cores for processing data (e.g., computer program instructions).

In particular implementations, the computer device may also include an output device 106 and an input device 107, as one embodiment. The output device 106 is in communication with the processor 101 and may display information in a variety of ways. For example, the output device 106 may be a Liquid Crystal Display (LCD), a Light Emitting Diode (LED) display device, a Cathode Ray Tube (CRT) display device, a projector (projector), or the like. The input device 107 is in communication with the processor 101 and may receive user input in a variety of ways. For example, the input device 107 may be a mouse, a keyboard, a touch screen device, a sensing device, or the like.

The computer device may be a general purpose computer device or a special purpose computer device. In a specific implementation, the computer device may be a desktop computer, a laptop computer, a network server, a Personal Digital Assistant (PDA), a mobile phone, a tablet computer, a wireless terminal device, a communication device, or an embedded device. The embodiment of the application does not limit the type of the computer equipment.

The memory 103 is used for storing program codes for executing the scheme of the application, and is controlled by the processor 101 to execute. The processor 101 is used to execute program code stored in the memory 103. One or more software modules may be included in the program code.

Next, a method for predicting a KQI of a video service according to an embodiment of the present application will be explained in detail. It should be noted that, in the embodiment of the present application, in order to facilitate quick prediction of a KQI of a video service, a KQI model library is created in advance, and the KQI model library includes a plurality of KQI models. In this way, when the KQI of the video service in the target geographical region after the station address planning needs to be predicted subsequently, a KQI model matched with the target geographical region can be searched from the KQI model library, and then the KQI of the video service in the network of the target geographical region can be predicted according to the searched KQI model. That is, the method for predicting the KQI of the video service provided in the embodiment of the present application includes two parts, where the first part is to create a KQI model library, and the second part is to predict the KQI of the video service in the target geographic region according to the KQI model library. The following two embodiments will be described with respect to these two parts.

Fig. 2 is a flowchart of a method for creating a KQI model library according to an embodiment of the present application, and as shown in fig. 2, the method includes the following steps:

step 201: the method comprises the steps of obtaining a video receipt data source and at least one associated data source, wherein the video receipt data source comprises a plurality of video receipts, each associated data source comprises a plurality of pieces of data, and the at least one associated data source is a data source related to video services.

The video bill data source comprises a plurality of video bills, and each video bill is used for recording related information of one video service. For example, the video document may include information of the video service, such as KQI, video download rate, video definition, video buffering time, video playing time, video pause duration, and video size. The video documents in the video document data source can be acquired manually or automatically by a probe deployed in a network element.

In addition, the KQI of the video service has many expressions, for example, the KQI of the video service may be a video Mean opinion Score (vMOS) of the video service, where the vMOS is a system for evaluating the quality of the video service, and the quality of each video service is evaluated in 1 to 5 minutes, for example, the vMOS of the video service 1 is 4 minutes, and the vMOS of the video service 2 is 4.2 minutes, so that the quality of the video service 2 is higher than that of the video service 1.

In addition, the at least one associated data source refers to a data source related to a video service, and in the present application, the at least one associated data source may include a speech system data source, a measurement report data source, a cell parameter data source, and the like.

In the working process of the base station, the base station counts the cell network characteristic information of each cell at regular intervals to obtain a piece of speech system data corresponding to each cell, so that the data included in a speech system data source is speech system data, each piece of speech system data corresponds to one cell, each piece of speech system data is used for describing the cell network characteristic information of the corresponding cell within a certain time, and the cell network characteristic information comprises the number of activated users, the cell throughput rate and other information. That is, each piece of data in the speech system data source includes a cell identifier and a speech system statistics period, and each piece of data is used to describe the cell network characteristics of the cell corresponding to the cell identifier in the speech system statistics period.

In the process of connecting the user terminal with the network, the user terminal reports a measurement report to the base station, wherein the measurement report comprises information such as the level strength, the channel quality and the Used resource blocks (RB _ Used) of a serving cell detected by the user terminal, and therefore, data included in a measurement report data source is a measurement report, and each measurement report corresponds to one user terminal. That is, each piece of data in the measurement report data source includes a ue identifier and a reporting time of the measurement report, and each piece of data is used to describe a characteristic of the ue corresponding to the ue identifier in the network connection process.

In addition, the data included in the cell operating parameter data source is cell operating parameter data, each cell operating parameter data corresponds to one cell, and each cell operating parameter data is used for describing the physical characteristics of the corresponding cell, and the physical characteristics include information such as cell transmission power, cell frequency band, cell bandwidth and the like. That is, each piece of data in the cell parameter data source includes a cell identifier, and each piece of data is used to describe the physical characteristics of the cell corresponding to the cell identifier.

Step 202: and for any one video document B in the multiple video documents, determining data associated with the video document B in each associated data source, and forming a piece of structured data by the determined data and the video document B to obtain multiple pieces of structured data.

Since the at least one associated data source includes data sources such as a speech system data source, a measurement report data source, and a cell parameter data source, the following description can be made to determine the data associated with the video document B in each associated data source:

(1) and when the at least one associated data source comprises a speech system data source, determining data associated with the video document B in the speech system data source.

Since each piece of data in the speech system data source includes a cell identifier and a speech system statistics period, for example, data that the cell identifier is the same as the cell identifier in the video document B and the video start time and the video end time in the video document B are both within the speech system statistics period may be searched from the speech system data source, and the searched data is determined as data associated with the video document B.

(2) When the at least one associated data source comprises a measurement report data source, determining data associated with the video document B in the measurement report data source.

Each piece of data in the measurement report data source includes a user terminal identifier and the reporting time of the measurement report, so, for example, data that the user terminal identifier is the same as the user terminal identifier in the video document B and the reporting time of the measurement report is within the video cycle of the video service corresponding to the video document B may be searched from the measurement report data source, and the searched data is determined as data associated with the video document B.

The starting time of the video period is the video starting time in the video document B minus the time threshold, and the ending time of the video period is the video ending time in the video document B plus the time threshold. The duration threshold is a set redundant duration, and the duration threshold may be 2 seconds, 5 seconds, or the like.

(3) And when the at least one associated data source comprises a cell working parameter data source, determining data associated with the video document B in the cell working parameter data source.

Since each piece of data in the cell parameter data source includes a cell identifier, for example, data having the same cell identifier as the cell identifier in the video document B may be searched from the cell parameter data source, and the searched data may be determined as data associated with the video document B.

After data associated with the video document B in the telephone system data source, data associated with the video document B in the measurement report data source and data associated with the video document B in the cell parameter data source are respectively determined, the determined associated data and the video document B can be combined to obtain a piece of structural data corresponding to the video document B. When the above operation is performed on each video document in the video document data source, one piece of structured data corresponding to each video document can be obtained, that is, a plurality of pieces of structured data are obtained.

After the plurality of pieces of structured data are obtained through step 202, each of the KQI models in the KQI model library may be determined according to steps 203 to 205 described below. Further, in order to improve the accuracy of the KQI model provided in the embodiment of the present application, after obtaining the plurality of pieces of structured data, data cleaning may be performed on the plurality of pieces of structured data to delete the structured data with higher noise and the wrong structured data in the plurality of pieces of structured data.

For example, the implementation manner of deleting the structural data with higher noise in the plurality of pieces of structural data may be: clustering the multiple pieces of structured data to obtain multiple sets, wherein each set corresponds to a central point, and deleting data farther away from the central point in any set. The data farther from the center point is the data with larger noise.

For example, the implementation manner of deleting erroneous structured data in the plurality of pieces of structured data may be: and deleting null data in the plurality of pieces of structured data, wherein the null data refers to data of which the variable value corresponding to a certain variable in the structured data is null, and if the variable value corresponding to a certain variable in a certain piece of structured data is null, the structured data is likely to be wrong structured data, so that the piece of structured data needs to be deleted.

For example, the implementation manner of deleting the wrong structured data in the plurality of pieces of structured data may further be: deleting data of which the numerical value corresponding to a certain variable in the plurality of pieces of structured data is not in a specified range, wherein the specified range refers to the actual value range of the variable. For example, if the variable is signal strength, the actual value of the signal strength is usually between-130 and-45, and if the value of the signal strength in a piece of structured data is not within the range, the piece of structured data is likely to be wrong structured data, and therefore the piece of structured data needs to be deleted.

In addition, in the embodiment of the present application, after obtaining the plurality of pieces of structured data, since the geographic location of the user terminal of the video service in each piece of structured data is known, the plurality of pieces of structured data may be mapped into the geographic location coordinate system, so that other operations may be performed subsequently according to the structured data mapped into the geographic location coordinate system.

Illustratively, for any piece of structured data, longitude and latitude information of a user terminal of a video service in the structured data is acquired through a measurement report positioning technology, and the acquired longitude and latitude information is used as the longitude and latitude information of the structured data. And mapping each piece of structured data to a geographic position coordinate system according to the longitude and latitude information of each piece of structured data to obtain a plurality of points in the geographic position coordinate system, wherein each point corresponds to one piece of structured data, and the value of the KQI of the video service in the structured data corresponding to each point can be marked in the geographic position coordinate system, so that a manager can know the distribution condition of the KQI of the video service in each area through the geographic position coordinate system.

In addition, different colors can be used for representing different values of the KQI, and then the corresponding color is adopted to mark the point in the geographic position coordinate system according to the value of the KQI of the video service in the structured data corresponding to the point, so that a manager can quickly know the distribution condition of the KQI of the video service in each area through the color distribution condition in the geographic position coordinate system.

In addition, in the embodiment of the application, since the network distribution in each region included in the target geographic region may be different, so that the KQIs of the video services in each region may also be different, when the KQI of the video service in the target geographic region is predicted subsequently, the target geographic region is divided into a plurality of grids in advance, and then the KQI of the video service in each grid is predicted, so as to realize accurate prediction of the KQI of the video service in the target geographic region. Therefore, in the process of mapping the pieces of structured data to the geographic position coordinate system, the geographic position coordinate system may be divided into a plurality of grids in advance, and each grid has the same size, for example, each grid has a size of 50 × 50 meters. An identifier is assigned to each grid, and the longitude and latitude information of each grid can be represented by the longitude and latitude information of a certain point in the grid, for example, the longitude and latitude information at the position of the upper left corner of the grid. In this way, when the subsequent manager predicts the KQI of the video service in a certain grid, the predicted KQI may be compared according to the distribution of the KQI of the video service existing in the grid. For convenience of description, the above process is referred to as data rasterization.

Step 203: the KQI included in each piece of the structured data and the variable value of each of the at least one characteristic variable are analyzed.

In which at least one characteristic variable refers to a variable affecting the KQI, since there is a relatively large amount of information included in each piece of structured data, but in the embodiment of the present application, only the information affecting the KQI is useful, it is necessary to analyze the KQI included in each piece of structured data and the variable value of each of the at least one characteristic variable through step 203.

Illustratively, the at least one characteristic variable includes End-To-End loop-back Time (E2E _ RTT), Signal-To-Interference plus Noise Ratio (SINR), Reference Signal Received Power (RSRP), RB Used, and Channel Quality Indicator (CQI), among other information. For example, when at least one of the characteristic variables is E2E _ RTT, SINR, RSRP, RB _ Used, and CQI, for each piece of structured data, a value corresponding to KQI, a value corresponding to the characteristic variable E2E _ RTT, a value corresponding to the characteristic variable SINR, a value corresponding to the characteristic variable RSRP, a value corresponding to the characteristic variable RB _ Used, and a value corresponding to the characteristic variable CQI in the structured data need to be analyzed.

In addition, the at least one characteristic variable is predetermined, and in one possible implementation, the determining the at least one characteristic variable may be: acquiring a plurality of variables related to KQI of the video service; and removing variables of which the correlation coefficient with the KQI of the video service is smaller than a first threshold value from the plurality of variables, and/or removing variables of which the sample distribution variance is smaller than a second threshold value, and/or removing variables of which the importance coefficient with the KQI of the video service is smaller than a third threshold value, wherein the removed variables are used as at least one characteristic variable. That is, after acquiring a plurality of variables related to the KQI of the video service, at least one of the above three culling operations is performed from the plurality of variables, and then the culled variable is used as at least one feature variable. The following describes how to determine at least one feature variable from a plurality of variables by taking the above three culling operations as examples:

(1) and removing the variable with the correlation coefficient smaller than the first threshold value with the KQI of the video service from the plurality of variables.

The removing of the variable having a correlation coefficient with the KQI of the video service smaller than the first threshold from the plurality of variables may be implemented by: and analyzing the correlation coefficient between each variable and the KQI of the video service by a Pearson correlation coefficient method, and then rejecting the variable of which the correlation coefficient is smaller than a first threshold value. The first threshold is a preset value. Of course, the correlation coefficient between each variable and the KQI of the video service may also be analyzed by other correlation coefficient analysis methods, and the embodiment of the present application is not specifically limited herein.

In addition, when analyzing the correlation coefficient between each variable and the KQI of the video service by using the pearson correlation coefficient method, data related to the video service needs to be referred to, in this embodiment, the structured data determined in step 202 may be directly referred to, and of course, other data related to the video service may also be referred to, which is not limited specifically.

(3) And continuously eliminating the variables with the sample distribution variance smaller than the second threshold from the variables after the first elimination.

For any one of the variables after the first elimination, according to the pre-counted sample data, determining all variable values of the variable appearing in the sample data, determining the sample distribution variance of the variable according to all the determined variable values, and then eliminating the variable with the sample distribution variance smaller than the second threshold value.

Wherein the second threshold is a preset value. In addition, the sample data counted in advance may be the multiple pieces of structured data determined in step 202, or may also be sample data acquired through other ways, which is not specifically limited herein in this embodiment of the application.

(4) And continuously removing the variables of which the importance degree coefficient with the KQI of the video service is smaller than a third threshold value from the variables after the second removal, and determining the variables after the third removal as at least one characteristic variable.

For example, for each of the variables after the second culling, an importance coefficient between the variable and a KQI of the video service may be analyzed through a Gradient Boosting Decision Tree (GBDT) algorithm, and then the variable with the importance coefficient smaller than a third threshold may be culled. Wherein the third threshold is a preset value.

Similarly, when analyzing the correlation coefficient between each variable and the KQI of the video service through the GBDT algorithm, data related to the video service needs to be referred to, in this embodiment of the application, the structured data determined in step 202 may be directly referred to, and of course, other data related to the video service may also be referred to, which is not limited specifically.

And (3) after the plurality of variables in the step (1) are subjected to the three times of elimination in the steps (2) to (4), the remaining variables are at least one characteristic variable. For example, the plurality of variables are 50 variables, and the at least one characteristic variable is E2E _ RTT, SINR, RSRP, RB _ Used, CQI, which is equivalent to selecting the 5 variables from the 50 variables as the at least one characteristic variable.

It should be noted that the above-mentioned three-time elimination process is only one possible implementation manner, and for example, the order of the three-time elimination process may be adjusted according to requirements, that is, the three-time elimination process does not have a strict sequence. And in practice, only one or two of the above-mentioned culling operations may be performed, which will not be described in detail herein.

In addition, the implementation manner of acquiring multiple variables related to the KQI of the video service may be: all variables that may be related to the KQI of the video service are manually collected, and all of the collected variables are determined as the plurality of variables. Optionally, the implementation manner of obtaining multiple variables related to the KQI of the video service may further be: in the searching step 203, all variables present in the structured data are determined, and all variables searched are determined as the plurality of variables.

Step 204: determining at least one feature tag of each piece of structured data, classifying the plurality of pieces of structured data according to the at least one feature tag of each piece of structured data to obtain a plurality of data sets, wherein each data set comprises at least one piece of structured data, each data set corresponds to one feature tag table, and the feature tag table corresponding to each data set is determined according to the feature tags of the structured data included in the corresponding data set.

In the embodiment of the present application, each KQI model in the KQI model library is distinguished by using the feature tag table, that is, each KQI model corresponds to one feature tag table. Therefore, after the plurality of pieces of structured data are determined, the determined structured data need to be classified according to at least one feature tag through step 204, so as to determine the KQI model according to the classified data set. The feature tag table corresponding to each data set is obtained by merging at least one tag of the structured data included in each data set.

Illustratively, the at least one tag may include an operator name, a network type, a data source for recording a video document of the video service, a video resolution and Carrier Aggregation (CA) characteristic, a city name, a data source of the video service, a buffering threshold and a scene of the video service, and the like. In the embodiment of the present application, at least one tag is manually set, and in practical application, at least one tag may also include other information according to requirements, which is not specifically limited herein.

Correspondingly, the feature tag table corresponding to each data set may also include feature tags such as an operator name, a network type, a data source for recording a video document of a video service, video resolution and CA characteristics, a city name, a data source of the video service, and an initial/slow threshold and scene of the video service. Table 1 is a feature tag table provided in the embodiment of the present application, and as shown in table 1, feature tags included in the feature tag table are respectively: the system comprises an operator name, a network system, a data source for recording a video document of a video service, video resolution and CA characteristics, a city name, a data source of the video service, a buffering threshold value and a scene of the video service. The feature tag table further includes a tag value corresponding to each feature tag, and table 1 does not list what the tag value of each feature tag is, but replaces with "xxx", and does not constitute a limitation to the embodiments of the present application.

In addition, some tags in the at least one tag are used to describe key features in the structured data, and some tags are used to describe non-key features in the structured data. Thus, in one possible implementation, the at least one tag includes a first class of feature tags and a second class of feature tags, where the first class of feature tags refers to tags used to describe key features in the structured data and the second class of feature tags refers to tags used to describe non-key features in the structured data. At this time, for the first-class feature tag, when classifying a plurality of pieces of structured data, it is necessary that the hard tags of the structured data included in each data set after the classification are the same. For the second class feature labels, it is only necessary that the second class feature labels of the structured data included in each data set after classification are similar. The first type of feature tag and the second type of note tag are set by a manager according to requirements.

Illustratively, the first type of feature tags include tags such as an operator name, a network type, a data source for recording a video document of a video service, a video resolution, and CA characteristics. The second type of feature tags include tags such as city names, data sources of video services, and initial and slow thresholds and scenes of video services.

The operator name refers to a name of an operator of a network where the video service in the structured data is located. The network standard refers to a network standard adopted by a network where the video service in the structured data is located, for example, the network standard may be a Global System for Mobile Communications (GSM) network, a Universal Mobile Telecommunications System (UMTS) network, a Code Division Multiple Access (CDMA) network, a Long Term evolution (Long Term evolution lte) network, and the like. The data source for recording the video documents of the video service comprises a drive test data source or a probe data source, wherein the drive test data source refers to that the video documents in the structured data are acquired manually, and the probe data source refers to that the video documents in the structured data are acquired automatically by adopting a probe deployed in a network element. The CA characteristic refers to a carrier aggregation mode acquired by a network where a video service in the structured data is located, where carrier aggregation refers to aggregation of carriers of different frequency bands to improve a bandwidth of the network. The city name refers to a name of a city where a network corresponding to the video service in the structured data is located. The data source of the video service refers to a source of video data of the video service in the structured data, for example, the data source of the video service may be a specific video website. The buffering threshold of the video service refers to a minimum buffering amount when the video corresponding to the video service in the structured data can be played, for example, the buffering threshold of the video service is 10%, which indicates that the terminal can play the video corresponding to the video service only when the buffering amount of the video reaches 10%. A scene refers to the type of area in which the video service is located in the structured data, for example, a scene may include an urban area or a suburban area.

When at least one feature tag includes a scene, the scene of each piece of structured data may be determined according to the longitude and latitude information of each piece of structured data determined in step 202, which is not described in detail herein.

TABLE 1

Feature tag	Tag value
		Operator	×××
Network system	×××
		Data source for recording video documents of video services	×××
Video resolution	×××
		CA characteristics	×××
Name of city	×××
		Data source of video service	×××
Initial buffering threshold of video service	×××
		Scene	×××

Step 205: for any data set C in the multiple data sets, taking the KQI included in each piece of structured data in the data set C as output, taking the variable value of each feature variable in at least one feature variable included in each piece of structured data as input, training the initialized algorithm model to obtain a KQI model corresponding to the data set C, taking the obtained multiple KQI models corresponding to the multiple data sets one by one as models in a KQI model library, and taking a feature tag table corresponding to each data set as a feature tag table of the corresponding KQI model.

For any data set C in the plurality of data sets, the initialized algorithm model is trained by using the KQI of each piece of structured data in the data set C as an output and using the variable value of each feature variable in at least one feature variable of each piece of structured data as an input, so that the trained KQI model can be used for predicting the KQI of the video service. In addition, each data set corresponds to one feature tag table, so when a KQI model corresponding to each data set is obtained, the feature tag table corresponding to the KQI model is the feature tag table corresponding to the data set.

In this embodiment of the present application, the implementation manner of selecting the initialized algorithm model may be: determining a plurality of algorithm models; selecting a data set from a plurality of data sets, and dividing structured data in the selected data set into two types, wherein the first type is used for training a model, and the second type is used for evaluating the model; training each algorithm model through the first type of structured data, and then evaluating each trained algorithm through the second type of structured data to obtain an evaluation index of the trained algorithm model; an algorithm model with the best evaluation index is determined from the trained algorithm models, and the obtained algorithm model is determined as the initialized algorithm model in step 205. Illustratively, in the embodiment of the present application, after the above operations, the preselected algorithm model is a random forest algorithm.

The evaluation index can be a model judgment coefficient, and the closer the model judgment coefficient is to 1, the closer the predicted value and the true value determined by the corresponding algorithm model are. The evaluation index can also be the root mean square error between the predicted value and the actual value, and the smaller the root mean square error is, the higher the usability of the corresponding model is. The evaluation index may also be fn _ cdf0.3, where fn _ cdf0.3 refers to the prediction accuracy of the model when the deviation range of the model is kept within 0.3, and generally, if fn _ cdf0.3 is greater than 0.7, the model can be considered to have high availability.

In addition, after the KQI model library is constructed, the KQI model library needs to be updated in order to further improve the accuracy of each KQI model in the KQI model library in predicting the KQI of the video service. And updating the KQI model library, wherein the updating comprises operations of data updating, model deleting or model merging and the like.

Wherein, the data updating means: and acquiring a new data source, and retraining the KQI model by adopting the new data source. Model deletion means: some of the KQI models in the KQI model library are deleted manually. Model merging means: if the similarity between the feature label tables of the two KQI models is high, the data for training the two KQI models can be merged, and the merged data is adopted to retrain the data to obtain one KQI model.

In the embodiment of the application, a KQI model base can be constructed according to the obtained video receipt data source and at least one associated data source, so that a KQI model matched with a target geographic area can be directly searched from the model base subsequently, and the efficiency of predicting the KQI of the video service in the target geographic area is improved.

The above-described embodiment shown in fig. 2 is used to explain a specific process of creating a KQI model library, and the following embodiment is used to explain how to predict a KQI of a target geographical region based on the created KQI model library.

Fig. 3 is a flowchart of a method for predicting a KQI of a video service according to an embodiment of the present disclosure, where as shown in fig. 3, the method includes the following steps:

step 301: and acquiring a characteristic label table of the target geographic area, wherein the characteristic label table comprises at least one characteristic label and a label value of each characteristic label, and the at least one characteristic label is used for describing video services in a network of the target geographic area.

Because the KQI models in the KQI model base are distinguished by the feature tag table, when the KQI of the video service in the target geographic area needs to be predicted, the feature tag table of the target geographic area needs to be determined first, so that the KQI model matched with the target geographic area can be searched from the KQI model base according to the feature tag table of the target geographic area.

The feature tag included in the feature tag table may be a feature tag in a feature tag table of each KQI model in the KQI model library, for example, the feature tag in the feature tag table is the feature tag in table 1 in the embodiment of fig. 2, and at this time, when the KQI of the video service of the target geographic area needs to be predicted, the tag value of the target geographic area on each feature tag in table 1 is determined, so as to obtain the feature tag table of the target geographic area. At least one tag has been explained in detail in the embodiment shown in fig. 2 and will not be explained further here.

In addition, for a certain feature tag, different regions in the target geographic area may have different tag values for the feature tag, in which case, in the feature tag table of the target geographic area, the tag value of the feature tag includes all possible tag values in the target geographic area. Thus, in the feature tag table for the target geographic area, each feature tag includes at least one tag value. For example, the video resolutions of video services in different areas within the target geographic area may be different, and thus the tag values for the feature tag video resolutions in the feature tag table of the target geographic area may have a plurality of values, such as video resolution 1, video resolution 2, and video resolution 3 for the feature tag video resolutions, respectively.

Step 302: and searching a KQI model matched with the characteristic tag table of the target geographical region from the KQI model library according to the characteristic tag table of the target geographical region.

Since the feature tag table includes at least one feature tag, in a possible implementation, step 302 may be: and searching a KQI model matched with all the feature tags in the feature tag table of the target geographic region from the KQI model library, and taking the searched KQI model as the KQI model matched with the feature tag table of the target geographic region. For example, for any feature tag a in the feature tag table of the target geographic region, a KQI model matching the feature tag a is searched from the KQI model library, and the KQI model matching each feature tag in the feature tag table of the target geographic region is determined as the KQI model matching the feature tag table of the target geographic region. That is, the searched KQI model requires matching of each signature in the signature table of the target geographical area.

As can be seen from the embodiment shown in fig. 2, at least one feature tag includes a first-class feature tag and a second-class feature tag, and the first-class feature tag is used to describe a key feature, so that for any first-class feature tag, a tag value of the first-class feature tag in the searched KQI model needs to be the same as a certain tag value of the first-class feature tag in the feature tags of the target geographic area. The second type of feature tags are used for describing non-critical features, and therefore, for any second type of feature tag, the tag value of the soft tag in the searched KQI model needs to be similar to a certain tag value of the soft tag in the feature tags of the target geographic area.

In addition, each feature tag in the feature tag table of each KQI model in the KQI model library includes a tag value, and therefore, the implementation manner of searching for the KQI model matching the feature tag a from the KQI model library may be: if the characteristic label A of the target geographic region belongs to the first class of characteristic labels, searching a KQI model with the label value of the characteristic label A in the KQI model being the same as any one of at least one label value of the characteristic label A of the target geographic region from a KQI model library; and if the characteristic label A of the target geographic region belongs to the second type of characteristic labels, searching a KQI model with the label value of the characteristic label A in the KQI model being similar to any one of at least one label value of the characteristic label A of the target geographic region from the KQI model. In the embodiment of the present application, a method for matching the first type of feature tag may be referred to as an exact matching method, and a method for matching the second type of feature tag may be referred to as a fuzzy matching method.

Since the tag value corresponding to the tag may be numeric or enumerated. For numerical tag values, the similarity between two tag values may be that the similarity between two tag values is larger, that is, the difference between two tag values is smaller. For enumerated tag values, similarity between two tag values may be such that an element in one tag value includes an element in the other tag value.

Step 303: and predicting the KQI of the video service of the target geographic area according to the searched KQI model.

Since, in this embodiment of the present application, predicting the KQI of the video service in the target geographic area refers to predicting the KQI of the video service in the target geographic area after planning the station address, in a possible implementation manner, step 303 may be: determining simulation parameters for planning the station address of the target geographic area, wherein the simulation parameters at least comprise the number of base stations to be deployed, the address of each base station in the base stations to be deployed and the base station engineering parameters of each base station; performing site planning simulation on a target geographic area according to simulation parameters, dividing the target geographic area into a plurality of grids in advance, and determining a variable value of each characteristic variable in at least one characteristic variable of each grid after simulation; and determining the KQI of the video service of each grid according to the variable value of each characteristic variable in the at least one characteristic variable of each grid after simulation and the searched KQI model.

If only one searched KQI model is available, the KQI of the video service of each grid can be directly predicted according to the searched KQI model. However, since the feature tag in the feature tag table of the target geographic area may include a plurality of tag values, the number of searched KQI models may also be multiple, and thus, the determination of the KQI of the video service of each grid according to the variable value of each of the at least one feature variable of each grid after the simulation and the searched KQI model may be implemented as follows: continuously searching a KQI model matched with a first grid from the searched KQI models, wherein the first grid is one of the grids; and determining the KQI of the video service of the first grid according to the variable value of each characteristic variable in the at least one characteristic variable of the first grid after simulation and the searched KQI model matched with the first grid. That is, if there are multiple searched KQI models, the KQI models matching the respective grids need to be continuously searched from the multiple searched KQI models, and the KQI models corresponding to the respective grids one to one are obtained.

The implementation manner of continuously searching the KQI model matched with the first grid from the searched KQI models may be as follows: and determining a characteristic label table of the first grid, and searching a KQI model matched with the characteristic label table of the first grid from the searched KQI models. Searching for a KQI model matching the tag list of the first grid from the searched KQI models may refer to searching for a KQI model matching the tag list of the target geographic area from a KQI model library, which will not be described in detail herein.

The KQI of the video service of each grid in the target geographic area can be predicted through step 302, and since the KQI method for predicting video service provided by the present application is used to ensure that the KQI of the video service of the network after the base station is planned meets the video service requirement of the user, for any grid in the target geographic area, if the determined KQI does not reach the KQI threshold, the simulation parameters are adjusted, and the site planning simulation is performed again on the target geographic area according to the adjusted simulation parameters. Determining a variable value of each characteristic variable in the at least one characteristic variable of the grid after re-simulation, determining the KQI of the video service of the grid according to the variable value of each characteristic variable in the at least one characteristic variable of each grid after re-simulation and the searched KQI model matched with the grid, continuously judging whether the predicted KQI of the video service reaches a KQI threshold value, and repeating the operation until the KQI of the video service of the grid is determined to reach the KQI threshold value. At this time, the simulation parameters after the last adjustment can be used as a reference for site planning of the target geographical area.

In addition, in this embodiment of the present application, when the KQI of the video service of each grid in the target geographic area reaches the KQI threshold, the following may be performed: the KQI of the video traffic for a certain proportion of the grid reaches the KQI threshold. The KQIs of the video services of all grids do not need to reach the KQI threshold, so that the efficiency of determining the site planning parameters of the target geographic area can be improved.

Wherein the KQI threshold may be a threshold set for the entire target geographical area. Optionally, the grids in the target geographic area may be classified in advance, and a threshold may be set for each type of grid, so as to improve accuracy of the KQI of the predicted video service. In addition, if a certain grid in the target geographic area is the grid after data rasterization in step 202 in the embodiment shown in fig. 2, the KQI threshold at this time may be an average of the KQI values in the structured data that the grid falls within the grid after data rasterization.

Referring to fig. 4, an embodiment of the present application provides an apparatus for predicting a KQI of a video service, including:

a first obtaining module 401, configured to perform step 301 in the embodiment of fig. 3;

a lookup module 402 for performing step 302 in the embodiment of fig. 3;

a prediction module 403, configured to perform step 303 in the embodiment of fig. 3.

Optionally, the lookup module 402 includes:

and the searching unit is used for searching the KQI model matched with all the characteristic labels in the characteristic label table of the target geographic region from the KQI model library, and the searched KQI model is used as the KQI model matched with the characteristic label table of the target geographic region.

Optionally, the at least one feature tag includes a first type of feature tag and a second type of feature tag, each feature tag in the feature tag table of the target geographic region includes at least one tag value, and each feature tag in the feature tag table of each KQI model in the KQI model library includes one tag value;

the search unit is specifically configured to:

for any characteristic label A in the characteristic label table of the target geographic region, if the characteristic label A of the target geographic region belongs to a first type of characteristic label, searching a KQI model with the same label value of the characteristic label A in the KQI model and any label value in at least one label value of the characteristic label A of the target geographic region from a KQI model library;

and if the characteristic label A of the target geographic region belongs to the second type of characteristic labels, searching a KQI model with the label value of the characteristic label A in the KQI model being similar to any one of at least one label value of the characteristic label A of the target geographic region from the KQI model.

Optionally, the first type of feature tag includes an operator name, a network type, a data source for recording a video document of a video service, a video resolution, and a carrier aggregation CA characteristic, and the second type of feature tag includes a city name, a data source of a video service, a threshold of an initial buffer level of a video service, and a scene.

Optionally, the target geographic area is pre-divided into a plurality of grids;

the prediction module 403 includes:

the system comprises a first determining unit, a second determining unit and a control unit, wherein the first determining unit is used for determining simulation parameters for planning the station address of a target geographic area, and the simulation parameters comprise the number of base stations to be deployed, the address of each base station in the base stations to be deployed and the base station engineering parameters of each base station;

the second determining unit is used for carrying out site planning simulation on the target geographic area according to the simulation parameters and determining the variable value of each characteristic variable in at least one characteristic variable of each grid after simulation, wherein the at least one characteristic variable is a variable influencing the KQI;

and the third determining unit is used for determining the KQI of the video service of each grid according to the variable value of each characteristic variable in the at least one characteristic variable of each grid after simulation and the searched KQI model.

Optionally, the searched KQI models include at least two KQI models:

a third determining unit, specifically configured to:

continuously searching a KQI model matched with a first grid from the searched KQI models, wherein the first grid is one of the grids;

and determining the KQI of the video service of the first grid according to the variable value of each characteristic variable in the at least one characteristic variable of the first grid after simulation and the searched KQI model matched with the first grid.

Optionally, referring to fig. 5, the apparatus 400 further comprises:

a second obtaining module 404, configured to perform step 201 in the embodiment of fig. 2;

a first determining module 405, configured to perform step 202 in the embodiment of fig. 2;

an analysis module 406 for performing step 203 in the embodiment of fig. 2;

a second determining module 407, configured to perform step 204 in the embodiment of fig. 2;

a training module 408, configured to perform step 205 in the embodiment of fig. 2.

Optionally, the at least one associated data source includes a speech system data source, and each piece of data in the speech system data source includes a cell identifier and a speech system statistics period, and each piece of data is used to describe a cell network characteristic of a cell corresponding to the cell identifier in the speech system statistics period;

the first determining module 405 is specifically configured to:

and searching data which has the same cell identification as the cell identification in the video document B and the video starting time and the video ending time in the video document B in a speech system counting period from a speech system data source, and determining the searched data as data associated with the video document B.

Optionally, the at least one associated data source includes a measurement report data source, and each piece of data in the measurement report data source includes a user terminal identifier and reporting time of a measurement report, and each piece of data is used to describe a characteristic of a user terminal corresponding to the user terminal identifier in a network connection process;

the first determining module 405 is specifically configured to:

and searching data, of which the user terminal identification is the same as that in the video document B and the reporting time of the measurement report is in the video period of the video service corresponding to the video document B, from a measurement report data source, determining the searched data as data associated with the video document B, wherein the starting time of the video period is the video starting time in the video document B minus a duration threshold, and the ending time of the video period is the video ending time in the video document B plus the duration threshold.

Optionally, the at least one associated data source includes a cell parameter data source, and each piece of data in the cell parameter data source includes a cell identifier, and each piece of data is used to describe a physical feature of a cell corresponding to the cell identifier;

the first determining module 405 is specifically configured to:

and searching data with the same cell identification as the cell identification in the video document B from the cell parameter data source, and determining the searched data as the data associated with the video document B.

Optionally, the apparatus 400 further comprises:

the third acquisition module is used for acquiring a plurality of variables related to the KQI of the video service;

and the removing module is used for removing a variable with a correlation coefficient with the KQI of the video service smaller than a first threshold value, or with a sample distribution variance smaller than a second threshold value, or with an importance degree coefficient with the KQI of the video service smaller than a third threshold value from the plurality of variables, and taking the removed variable as at least one characteristic variable.

It should be noted that: in the apparatus for predicting a KQI of a video service according to the above embodiment, only the above-mentioned division of each functional module is used for illustration when predicting a KQI of a video service, and in practical applications, the above-mentioned function allocation may be completed by different functional modules according to needs, that is, the internal structure of the apparatus for predicting a KQI of a video service is divided into different functional modules, so as to complete all or part of the above-mentioned functions. In addition, the apparatus for predicting a KQI of a video service and the method embodiment for predicting a KQI of a video service provided in the above embodiments belong to the same concept, and specific implementation processes thereof are detailed in the method embodiment and are not described herein again.

In the above embodiments, the implementation may be wholly or partly realized by software, hardware, firmware, or any combination thereof. When implemented in software, may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When loaded and executed on a computer, cause the processes or functions described in accordance with the embodiments of the application to occur, in whole or in part. The computer may be a general purpose computer, a special purpose computer, a network of computers, or other programmable device. The computer instructions may be stored on a computer readable storage medium or transmitted from one computer readable storage medium to another, for example, from one website, computer, server, or data center to another website, computer, server, or data center via wire (e.g., coaxial cable, fiber optic, Digital Subscriber Line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device, such as a server, a data center, etc., that incorporates one or more of the available media. The usable medium may be a magnetic medium (e.g., floppy Disk, hard Disk, magnetic tape), an optical medium (e.g., Digital Versatile Disk (DVD)), or a semiconductor medium (e.g., Solid State Disk (SSD)), among others.

It will be understood by those skilled in the art that all or part of the steps for implementing the above embodiments may be implemented by hardware, or may be implemented by a program instructing relevant hardware, where the program may be stored in a computer-readable storage medium, and the above-mentioned storage medium may be a read-only memory, a magnetic disk or an optical disk, etc.

The above-mentioned embodiments are provided not to limit the present application, and any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. A method of predicting a key quality indicator of a video service, the method comprising:

acquiring a feature tag table of a target geographic area, wherein the feature tag table comprises at least one feature tag and a tag value of each feature tag, and the at least one feature tag is used for describing video services in a network of the target geographic area;

searching a KQI model matched with the feature tag table of the target geographic region from a key quality index KQI model library according to the feature tag table of the target geographic region, wherein the KQI model library comprises a plurality of KQI models, each KQI model corresponds to one feature tag table, and each KQI model is used for predicting the KQI of the video service;

and predicting the KQI of the video service of the target geographic area according to the searched KQI model.

2. The method as claimed in claim 1, wherein said searching the KQI model matching the signature table of the target geographical area from the KQI model library according to the signature table of the target geographical area comprises:

and searching a KQI model matched with all the feature tags in the feature tag table of the target geographic region from the KQI model library, and taking the searched KQI model as the KQI model matched with the feature tag table of the target geographic region.

3. The method of claim 2, wherein the at least one feature tag comprises a first class of feature tags and a second class of feature tags, each feature tag in the list of feature tags for the target geographic region comprises at least one tag value, and each feature tag in the list of feature tags for each KQI model in the library of KQI models comprises a tag value;

the searching, from the KQI model library, a KQI model that matches all the feature tags in the feature tag table of the target geographic region includes:

for any feature tag A in the feature tag table of the target geographic area, if the feature tag A of the target geographic area belongs to a first class of feature tags, searching a KQI model with the same tag value of the feature tag A in the KQI model as any tag value in at least one tag value of the feature tag A of the target geographic area from the KQI model library;

and if the characteristic label A of the target geographic region belongs to a second type of characteristic label, searching a KQI model with the label value of the characteristic label A in the KQI model similar to any one of at least one label value of the characteristic label A of the target geographic region from the KQI model library.

4. The method of claim 3, wherein the first type of feature tag comprises an operator name, a network type, a data source for recording a video receipt of a video service, a video resolution, and Carrier Aggregation (CA) characteristics, and wherein the second type of feature tag comprises a city name, a data source for a video service, a buffering threshold for a video service, and a scene.

5. The method according to any one of claims 1 to 4, wherein the target geographical area is divided into a plurality of grids in advance;

the predicting the KQI of the video service of the target geographic area according to the searched KQI model comprises the following steps:

determining simulation parameters for planning the station address of the target geographic area, wherein the simulation parameters at least comprise the number of base stations to be deployed, the address of each base station in the base stations to be deployed and the base station engineering parameters of each base station;

performing site planning simulation on the target geographic area according to the simulation parameters, and determining a variable value of each characteristic variable in at least one characteristic variable of each grid after simulation, wherein the at least one characteristic variable is a variable influencing the KQI;

and determining the KQI of the video service of each grid according to the variable value of each characteristic variable in the at least one characteristic variable of each grid after simulation and the searched KQI model.

6. The method of claim 5, wherein the located KQI models comprise at least two KQI models:

determining a KQI of the video service of each grid according to the variable value of each of the at least one characteristic variable of each grid after simulation and the searched KQI model, including:

7. The method of any of claims 1 to 4, 6, further comprising:

the method comprises the steps of obtaining a video receipt data source and at least one associated data source, wherein the video receipt data source comprises a plurality of video receipts, each associated data source comprises a plurality of pieces of data, and the at least one associated data source is a data source related to video services;

for any video document B in the multiple video documents, determining data associated with the video document B in each associated data source, and forming the determined data and the video document B into a piece of structured data to obtain multiple pieces of structured data;

analyzing the KQI included in each piece of structured data and the variable value of each characteristic variable in at least one characteristic variable;

determining at least one feature tag of each piece of structured data, classifying the plurality of pieces of structured data according to the at least one feature tag of each piece of structured data to obtain a plurality of data sets, wherein each data set comprises at least one piece of structured data, each data set corresponds to one feature tag table, and the feature tag table corresponding to each data set is determined according to the feature tags of the structured data included in the corresponding data set;

and for any data set C in the plurality of data sets, taking the KQI included in each piece of structured data in the data set C as output, taking the variable value of each feature variable in at least one feature variable included in each piece of structured data as input, training an initialized algorithm model to obtain a KQI model corresponding to the data set C, taking a plurality of obtained KQI models corresponding to the data sets one by one as models in the KQI model library, and taking a feature tag table corresponding to each data set as a feature tag table of the corresponding KQI model.

8. The method of claim 7, wherein the at least one associated data source comprises a session data source, and each piece of data in the session data source comprises a cell id and a session statistics period, and each piece of data is used to describe a cell network characteristic of a cell corresponding to the cell id within the session statistics period;

the determining data associated with the video document B in each associated data source comprises:

and searching data which has the same cell identification as the cell identification in the video bill B and has the video starting time and the video ending time in the video bill B within a session statistics period from the session data source, and determining the searched data as the data associated with the video bill B.

9. The method of claim 7, wherein the at least one associated data source includes a measurement report data source, and each piece of data in the measurement report data source includes a ue id and a reporting time of a measurement report, and each piece of data is used to describe a characteristic of a ue during a network connection process corresponding to the ue id;

searching data, in which a user terminal identifier is the same as a user terminal identifier in the video document B and the reporting time of a measurement report is in a video period of a video service corresponding to the video document B, from the measurement report data source, and determining the searched data as data associated with the video document B, wherein the starting time of the video period is the video starting time in the video document B minus a duration threshold, and the ending time of the video period is the video ending time in the video document B plus a duration threshold.

10. The method of claim 7, wherein the at least one associated data source comprises a cell parameter data source, and each piece of data in the cell parameter data source comprises a cell identifier, and each piece of data is used for describing a physical characteristic of a cell corresponding to the cell identifier;

and searching data with the same cell identification as the cell identification in the video document B from the cell working parameter data source, and determining the searched data as the data associated with the video document B.

11. The method of any of claims 8 to 10, wherein prior to determining the KQI in each piece of structured data and the value of each of the at least one characteristic variable, further comprising:

acquiring a plurality of variables related to KQI of the video service;

and removing variables of which the correlation coefficient with the KQI of the video service is smaller than a first threshold value and/or variables of which the sample distribution variance is smaller than a second threshold value and/or variables of which the importance coefficient with the KQI of the video service is smaller than a third threshold value from the plurality of variables, and taking the removed variables as the at least one characteristic variable.

12. An apparatus for predicting a key quality indicator of a video service, the apparatus comprising:

the system comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is used for acquiring a feature tag table of a target geographic area, the feature tag table comprises at least one feature tag and a tag value of each feature tag, and the at least one feature tag is used for describing video services in a network of the target geographic area;

the searching module is used for searching a KQI model matched with the characteristic tag table of the target geographic region from a key quality index KQI model base according to the characteristic tag table of the target geographic region, wherein the KQI model base comprises a plurality of KQI models, each KQI model corresponds to one characteristic tag table, and each KQI model is used for predicting the KQI of the video service;

and the prediction module is used for predicting the KQI of the video service of the target geographic area according to the searched KQI model.

13. The apparatus of claim 12, wherein the lookup module comprises:

and the searching unit is used for searching a KQI model matched with all the characteristic labels in the characteristic label table of the target geographic region from the KQI model library, and the searched KQI model is used as the KQI model matched with the characteristic label table of the target geographic region.

14. The apparatus of claim 13, wherein the at least one feature tag comprises a first class of feature tags and a second class of feature tags, each feature tag in the list of feature tags for the target geographic region comprises at least one tag value, and each feature tag in the list of feature tags for each KQI model in the library of KQI models comprises one tag value;

the search unit is specifically configured to:

and if the characteristic label A of the target geographic region belongs to a second type of characteristic label, searching a KQI model with the label value of the characteristic label A in the KQI model being similar to any one of at least one label value of the characteristic label A of the target geographic region from the KQI model.

15. The apparatus of claim 14, wherein the first type of feature tag comprises an operator name, a network type, a data source for recording a video receipt for video traffic, a video resolution, and Carrier Aggregation (CA) characteristics, and wherein the second type of feature tag comprises a city name, a data source for video traffic, a buffering threshold for video traffic, and a scene.

16. The apparatus of any one of claims 12 to 15, wherein the target geographic area is pre-divided into a plurality of grids;

the prediction module comprises:

a first determining unit, configured to determine simulation parameters for planning a site of the target geographic area, where the simulation parameters at least include the number of base stations to be deployed, an address of each base station in the base stations to be deployed, and a base station engineering parameter of each base station;

a second determining unit, configured to perform site planning simulation on the target geographic area according to the simulation parameters, and determine a variable value of each of at least one feature variable of each grid after the simulation, where the at least one feature variable is a variable affecting the KQI;

and a third determining unit, configured to determine a KQI of the video service of each grid according to the variable value of each of the at least one feature variable of each grid after simulation and the searched KQI model.

17. The apparatus of claim 16, wherein the located KQI models comprise at least two KQI models:

the third determining unit is specifically configured to:

18. The apparatus of any of claims 12 to 15, 17, further comprising:

the second acquisition module is used for acquiring a video receipt data source and at least one associated data source, wherein the video receipt data source comprises a plurality of video receipts, each associated data source comprises a plurality of pieces of data, and the at least one associated data source is a data source related to video services;

the first determining module is used for determining data related to the video document B in each related data source for any one video document B in the plurality of video documents, and forming a piece of structural data by the determined data and the video document B to obtain a plurality of pieces of structural data;

the analysis module is used for analyzing the KQI included in each piece of structured data and the variable value of each characteristic variable in at least one characteristic variable;

a second determining module, configured to determine at least one feature tag of each piece of structured data, and classify the pieces of structured data according to the at least one feature tag of each piece of structured data to obtain a plurality of data sets, where each data set includes at least one piece of structured data, and each data set corresponds to one feature tag table, and the feature tag table corresponding to each data set is determined according to the feature tag of the piece of structured data included in the corresponding data set;

and the training module is used for taking the KQI included in each piece of structured data in the data set C as output and the variable value of each feature variable in at least one feature variable included in each piece of structured data as input for any one of the data sets C, training the initialized algorithm model to obtain a KQI model corresponding to the data set C, taking the obtained multiple KQI models corresponding to the data sets one by one as models in the KQI model library, and taking the feature tag table corresponding to each data set as the feature tag table of the corresponding KQI model.

19. The apparatus of claim 18, wherein the at least one associated data source comprises a session data source, and each piece of data in the session data source comprises a cell id and a session statistics period, and each piece of data is used to describe a cell network characteristic of a cell corresponding to the cell id within the session statistics period;

the first determining module is specifically configured to:

20. The apparatus of claim 18, wherein the at least one associated data source comprises a measurement report data source, and each piece of data in the measurement report data source comprises a ue id and a reporting time of a measurement report, and each piece of data is used for describing a characteristic of a ue during a network connection process corresponding to the ue id;

the first determining module is specifically configured to:

21. The apparatus of claim 18, wherein the at least one associated data source comprises a cell parameter data source, and each piece of data in the cell parameter data source comprises a cell identifier, each piece of data describing physical characteristics of a cell corresponding to the cell identifier;

the first determining module is specifically configured to:

22. The apparatus of any of claims 19 to 21, wherein the apparatus further comprises:

and the rejecting module is used for rejecting variables of which the correlation coefficient with the KQI of the video service is smaller than a first threshold value, and/or variables of which the sample distribution variance is smaller than a second threshold value, and/or variables of which the importance degree coefficient with the KQI of the video service is smaller than a third threshold value from the plurality of variables, and taking the rejected variables as the at least one characteristic variable.

23. An apparatus for predicting a key quality indicator of a video service, the apparatus comprising:

a processor and a memory;

the memory is used for storing a program for implementing the method of any one of claims 1 to 11 and for storing data involved in implementing the method of any one of claims 1 to 11, and the processor is configured for executing the program stored in the memory.

24. A computer-readable storage medium having stored therein instructions which, when executed on a computer, cause the computer to perform the method of any one of claims 1 to 11.