CN113497717A

CN113497717A - Network flow prediction method, device, equipment and storage medium

Info

Publication number: CN113497717A
Application number: CN202010194429.4A
Authority: CN
Inventors: 邱亚星; 边森; 王希栋
Original assignee: China Mobile Communications Group Co Ltd; China Mobile Communications Ltd Research Institute
Current assignee: China Mobile Communications Group Co Ltd; China Mobile Communications Ltd Research Institute
Priority date: 2020-03-19
Filing date: 2020-03-19
Publication date: 2021-10-12
Anticipated expiration: 2040-03-19
Also published as: CN113497717B

Abstract

The invention discloses a method, a device, equipment and a storage medium for predicting network flow, wherein the method comprises the following steps: acquiring historical service related data of a plurality of user equipment; extracting the characteristics of the historical service related data to obtain service related characteristics corresponding to each user equipment; based on the similarity of service related characteristics between every two pieces of user equipment in the plurality of pieces of user equipment, clustering and dividing the plurality of pieces of user equipment to obtain at least two user equipment groups; taking historical position data and historical network traffic data of user equipment in each user equipment group as training sample data, and respectively carrying out prediction model training on the position and the network traffic of the corresponding user equipment group to obtain a prediction model corresponding to the position and the network traffic of each user equipment group; the prediction model is used for predicting a target position and network traffic of the target user equipment group, and the target position and the network traffic are used for determining a total network traffic predicted value corresponding to the target position.

Description

Network flow prediction method, device, equipment and storage medium

Technical Field

The present invention relates to the field of wireless communications, and in particular, to a method, an apparatus, a device, and a storage medium for predicting network traffic.

Background

With the development of the internet, the number of network users is gradually increased, and in order to ensure the network security and the perception of the users, the trend of dynamic change of the network traffic is mastered by predicting the network traffic of the whole target position of the access network, such as the whole cell, so that the network structure and the bandwidth are continuously adjusted, and the method has important significance in the actual network management application.

Currently, in the related art, when predicting the network traffic of the whole cell, a method of analyzing and predicting the network traffic based on the network or the traffic data of the cell itself, or a method of predicting the network traffic based on the mobility analysis of a single user in the cell is generally adopted. However, the two prediction methods still have certain limitations, which results in the accuracy of predicting the network traffic of the whole cell being reduced.

Disclosure of Invention

In order to solve the technical problems in the related art, embodiments of the present invention provide a method, an apparatus, a device, and a storage medium for predicting network traffic, which can effectively improve the accuracy of predicting the network traffic of the entire target location.

The technical scheme of the embodiment of the invention is realized as follows:

the embodiment of the invention provides a method for predicting network flow, which comprises the following steps:

acquiring historical service related data of a plurality of user equipment;

extracting the characteristics of the historical service related data to obtain service related characteristics corresponding to each user equipment;

based on the similarity of service related features between every two pieces of user equipment in the plurality of pieces of user equipment, clustering and dividing the plurality of pieces of user equipment to obtain at least two user equipment groups;

taking historical position data and historical network traffic data of user equipment in each user equipment group as training sample data, and respectively carrying out position and network traffic prediction model training on the corresponding user equipment group to obtain a prediction model corresponding to the position and network traffic of each user equipment group;

the prediction model is used for predicting a target position and network traffic of a target user equipment group, and the target position and the network traffic are used for determining a total network traffic predicted value corresponding to the target position.

In the above scheme, the method further comprises:

and determining similarity of the service related characteristics between every two user equipment in the plurality of user equipment based on the service related characteristics of the user equipment.

In the foregoing solution, the determining a similarity of service-related features between every two pieces of user equipment in the multiple pieces of user equipment based on the service-related features of the user equipment includes:

respectively determining the motion track similarity between every two user equipments in the plurality of user equipments under the condition that the service related characteristics comprise the motion track, the stay area and the service flow, and

determining a stay area similarity between each two of the plurality of user equipments, an

And determining the similarity of the service flow between every two user equipments in the plurality of user equipments.

In the foregoing solution, the determining the motion trajectory similarity between every two pieces of user equipment in the multiple pieces of user equipment includes:

acquiring a first motion track corresponding to a first user device and a second motion track corresponding to a second user device in two user devices in the same time period;

determining the distance between the track points of the first motion track and the corresponding serial numbers in the second motion track;

and determining the similarity of the motion tracks between the two user devices based on the summation of the distances between the track points of the corresponding numbers in the first motion track and the second motion track.

In the foregoing solution, the determining the similarity of the staying areas between every two pieces of the user equipment includes:

acquiring a first staying area corresponding to a first user device and a second staying area corresponding to a second user device in two user devices in the same time period;

determining the number of stop points in the intersection area of the first stop area and the second stop area;

determining a number of total dwell points in the first dwell area and the second dwell area;

determining dwell region similarity between the two user devices based on a ratio of the number of dwell points within the intersection region to the total number of dwell points.

In the foregoing solution, the determining the traffic similarity between every two pieces of user equipment in the multiple pieces of user equipment includes:

acquiring a first service class corresponding to a first user equipment and a second service class corresponding to a second user equipment in two user equipments in the same time period;

determining a first service flow corresponding to the first service class and a second service flow corresponding to the second service class;

and determining the similarity of the service flow between the two user equipment based on the difference value between the first service flow and the second service flow.

In the foregoing solution, the clustering and partitioning the multiple pieces of user equipment based on the similarity of the service-related features between every two pieces of user equipment in the multiple pieces of user equipment to obtain at least two user equipment groups includes:

determining a time-space distribution similarity between each two user equipments in the plurality of user equipments based on the similarity of the service-related features between each two user equipments in the plurality of user equipments;

constructing a similarity matrix based on matrix representation of space-time distribution similarity between every two user equipment in the plurality of user equipment;

and establishing a graph model based on the similarity matrix, and clustering and dividing the plurality of user equipment based on the weight of the excess edges in the graph model to obtain at least two user equipment groups.

In the above scheme, the method further comprises: normalizing the similarity matrix to obtain a normalized similarity matrix;

the establishing of the graph model based on the similarity matrix comprises the following steps:

and carrying out graph conversion on the normalized similarity matrix to obtain the graph model.

In the above scheme, the method further comprises:

acquiring real-time service related data of the target user equipment group at the current moment, wherein the real-time service related data at least comprises real-time position data and real-time network flow data;

respectively inputting the real-time position data and the real-time network traffic data to the prediction model to obtain a target position of the target user equipment group at a target moment and a network traffic prediction value of the target user equipment group at the target moment, wherein the target position is output by the prediction model; the target time is used for representing any time taking the current time as the starting time.

In the above scheme, the method further comprises:

determining a target user equipment group set positioned at the same target position based on the target position of the target user equipment group at the target moment;

and summarizing the network traffic predicted values of all target user equipment groups in the target user equipment group set at the same target position to obtain a total network traffic predicted value corresponding to the target position.

The embodiment of the invention also provides a method for predicting network flow, which comprises the following steps:

acquiring real-time service related data of a target user equipment group at the current moment, wherein the real-time service related data at least comprises real-time position data and real-time network flow data;

respectively inputting the real-time position data and the real-time network traffic data into a prediction model to obtain a target position of the target user equipment group at a target moment and a network traffic prediction value of the target user equipment group at the target moment, wherein the target position is output by the prediction model; the target time is used for representing any time taking the current time as the starting time;

determining a total network traffic predicted value of the target position based on the target position of the target user equipment group at the target time and the network traffic predicted value of the target user equipment group at the target time;

the prediction model is obtained by performing model training based on historical position data and historical network traffic data of user equipment in the user equipment group as training sample data.

In the foregoing solution, the determining a total network traffic predicted value of the target location based on the target location of the target ue group at the target time and the network traffic predicted value of the target ue group at the target time includes:

In the above scheme, the method further comprises:

comparing the total network traffic predicted value of the target position with the total network traffic threshold value of the target position to obtain a comparison result;

and when the comparison result represents that the total network flow predicted value of the target position is greater than the total network flow threshold value, maintaining and optimizing the network equipment of the target position.

The embodiment of the invention also provides a device for predicting network flow, which comprises:

a first obtaining unit, configured to obtain historical service related data of a plurality of user equipments;

a feature extraction unit, configured to perform feature extraction on the historical service-related data to obtain service-related features corresponding to each piece of the user equipment;

a cluster division unit, configured to perform cluster division on the multiple pieces of user equipment based on similarity of service-related features between every two pieces of user equipment in the multiple pieces of user equipment, so as to obtain at least two user equipment groups;

the model training unit is used for respectively carrying out prediction model training on the position and the network flow of the corresponding user equipment group by taking historical position data and historical network flow data of the user equipment in each user equipment group as training sample data to obtain a prediction model corresponding to the position and the network flow of each user equipment group;

In the above scheme, the apparatus further comprises:

a first determining unit, configured to determine, based on the service-related features of the user equipments, a similarity of the service-related features between every two user equipments in the multiple user equipments.

In the foregoing solution, the first determining unit is specifically configured to:

In the foregoing solution, the cluster dividing unit is specifically configured to:

In the above scheme, the apparatus further comprises: the normalization unit is used for carrying out normalization processing on the similarity matrix to obtain a normalized similarity matrix;

the cluster partitioning unit is specifically configured to:

In the above scheme, the apparatus further comprises:

a second obtaining unit, configured to obtain real-time service related data of the target user equipment group at a current time, where the real-time service related data at least includes real-time location data and real-time network traffic data;

a data input unit, configured to input the real-time location data and the real-time network traffic data to the prediction model, respectively, so as to obtain a target location of the target user equipment group at a target time and a network traffic prediction value of the target user equipment group at the target time, where the target location and the network traffic data are output by the prediction model; the target time is used for representing any time taking the current time as the starting time.

In the above scheme, the apparatus further comprises:

a second determining unit, configured to determine, based on a target location of the target user equipment group at the target time, a target user equipment group set located at the same target location;

and the summarizing unit is used for summarizing the network traffic predicted values of all target user equipment groups in the target user equipment group set at the same target position to obtain a total network traffic predicted value corresponding to the target position.

a second obtaining unit, configured to obtain real-time service related data of a target user equipment group at a current time, where the real-time service related data at least includes real-time location data and real-time network traffic data;

a data input unit, configured to input the real-time location data and the real-time network traffic data to a prediction model, respectively, so as to obtain a target location of the target user equipment group at a target time and a network traffic prediction value of the target user equipment group at the target time, where the target location and the network traffic data are output by the prediction model; the target time is used for representing any time taking the current time as the starting time;

a third determining unit, configured to determine a total network traffic predicted value of the target location based on a target location of the target user equipment group at a target time and a network traffic predicted value of the target user equipment group at the target time;

In the foregoing solution, the third determining unit is specifically configured to:

In the above scheme, the apparatus further comprises:

the comparison unit is used for comparing the total network traffic predicted value of the target position with the total network traffic threshold value of the target position to obtain a comparison result;

and the management unit is used for maintaining and optimizing the network equipment at the target position when the comparison result represents that the total network flow predicted value of the target position is greater than the total network flow threshold value.

An embodiment of the present invention further provides a device for predicting network traffic, including: a first processor and a first memory for storing a computer program operable on the processor;

wherein the first processor is configured to execute the steps of any one of the above methods of the model training side when running the computer program.

An embodiment of the present invention further provides a device for predicting network traffic, including: a second processor and a second memory for storing a computer program operable on the processor;

wherein the second processor is configured to execute the steps of any one of the above methods of the model application side when running the computer program.

An embodiment of the present invention further provides a storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the steps of any one of the methods on the model training side or the steps of any one of the methods on the model application side.

The method, the device, the equipment and the storage medium for predicting the network traffic provided by the embodiment of the invention perform feature extraction on historical service related data of a plurality of user equipment to obtain service related features corresponding to each user equipment, then perform cluster division on the plurality of user equipment based on the similarity of the service related features between every two user equipment in the plurality of user equipment to obtain at least two user equipment groups, and perform prediction model training of the position and the network traffic aiming at each divided user equipment group to obtain a prediction model corresponding to the position and the network traffic of each user equipment group, wherein the prediction model is used for predicting the target position and the network traffic of the target user equipment group. Therefore, based on the consideration of the influence of the user equipment on the network flow change, the similarity of service related characteristics among different user equipment is fully mined, the division of user equipment groups is realized by means of the similarity, the position and network flow association modeling is respectively carried out on each divided user equipment group, further the total network flow corresponding to the target position can be determined, and the accuracy of predicting the network flow of the whole target position can be effectively improved.

Drawings

Fig. 1 is a schematic flowchart of a method for predicting network traffic according to an embodiment of the present invention;

fig. 2 is a schematic diagram of a structure of similarity between service-related features according to an embodiment of the present invention;

FIG. 3 is a schematic diagram of a motion trajectory provided by an embodiment of the present invention;

FIG. 4 is a schematic view of a dwell area provided by an embodiment of the present invention;

fig. 5 is a schematic flowchart illustrating a location and network traffic prediction performed for a ue group according to an embodiment of the present invention;

fig. 6 is a flowchart illustrating another method for predicting network traffic according to an embodiment of the present invention;

fig. 7 is a flowchart illustrating another method for predicting network traffic according to an embodiment of the present invention;

fig. 8 is a schematic structural diagram of a device for predicting network traffic according to an embodiment of the present invention;

fig. 9 is a schematic structural diagram of another network traffic prediction apparatus according to an embodiment of the present invention;

fig. 10 is a schematic structural diagram of a device for predicting network traffic according to an embodiment of the present invention;

fig. 11 is a schematic structural diagram of another network traffic prediction device according to an embodiment of the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the present invention will be further described in detail with reference to the accompanying drawings, the described embodiments should not be construed as limiting the present invention, and all other embodiments obtained by a person of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

In the following description, reference is made to "some embodiments" which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and that the technical solutions described in the embodiments of the present invention may be combined with each other without conflict.

The following analyzes the scheme provided by the related art regarding predicting the network traffic of the target location.

In some solutions of the related art, when predicting network traffic of a target location, a method of analyzing and predicting based on network or cell traffic data itself is adopted, and the method mainly includes:

1. the network traffic prediction based on the traditional algorithm model is realized by using a Markov (Markov) model or an autoregressive Moving Average (ARMA) model, for example. However, in the process of predicting the network traffic by using the Markov model, the future trend of the network traffic is only related to the current state and is unrelated to the past state, and the prediction accuracy cannot be guaranteed; the ARMA model belongs to a linear model, and along with the increase of the complexity of network flow, the network flow is often a non-stable chaotic time sequence, and at the moment, the linear model similar to the ARMA model cannot ensure the prediction accuracy.

2. The prediction method comprises the steps of mining and modeling historical network flow data of a network or a cell, learning a neural network model, and predicting the network flow of the next time granularity by using the obtained neural network model.

3. The network traffic prediction based on the time sequence comprises a network traffic prediction method based on the traditional time sequence, for example, a method for realizing the prediction of network traffic based on a Long Short-Term Memory network (LSTM), and a method for dividing the network traffic prediction into two different angles of working days and rest days and the like on the basis of considering the relevance between the network traffic characteristics and social factors.

However, in the above-mentioned several implementation methods based on the analysis and prediction of the network or cell own traffic data, there is no problem that the influence of the behavior rule of the mobile user on the network traffic is analyzed from the user perspective, and the prediction accuracy of the network traffic is low. This is because the change of the network traffic is mainly determined by the mobile user, and if the influence of the service-related characteristics of the mobile user on the network traffic is not considered, it is difficult to ensure the accuracy of predicting the network traffic.

In other solutions of related technologies, when predicting a target location, such as a network traffic of a target cell, a method for predicting the network traffic based on mobility analysis of a single user in the target cell is adopted, and the method mainly determines an influence of the single user on the network traffic of the target cell by analyzing data of the single user, such as inflow, outflow, residence, traffic and the like.

Although theoretically, the method can improve the accuracy of the prediction of the network traffic, the method still has limitations in practical application: on one hand, the application limitation of the prediction model is realized, and because each user needs to be modeled and analyzed respectively, when the number of the users is high, a large amount of computing resources and computing time need to be consumed, and meanwhile, the generalization capability of the prediction model is poor; on the other hand, the method is not applicable to the situation that mobility data of a single user is lost due to the limitation of a prediction scene or a data set; on the other hand, the method does not fully consider and mine the influence of the relationship between the users on the network traffic, so that the accuracy of predicting the network traffic cannot be guaranteed.

Based on this, in various embodiments of the present invention, analysis and modeling of a user equipment group is implemented based on similarity of traffic-related features between every two user equipments in a plurality of user equipments, so as to predict total network traffic of a target location based on group identification.

By adopting the scheme of the embodiment of the invention, based on the consideration of the influence of the user equipment on the network flow change, the similarity of the service related characteristics among different user equipment is fully excavated, the division of the user equipment groups is realized by means of the similarity, and the correlation modeling of the position and the network flow is respectively carried out aiming at each divided user equipment group, so that the total network flow corresponding to the target position can be determined, and the accuracy of predicting the network flow of the whole target position can be effectively improved.

The following describes implementation of the method for predicting network traffic provided by the embodiment of the present invention.

Referring to fig. 1, fig. 1 is a flowchart illustrating a method for predicting network traffic according to an embodiment of the present invention, where the method for predicting network traffic may be implemented by a prediction device of network traffic, and the prediction device of network traffic according to the embodiment of the present invention may be implemented as various types of terminal devices, such as a notebook computer, a desktop computer, and the like, or may be implemented as a network management server, such as a cloud server for network management or a local server for network management. The following describes a method for predicting network traffic according to an embodiment of the present invention, with reference to the steps shown in fig. 1, by taking a network management server as an example.

Step 101, obtaining historical service related data of a plurality of user equipments.

Here, the plurality of user equipments may be mobile equipments respectively held by a plurality of mobile users accessing the network, where the plurality of mobile users include two or more mobile users and the plurality of user equipments include two or more user equipments.

In practical application, the mobile device may be various types of portable devices such as a smart phone, a tablet computer, and a notebook computer used by a mobile user when the mobile user meets a service requirement. Historical service-related data, including data generated during a historical period of time to fulfill a service requirement, i.e., historical service-related data associated with services selected by a mobile subscriber during the historical period of time.

In some embodiments, the network management server may obtain historical traffic related data for a plurality of user equipments by: and acquiring historical service related data sent by the plurality of user equipment, wherein the historical service related data is acquired by calling an acquisition device of the plurality of user equipment.

Specifically, the method may first acquire historical service related data of the mobile user by calling an acquisition device of the user equipment, that is, acquire historical service related data of a plurality of user equipments, and then send the acquired historical service related data to the network management server through the network, so that the network management server may acquire the historical service related data of the plurality of user equipments. The network management server is connected with the user equipment through a network, the network can be a wide area network or a local area network, or the combination of the wide area network and the local area network, and data transmission is realized by using a wireless link.

Here, in practical application, the historical service related data obtained by the network management server may be data associated with a service selected by the mobile subscriber according to a time dimension. The historical traffic related data comprises at least historical location data and historical traffic data. The historical position data is used for representing the position information of the user equipment passing through in a historical time period, and the historical position data of a plurality of user equipment can form a historical position data list; the historical service data is used for representing service information adopted by the user equipment in a historical time period, and the historical service data of the plurality of user equipment can form a historical service data list.

Table 1 is a historical location data list, and location information of each ue passing through in a certain historical time period can be obtained through each piece of information in table 1, as shown in table 1, each piece of information may include a ue ID, a start time, an end time, a cell ID, a cell name, a base station location, and latitude and longitude information (GPS, optional):

TABLE 1

Table 2 is a historical service data list, and information such as a service type (including a service major class and a service minor class) and a corresponding service traffic (including an uplink traffic and a downlink traffic) used by each user equipment in a certain historical time period can be obtained through each piece of information in table 2, as shown in table 2, each piece of information may include user equipment ID, start time, end time, a service major class, a service minor class, a service traffic, and service longitude and latitude information (service GPS, optional):

TABLE 2

And 102, extracting the characteristics of the historical service related data to obtain the service related characteristics corresponding to each user equipment.

Here, in practical application, the service-related features include a motion trajectory, a stay area, and a service flow.

Specifically, the motion trajectory may be formed by combining trace points that the user equipment passes through in a certain historical time period, where the trace points may be represented by a cell ID or a cell name that the user equipment arrives in table 1 above; the dwell time is generally greater than a time threshold T_dIs defined as the stay area of the user equipment, i.e. the stay area of the user equipment comprises a plurality of stay points, the available stay time of the stay points is larger than the time threshold T_dA location of; the service traffic here includes uplink traffic and downlink traffic generated by the user equipment in the historical time period to realize the service requirement.

And 103, clustering and dividing the plurality of user equipment based on the similarity of the service related characteristics between every two user equipment in the plurality of user equipment to obtain at least two user equipment groups.

In practical application, the network management server further needs to determine similarity of service-related features between every two user equipments in the plurality of user equipments.

Based on this, in some embodiments, before the network management server performs step 103, the method further comprises: and determining the similarity of the service related characteristics between every two user equipment in the plurality of user equipment based on the service related characteristics of the user equipment.

Here, in actual application, the similarity of the service-related features between every two pieces of user equipment in the plurality of pieces of user equipment is determined based on the service-related features of the pieces of user equipment, and may be implemented by:

And determining the similarity of the service flow between every two user equipment in the plurality of user equipment.

Referring to fig. 2, fig. 2 is a schematic diagram illustrating a structure of similarity between service-related features according to an embodiment of the present invention, where the similarity between service-related features includes a motion trajectory similarity, a stay area similarity, and a service traffic similarity.

Specifically, the network management server may determine a motion trajectory similarity between two user equipments in the plurality of user equipments based on a motion trajectory between the two user equipments; determining a stay area similarity between two user equipment in the plurality of user equipment based on a stay area between the two user equipment; and determining the similarity of the service flow between the user equipment in the plurality of user equipment based on the service flow between the user equipment in pairs.

The following describes the determination methods of the motion trajectory similarity, the stay area similarity, and the traffic flow similarity, respectively.

In some embodiments, determining the motion trajectory similarity between each two of the plurality of user devices comprises:

determining the distance between track points of corresponding numbers in the first motion track and the second motion track;

Here, in actual application, the distance between the track points corresponding to the numbers may be an euclidean distance, and after determining the euclidean distance between the track points corresponding to the numbers in the first motion trajectory and the second motion trajectory, summing the euclidean distances to obtain a summed distance, and determining the reciprocal of the summed distance as the motion trajectory similarity between the two user devices.

For example, referring to fig. 3, fig. 3 is a schematic diagram of a motion trajectory according to an embodiment of the present invention, where fig. 3 includes two motion trajectories, which are a first motion trajectory of a user equipment a and a second motion trajectory of a user equipment B, respectively, the motion trajectories of the user equipment a and the user equipment B in a certain time period may be represented by network nodes, and the first motion trajectory corresponding to the user equipment a may be represented as a set la { la ═ la₁,la₂,...,la_nThe second motion track corresponding to the user equipment B may be represented as a set lb ═ lb₁,lb₂,...,lb_nWhere n denotes the number of trace points, la_nRepresenting the points of track, lb, traversed by the user equipment a_nRepresenting the points of track traversed by user device B.

In practical application, the motion trajectory similarity between the user equipment a and the user equipment B can be determined by the following formula:

wherein sim (la, lb) represents the motion track similarity between the user equipment A and the user equipment B; d (la)_i,lb_i) And the distance between the track points of the corresponding numbers in the first motion track and the second motion track is represented, and specifically, the distance may be an euclidean distance. After Euclidean distances between track points of corresponding numbers in a first motion track of the user equipment A and a second motion track of the user equipment B are calculated, all the obtained Euclidean distances are summed to obtain a summed distance, and the reciprocal of the summed distance is determined as the motion track similarity between the user equipment A and the user equipment B.

In some embodiments, determining the stay area similarity between each two of the plurality of user devices comprises:

determining the number of the stop points in the intersection area of the first stop area and the second stop area;

determining the number of total stay points in the first stay area and the second stay area;

determining the stay region similarity between the two user equipments based on the ratio of the number of stay points in the intersection region to the total number of stay points.

For example, stay areas of the user equipment a and the user equipment B in a certain time period may be represented by the following set, and a first stay area corresponding to the user equipment a may be represented by the set ta ═ ta₁,ta₂,...,ta_pThe second stay area corresponding to the user equipment B can be expressed as a set tb ═ tb₁,tb₂,...,tb_qP represents the number of stay points of the user equipment A in the first stay area; q represents the number of stay points of the user equipment B in the second stay area; ta is_pA dwell point representing user equipment a in the first dwell area; tb_qIndicating the dwell point of user equipment B in the second dwell area.

Referring to fig. 4, fig. 4 is a schematic diagram of a stay area according to an embodiment of the present invention, where fig. 4 includes two stay areas, which are a first stay area ta corresponding to a user equipment a and a second stay area tb corresponding to a user equipment B, and in actual application, a stay area similarity between the user equipment a and the user equipment B may be determined through the following formula:

where sim (ta, tb) represents the stay area similarity between user equipment a and user equipment B; ta represents a first stay area corresponding to the user equipment a; tb denotes a second stay area corresponding to the user equipment B.

In actual application, the number of the stay points in the intersection area of the first stay area ta corresponding to the user equipment a and the second stay area tb corresponding to the user equipment B and the number of the total stay points in the first stay area ta and the second stay area tb are determined, and then the stay area similarity between the user equipment a and the user equipment B can be determined based on the ratio of the number of the stay points in the intersection area to the number of the total stay points.

In some embodiments, determining the traffic flow similarity between every two user equipments in the plurality of user equipments includes:

Here, in practical application, after the first service traffic and the second service traffic are determined, the difference values of the service traffic corresponding to each service class may be determined first, then the difference values are summed to obtain a total difference value, and a reciprocal of the total difference value is determined as a service traffic similarity between two user equipments.

For example, the service categories selected by the user equipment a and the user equipment B in a certain time period may be represented by the following set, and the first service category corresponding to the user equipment a may be represented by the set sa ═ sa₁,sa₂,...,sa_kThe second service class corresponding to the user equipment B may be represented as a set sb ═ sb₁,sb₂,...,sb_kThe first traffic flow corresponding to the first traffic class may be represented as a set fa ═ fa₁,fa₂,...,fa_kAnd a second traffic flow corresponding to the second traffic class may be represented as a set fb ═ fb₁,fb₂,...,fb_kIn which fa is_kFor the traffic class sa_kCorresponding traffic flow, fb_kFor the traffic class sb_kAnd corresponding traffic flow.

In practical application, the service flow similarity between the user equipment a and the user equipment B can be determined by the following formula:

wherein sim (fa, fb) represents the traffic similarity between user equipment a and user equipment B, and is used for measuring the difference of the used traffic between user equipment a and user equipment B; fa (fa)_iRepresenting the service flow adopted by the user equipment A; fb_iIndicating the traffic volume employed by user equipment B.

Table 3 is a historical service data list of the user equipment a, as follows:

TABLE 3

Table 4 is a historical service data list of the user equipment B, as follows:

starting time	End time	Business subclass	Traffic flow
				2017-05-25-0720	2017-05-25-0721	WeChat	10k
2017-05-25-0725	2017-05-25-0726	Ink weather	4k
				2017-05-25-0730	2017-05-25-0731	WeChat	6k
2017-05-25-0735	2017-05-25-0736	Sina microblog	10k
				2017-05-25-0740	2017-05-25-0741	Taobao (treasure made of Chinese herbal medicine)	20k

TABLE 4

As can be seen from the data in tables 3 and 4, the first traffic flow of the user equipment a can be expressed as set fa ═ {10,5,8,10,15}, and the second traffic flow of the user equipment B can be expressed as set fb ═ {10,4,6,10,20}, and then the data in tables 3 and 4 show that the first traffic flow of the user equipment a can be expressed as set fb ═ 10,5,8,10,15}, respectively

It can be determined that the traffic similarity between the user equipment a and the user equipment B is 1/8.

It should be noted that, in practical application, the motion trajectory similarity, the stay area similarity, and the traffic flow similarity may also be calculated by using formulas in other forms, and are not limited to the above calculation formulas, and the embodiment of the present invention does not limit this.

In practical application, since the service-related features extracted from the features have the characteristics of high dimensionality, complexity and sparsity, for example, the motion trajectory has a very high dimensionality, and the service flow has a relatively complex attribute, if the clustering algorithm is directly applied to cluster and divide a plurality of user equipments, huge computing resources and computing time are consumed, and meanwhile, redundant data can also seriously affect the clustering effect, thereby reducing the clustering accuracy.

Based on this, in some embodiments, the network management server may first determine similarity of service-related features between every two user equipments in the multiple user equipments, and then perform cluster division on the multiple user equipments based on the similarity of the service-related features between every two user equipments in the multiple user equipments, that is, the motion trajectory similarity, the stay area similarity, and the service traffic similarity between every two user equipments in the multiple user equipments, so as to implement division of different user equipment groups. The method for realizing the division of the user equipment group by integrating the similarity of the relevant characteristics of various services is beneficial to reducing the data dimension and the calculated amount, thereby saving the calculation resource and the calculation time, and improving the clustering precision to solve the clustering problem of high-dimension and complex data types.

In some embodiments, the clustering the plurality of user equipments based on the similarity of the service related features between every two user equipments in the plurality of user equipments to obtain at least two user equipment groups includes:

determining the similarity of space-time distribution between every two pieces of user equipment in the plurality of pieces of user equipment based on the similarity of service related characteristics between every two pieces of user equipment in the plurality of pieces of user equipment;

Here, in the case where the service-related features include a motion trajectory, a stay area, and a service traffic, the spatial-temporal distribution similarity between each two user equipments among the plurality of user equipments may be determined based on a motion trajectory similarity between each two user equipments among the plurality of user equipments, a stay area similarity, and a square of a service traffic similarity. In the embodiment of the present invention, the spatial-temporal distribution similarity is used to represent the similarity of each two user equipments in the plurality of user equipments in time and space, respectively.

Still taking the user equipment a and the user equipment B as an example, the motion trajectory similarity between the two user equipments is sim (la, lb), the stay area similarity between the two user equipments is sim (ta, tb), and the traffic flow similarity between the two user equipments is sim (fa, fb), in practical application, after determining sim (la, lb), sim (ta, tb), and sim (fa, fb), the space-time distribution similarity between the user equipment a and the user equipment B can be determined by the following formula:

where Sim (a, B) represents the similarity of space-time distribution between user equipment a and user equipment B, and may also be expressed as Sim_AB. The space-time distribution similarity between the user equipment A and the user equipment B can be determined by calculating the square sum of the motion trajectory similarity, the stay area similarity and the service flow similarity and then opening the root of the value corresponding to the square sum.

Here, in practical application, after the space-time distribution similarity between two user equipments is determined by the above formula, a similarity matrix may be constructed based on the space-time distribution similarity between all different user equipments, the similarity matrix may be expressed as an SIM, and the expression form of the similarity matrix is:

wherein, Sim_mnRepresenting the spatio-temporal distribution similarity between user equipment m and user equipment n.

In practical application, in order to reduce the calculation load of subsequent network management server processing, the similarity matrix may be normalized, that is, the similarity matrix is normalized to a range from 0 to 1, so as to obtain a normalized similarity matrix.

Based on this, in some embodiments, the method further comprises: and carrying out normalization processing on the similarity matrix to obtain the normalized similarity matrix. Accordingly, a graph model is built based on the similarity matrix, comprising: and carrying out graph conversion on the normalized similarity matrix to obtain a graph model.

Here, a graph model is established based on the similarity matrix, and then clustering division of a plurality of user equipments is realized by segmenting the graph model, so that high-dimensional data and data with complex attributes can be effectively processed. In practical application, the relation of the similarity matrix of the high-dimensional space is converted into a graph model, the relation between the space points is described by using the weight of the super edges in the graph model, and then the space points contained in the super edges with large weight are placed in one class as much as possible based on a segmentation algorithm of the graph model, so that the clustering division process of a plurality of user equipment is realized by using the segmentation of the graph model, and at least two user equipment groups are finally obtained.

Step 104, taking the historical position data and the historical network flow data of the user equipment in each user equipment group as training sample data, and respectively carrying out position and network flow prediction model training on the corresponding user equipment group to obtain a prediction model corresponding to the position and the network flow of each user equipment group; the prediction model is used for predicting a target position and network traffic of a target user equipment group, and the target position and the network traffic are used for determining a total network traffic predicted value corresponding to the target position.

Here, the historical location data and the historical network traffic data are both associated with historical traffic related data, and specifically, the historical traffic related data includes the historical network traffic data and the historical location data.

In practical application, the prediction model of the position and the network traffic is a joint prediction model, which can predict the target position of the target user equipment group at the target time and also predict the network traffic of the target user equipment group at the target time. Here, the prediction model of the location and the network traffic may be constructed based on a deep learning method, such as a random forest model, a deep neural network model, and the like.

Here, the historical location data with labels of the ue in the ue group and the historical network traffic data with labels of the ue in the ue group may be used as training sample data, and location and network traffic prediction model training is performed on each ue group, that is, association modeling of location and network traffic is performed on each ue group, so as to obtain a location and network traffic prediction model corresponding to the ue group. In practical application, the model parameters of the prediction model of the position and the network flow can be updated based on a preset period.

In practical application, the prediction model of the position and the network flow can be replaced by two prediction models, namely a position prediction model and a flow prediction model, specifically, historical position data and historical network flow data of user equipment in each user equipment group are used as training sample data, and the prediction model training of the position and the network flow is respectively carried out on the corresponding user equipment group to obtain the position prediction model and the flow prediction model corresponding to each user equipment group; the location prediction model is used for predicting a target location of the target user equipment group, and the traffic prediction model is used for predicting network traffic of the target user equipment group.

In some embodiments, the method for obtaining a location prediction model and a traffic prediction model corresponding to each user equipment group by using historical location data and historical network traffic data of the user equipment in each user equipment group as training sample data and performing location and network traffic prediction model training on the corresponding user equipment group includes:

taking the marked historical position data of the user equipment in each user equipment group as first training sample data, and respectively carrying out position prediction model training on each user equipment group to obtain a position prediction model corresponding to each user equipment group;

and taking the marked historical network traffic data of the user equipment in each user equipment group as second training sample data, and respectively carrying out prediction model training of network traffic on each user equipment group to obtain a traffic prediction model corresponding to each user equipment group.

Specifically, in the process of training a prediction model of network traffic, using marked historical network traffic data of user equipment in a user equipment group as second training sample data, inputting the second training sample data into the traffic prediction model, and obtaining the predicted network traffic corresponding to the second training sample data output by the traffic prediction model; determining a value of a loss function of a traffic prediction model based on the output predicted network traffic and historical network traffic data; model parameters of the flow prediction model are updated based on the values of the loss function.

In practical application, a traffic prediction model, such as a random forest model and a deep neural network model, can be pre-constructed based on a deep learning method, the traffic prediction model comprises an input layer, a hidden layer and an output layer and is used for predicting network traffic of a target user equipment group, and after the traffic prediction model is constructed, the traffic prediction model is trained based on collected second training sample data to obtain optimized traffic prediction model parameters.

Before the model is trained, a large number of collected samples can be divided into a training set and a testing set according to a certain proportion, and the historical network traffic with labels of the user equipment in the user equipment group in the training set is used as first training sample data and is input into the traffic prediction model to obtain the predicted network traffic of the target user equipment group output by the traffic prediction model. Furthermore, the model training process is an updating and adjusting process of each parameter in the model, second training sample data is input to an input layer of the flow prediction model, passes through a hidden layer, finally reaches an output layer and outputs a result, because the output result of the flow prediction model and an actual result have errors, the error between the output result and the actual value needs to be calculated, and the error is reversely propagated from the output layer to the hidden layer until being propagated to the input layer, and then the value of the model parameter is adjusted according to the error in the process of reverse propagation; and continuously iterating the steps in the whole training process until convergence so as to reduce the error of the model output.

Based on the above, in actual application, in order to reduce possible errors between the predicted network traffic output by the traffic prediction model and the historical network traffic, a loss function is introduced, and a value of the loss function is determined based on the predicted network traffic output by the traffic prediction model and the historical network traffic; and updating parameters of the flow prediction model layer by using a back propagation algorithm based on the value of the loss function until the loss function is converged to realize the constraint and adjustment of the parameters of the flow prediction model, thereby obtaining the flow prediction model with high calculation precision and determining the network flow of the target user equipment group at the target moment based on the flow prediction model.

It should be noted that the training process of the location prediction model is similar to the training process of the traffic prediction model described above, and is not described here again.

In some embodiments, the method further comprises:

respectively inputting the real-time position data and the real-time network traffic data into a prediction model of the position and the network traffic to obtain a target position of a target user equipment group at a target moment and a network traffic prediction value of the target user equipment group at the target moment, wherein the target position is output by the prediction model; the target time is used for representing any time taking the current time as the starting time.

Referring to fig. 5, fig. 5 is a schematic flow chart of predicting the location and the network traffic for the ue groups according to the embodiment of the present invention, which is illustrated by taking a prediction model as a joint prediction model, that is, a prediction model of the location and the network traffic as an example, and for each ue group, taking the historical time-location data of the ue in the ue group and the historical time-traffic data of the ue in the ue group as training sample data, performing location and network traffic prediction model training for each ue group to obtain a prediction model corresponding to the location and the network traffic of each ue group, and then inputting the real-time-location data and the real-time-traffic data to the prediction model of the location and the network traffic to predict the average location of the ue in the target ue group at a target time, and the average network traffic of the user equipment in the target user equipment group at the target moment. In the process of training the prediction model of the position and the network flow, historical service related data of part of user equipment in the user equipment group can be used as a training sample, so that if the historical service related data of some user equipment in the user equipment group is lost, the scheme is still applicable and has strong realizability, thus the requirement on a data set is reduced, and the problem that the model training effect is influenced due to the small data quantity or the data loss in the data set is solved.

In practical applications, it is assumed that the number of user equipments included in the target user equipment group is N_cPredicted users of the target group of user devicesThe average position of the device at the target moment is l_nextThe average network flow of each user equipment at the target moment is f_nextIf the target user equipment group is located at the target time, the target position of the target user equipment group is set to be L_next＝l_nextThe network flow of the target user equipment group at the target moment is F_next＝f_next*N_c。

In practical application, the model parameters of the position prediction model and the model parameters of the flow prediction model can be updated based on a preset period.

In some embodiments, the method further comprises:

Here, the target location of the target ue group may be a target cell, for example, table 5 shows the target location and the network traffic of the target ue group at a target time based on a prediction model of location and network traffic, as shown in table 5:

target user equipment group identification	Target position	Network traffic
			1	Cell 1	100k
2	Cell 1	48k
			…	…	…
S	Cell 3	250k

TABLE 5

In practical application, the predicted network traffic predicted values of all target user equipment groups appearing in the same target cell, such as cell 1, are summed and summarized, so that a total network traffic predicted value corresponding to the target cell can be obtained. Similarly, the network management server according to the embodiment of the present invention may further determine, based on the total network traffic predicted value of each target cell, a total network traffic predicted value corresponding to a target area including each target cell.

In practical application, when the predicted total network traffic value of the target location exceeds the total network traffic threshold value, it is indicated that the future network load condition of the target location is not good, and the network device of the target location needs to be adjusted in advance to ensure the network security and the perception of the network user.

Based on this, in some embodiments, the method further comprises: comparing the total network traffic predicted value of the target position with the total network traffic threshold value of the target position to obtain a comparison result; and when the comparison result represents that the total network flow predicted value of the target position is greater than the total network flow threshold value, maintaining and optimizing the network equipment of the target position.

Correspondingly, the embodiment of the invention also provides another network traffic prediction method, which is an embodiment provided for the application of the prediction model of the position and the network traffic.

Referring to fig. 6, fig. 6 is a flowchart illustrating another method for predicting network traffic according to an embodiment of the present invention, where the method for predicting network traffic may be implemented by a network traffic prediction device, and the network traffic prediction device according to the embodiment of the present invention may be implemented as various types of terminal devices, such as a notebook computer, a desktop computer, and the like, or may be implemented as a network management server, such as a cloud server for network management or a local server for network management. The following describes a method for predicting network traffic according to an embodiment of the present invention, with reference to the steps shown in fig. 6, by taking an embodiment of a network management server as an example. For details which are not exhaustive in the following description of the steps, reference is made to the above for an understanding.

Step 601, acquiring real-time service related data of a target user equipment group at the current moment, wherein the real-time service related data at least comprises real-time position data and real-time network flow data.

Step 602, inputting the real-time location data and the real-time network traffic data to a prediction model of location and network traffic, respectively, to obtain a target location of the target user equipment group at a target time and a predicted value of the network traffic of the target user equipment group at the target time, which are output by the prediction model.

Here, the target time is used to represent any time starting from the current time.

In practical application, the prediction model of the position and the network traffic is obtained by performing model training based on historical position data and historical network traffic data of user equipment in the user equipment group as training sample data,

specifically, the network management server may train and obtain a prediction model of the location and the network traffic by: and respectively carrying out prediction model training on the position and the network flow of the corresponding user equipment group by taking the historical position data and the historical network flow data of the user equipment in each user equipment group as training sample data to obtain a prediction model corresponding to the position and the network flow of each user equipment group. The user equipment group can be clustered and divided based on the similarity of the service related characteristics between every two user equipments in the plurality of user equipments.

Here, the historical location data and the historical network traffic data are both associated with historical traffic related data, and specifically, the historical traffic related data includes the historical location data and the historical network traffic data. The prediction model of the position and the network traffic is used for predicting the target position and the network traffic of the target user equipment group.

Step 603, determining a total network traffic predicted value of the target location based on the target location of the target user equipment group at the target time and the network traffic predicted value of the target user equipment group at the target time.

In some embodiments, determining a total network traffic predicted value of the target location based on the target location of the target user equipment group at the target time and the network traffic predicted value of the target user equipment group at the target time comprises:

Here, the target location of the target user equipment group may be a target cell. In practical application, the predicted network traffic predicted values of all target ue groups appearing in the same target cell, such as cell 1 in table 5, are summed and summarized to obtain a total network traffic predicted value corresponding to the target cell. Similarly, the network management server according to the embodiment of the present invention may further determine, based on the total network traffic predicted value of each target cell, a total network traffic predicted value corresponding to a target area including each target cell.

The present invention will be described in further detail with reference to the following application examples.

Referring to fig. 7, fig. 7 is a schematic flow chart of another network traffic prediction method according to an embodiment of the present invention, where the network traffic prediction method includes an offline training part and an online prediction part, the offline training part mainly identifies and divides user equipment groups by using historical service related data of a plurality of user equipment, and constructs a prediction model of a corresponding location and network traffic for each user equipment group; the online prediction part mainly inputs the real-time service related data at the current moment into a trained position and network flow prediction model so as to predict a target cell of the target user equipment group at the target moment and the network flow of the target user equipment group at the target moment.

Specifically, in the offline training process, feature extraction is performed on the acquired historical service related data of a plurality of user equipment to obtain service related features corresponding to each user equipment, wherein the service related features comprise a motion track, a stay area and service flow, then the similarity of the service related features between every two user equipment in the plurality of user equipment is determined, namely the motion track similarity between every two user equipment in the plurality of user equipment is respectively determined, the stay area similarity between every two user equipment in the plurality of user equipment is determined, the service flow similarity between every two user equipment in the plurality of user equipment is determined, and then the plurality of user equipment are clustered according to the similarity of the service related features to obtain different user equipment groups; and then, performing associated modeling of the position and the network traffic on each user equipment group to obtain a prediction model corresponding to the position and the network traffic of each user equipment group.

In the online prediction process, firstly acquiring real-time service related data of a target user equipment group at the current moment, wherein the real-time service related data at least comprises real-time position data and real-time network flow data, respectively inputting the real-time position data and the real-time network flow data into a prediction model of a position and network flow, acquiring a target position of the target user equipment group at the target moment, namely a target cell of the target user equipment group at the target moment, and acquiring network flow of the target user equipment group at the target moment; next, all network traffic of the target ue group set of the same target cell is summarized to obtain a total predicted network traffic value corresponding to the target cell, and the total predicted network traffic value corresponding to the target cell is output, so as to manage the network devices of the target cell based on the total predicted network traffic value corresponding to the target cell.

In order to implement the method of the model training side in the embodiment of the present invention, an embodiment of the present invention further provides a device for predicting network traffic, referring to fig. 8, where fig. 8 is a schematic structural diagram of the device for predicting network traffic provided in the embodiment of the present invention, and the device includes:

a first obtaining unit 81, configured to obtain historical service related data of a plurality of user equipments;

a feature extraction unit 82, configured to perform feature extraction on the historical service-related data to obtain service-related features corresponding to each piece of the user equipment;

a cluster partitioning unit 83, configured to perform cluster partitioning on the multiple pieces of user equipment based on similarity of service-related features between every two pieces of user equipment in the multiple pieces of user equipment, so as to obtain at least two user equipment groups;

a model training unit 84, configured to use historical location data and historical network traffic data of user equipment in each user equipment group as training sample data, and perform prediction model training on location and network traffic for the corresponding user equipment group, respectively, to obtain a prediction model corresponding to the location and network traffic of each user equipment group;

In some embodiments, the apparatus further comprises:

In some embodiments, the first determining unit is specifically configured to:

In some embodiments, the cluster partitioning unit is specifically configured to:

In some embodiments, the apparatus further comprises: the normalization unit is used for carrying out normalization processing on the similarity matrix to obtain a normalized similarity matrix;

the cluster partitioning unit is specifically configured to:

In some embodiments, the apparatus further comprises:

In practical applications, the first obtaining unit 81 may be implemented by a communication interface in a prediction device of network traffic; the feature extraction unit 82, the cluster partitioning unit 83, and the model training unit 84 may be implemented by a processor in a prediction device of network traffic in combination with a communication interface.

In order to implement the method of the model application side in the embodiment of the present invention, another network traffic prediction apparatus is further provided in the embodiment of the present invention, referring to fig. 9, fig. 9 is a schematic structural diagram of another network traffic prediction apparatus provided in the embodiment of the present invention, where the apparatus includes:

a second obtaining unit 91, configured to obtain real-time service related data of a target user equipment group at a current time, where the real-time service related data at least includes real-time location data and real-time network traffic data;

a data input unit 92, configured to input the real-time location data and the real-time network traffic data to a prediction model, respectively, so as to obtain a target location of the target user equipment group at a target time and a predicted network traffic value of the target user equipment group at the target time, where the target location and the predicted network traffic value are output by the prediction model; the target time is used for representing any time taking the current time as the starting time;

a third determining unit 93, configured to determine a total network traffic predicted value of the target location based on a target location of the target ue group at a target time and a network traffic predicted value of the target ue group at the target time;

In some embodiments, the third determining unit is specifically configured to:

In some embodiments, the apparatus further comprises:

In practical applications, the second obtaining unit 91 may be implemented by a communication interface in a prediction device of network traffic; the data input unit 92 and the third determination unit 93 may be implemented by a communication interface in a prediction device of network traffic in combination with a processor.

It should be noted that, when the network traffic prediction apparatus provided in the foregoing embodiment performs network traffic prediction, the division of each program module is merely exemplified, and in practical applications, the above processing may be distributed to different program modules according to needs, that is, the internal structure of the apparatus is divided into different program modules to complete all or part of the above-described processing. In addition, the network traffic prediction apparatus provided in the foregoing embodiment and the network traffic prediction method embodiment belong to the same concept, and specific implementation processes thereof are described in the method embodiment and are not described herein again.

Based on the hardware implementation of the program module, and in order to implement the method at the model training side in the embodiment of the present invention, an embodiment of the present invention further provides a device for predicting network traffic, referring to fig. 10, where fig. 10 is a schematic structural diagram of the device for predicting network traffic provided in the embodiment of the present invention, and the device 1000 for predicting network traffic includes:

a first communication interface 1001 for acquiring historical service related data of a plurality of user equipments;

the first processor 1002 is connected to the first communication interface 1001, and is configured to execute a method provided by one or more technical solutions of the model training side when running a computer program. And the computer program is stored on the first memory 1003.

Specifically, the first processor 1002 is configured to perform feature extraction on the historical service related data, so as to obtain service related features corresponding to each piece of the user equipment; based on the similarity of service related features between every two pieces of user equipment in the plurality of pieces of user equipment, clustering and dividing the plurality of pieces of user equipment to obtain at least two user equipment groups; taking historical position data and historical network traffic data of user equipment in each user equipment group as training sample data, and respectively carrying out position and network traffic prediction model training on the corresponding user equipment group to obtain a prediction model corresponding to the position and network traffic of each user equipment group; the prediction model is used for predicting a target position and network traffic of a target user equipment group, and the target position and the network traffic are used for determining a total network traffic predicted value corresponding to the target position.

In some embodiments, the first processor 1002 is further configured to determine a similarity of the service related features between each two of the plurality of user equipments based on the service related features of the user equipments.

In some embodiments, the first processor 1002 is specifically configured to:

In some embodiments, the first processor 1002 is further configured to: normalizing the similarity matrix to obtain a normalized similarity matrix;

accordingly, the first processor 1002 is specifically configured to: and carrying out graph conversion on the normalized similarity matrix to obtain the graph model.

It should be noted that specific processing procedures of the first communication interface 1001 and the first processor 1002 are detailed in the method embodiment, and are not described herein again.

Of course, in practice, the various components of the predictive device 1000 of network traffic are coupled together by the bus system 1004. It is understood that the bus system 1004 is used to enable communications among the components. The bus system 1004 includes a power bus, a control bus, and a status signal bus in addition to a data bus. But for the sake of clarity the various busses are labeled in fig. 10 as the bus system 1004.

The first memory 1003 in the embodiment of the present invention is used to store various types of data to support the operation of the prediction apparatus 1000 for network traffic. Examples of such data include: any computer program for operating on a predictive device 1000 for network traffic.

The method disclosed in the above embodiments of the present invention may be applied to the first processor 1002, or implemented by the first processor 1002. The first processor 1002 may be an integrated circuit chip having signal processing capabilities. In implementation, the steps of the above method may be performed by integrated logic circuits of hardware or instructions in the form of software in the first processor 1002. The first Processor 1002 may be a general purpose Processor, a Digital Signal Processor (DSP), or other programmable logic device, discrete gate or transistor logic device, discrete hardware components, etc. The first processor 1002 may implement or perform the methods, steps, and logic blocks disclosed in embodiments of the present invention. A general purpose processor may be a microprocessor or any conventional processor or the like. The steps of the method disclosed by the embodiment of the invention can be directly implemented by a hardware decoding processor, or can be implemented by combining hardware and software modules in the decoding processor. The software module may be located in a storage medium located in the first memory 1003, and the first processor 1002 reads the information in the first memory 1003 and completes the steps of the foregoing method in combination with the hardware thereof.

In an exemplary embodiment, the Device 1000 for predicting network traffic may be implemented by one or more Application Specific Integrated Circuits (ASICs), DSPs, Programmable Logic Devices (PLDs), Complex Programmable Logic Devices (CPLDs), Field Programmable Gate Arrays (FPGAs), general purpose processors, controllers, Micro Controllers (MCUs), microprocessors (microprocessors), or other electronic components for performing the foregoing methods.

Based on the hardware implementation of the program module, and in order to implement the method on the model application side in the embodiment of the present invention, an embodiment of the present invention further provides another device for predicting network traffic, referring to fig. 11, where fig. 11 is a schematic structural diagram of another device for predicting network traffic provided in the embodiment of the present invention, and the device 1100 for predicting network traffic includes:

the second communication interface 1101 is configured to obtain real-time service related data of the target user equipment group at the current time, where the real-time service related data at least includes real-time location data and real-time network traffic data;

the second processor 1102 is connected to the second communication interface 1101, and configured to execute the method provided by one or more technical solutions of the application side of the model when running the computer program. And the computer program is stored on the second memory 1103.

Specifically, the second processor 1102 is configured to input the real-time location data and the real-time network traffic data to a prediction model, so as to obtain a target location of the target user equipment group at a target time and a network traffic prediction value of the target user equipment group at the target time, where the target location and the network traffic prediction value are output by the prediction model; the target time is used for representing any time taking the current time as the starting time; the network traffic prediction method is further used for determining a total network traffic prediction value of the target position based on the target position of the target user equipment group at the target moment and the network traffic prediction value of the target user equipment group at the target moment; the prediction model is obtained by performing model training based on historical position data and historical network traffic data of user equipment in the user equipment group as training sample data.

In some embodiments, the second processor 1102 is specifically configured to: determining a target user equipment group set positioned at the same target position based on the target position of the target user equipment group at the target moment; and summarizing the network traffic predicted values of all target user equipment groups in the target user equipment group set at the same target position to obtain a total network traffic predicted value corresponding to the target position.

In some embodiments, the second processor 1102 is further configured to compare the total network traffic predicted value of the target location with a total network traffic threshold of the target location, so as to obtain a comparison result; and when the comparison result represents that the total network flow predicted value of the target position is greater than the total network flow threshold value, maintaining and optimizing the network equipment of the target position.

It should be noted that specific processing procedures of the second communication interface 1101 and the second processor 1102 are detailed in the method embodiment, and are not described herein again.

Of course, in actual practice, the various components of the predictive device 1100 of network traffic are coupled together by a bus system 1104. It is understood that the bus system 1104 is used to enable communications among the components for connection. The bus system 1104 includes a power bus, a control bus, and a status signal bus in addition to the data bus. For clarity of illustration, however, the various buses are designated as the bus system 1104 in FIG. 11.

The second memory 1103 in the embodiment of the present invention is used to store various types of data to support the operation of the prediction device 1100 for network traffic. Examples of such data include: any computer program for operating on a predictive device 1100 of network traffic.

The method disclosed in the above embodiments of the present invention can be applied to the second processor 1102 or implemented by the second processor 1102. The second processor 1102 may be an integrated circuit chip having signal processing capabilities. In implementation, the steps of the method may be performed by integrated logic circuits of hardware or instructions in the form of software in the second processor 1102. The second processor 1102 described above may be a general purpose processor, a DSP, or other programmable logic device, discrete gate or transistor logic device, discrete hardware components, or the like. The second processor 1102 may implement or perform the methods, steps, and logic blocks disclosed in embodiments of the present invention. A general purpose processor may be a microprocessor or any conventional processor or the like. The steps of the method disclosed by the embodiment of the invention can be directly implemented by a hardware decoding processor, or can be implemented by combining hardware and software modules in the decoding processor. The software module may be located in a storage medium located in the second memory 1103, and the second processor 1102 reads the information in the second memory 1103 and performs the steps of the foregoing method in combination with its hardware.

The prediction device 1100 of network traffic may be implemented in an exemplary embodiment by one or more ASICs, DSPs, PLDs, CPLDs, FPGAs, general purpose processors, controllers, MCUs, microprocessors, or other electronic components for performing the aforementioned methods.

It is understood that the memories (the first memory 1003 and the second memory 1103) of the embodiments of the present invention may be volatile memories or nonvolatile memories, and may include both volatile and nonvolatile memories. Among them, the nonvolatile Memory may be a Read Only Memory (ROM), a Programmable Read Only Memory (PROM), an Erasable Programmable Read-Only Memory (EPROM), an Electrically Erasable Programmable Read-Only Memory (EEPROM), a magnetic random access Memory (FRAM), a Flash Memory (Flash Memory), a magnetic surface Memory, an optical disk, or a Compact Disc Read-Only Memory (CD-ROM); the magnetic surface storage may be disk storage or tape storage. Volatile Memory can be Random Access Memory (RAM), which acts as external cache Memory. By way of illustration and not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Synchronous Static Random Access Memory (SSRAM), Dynamic Random Access Memory (DRAM), Synchronous Dynamic Random Access Memory (SDRAM), Double Data Rate Synchronous Dynamic Random Access Memory (DDRSDRAM), Enhanced Synchronous Dynamic Random Access Memory (ESDRAM), Enhanced Synchronous Dynamic Random Access Memory (Enhanced Synchronous Dynamic Random Access Memory), Synchronous linked Dynamic Random Access Memory (DRAM, Synchronous Link Dynamic Random Access Memory), Direct Memory (DRmb Random Access Memory). The described memory for embodiments of the present invention is intended to comprise, without being limited to, these and any other suitable types of memory.

In an exemplary embodiment, the embodiment of the present invention further provides a storage medium, specifically a computer-readable storage medium, for example, the storage medium includes a first memory 1003 storing a computer program, and the computer program is executable by a first processor 1002 of the device 1000 for predicting network traffic, so as to complete the steps described in the model training side method. For example, the second memory 1103 may store a computer program, which may be executed by the second processor 1102 of the network traffic prediction device 1100, to perform the steps described in the model application side method. The computer readable storage medium can be FRAM, ROM, PROM, EPROM, EEPROM, Flash Memory, magnetic surface Memory, optical disk or CD-ROM; or may be various devices including one or any combination of the above memories.

In embodiments of the present invention, reference may be made to the terms "first," "second," etc. merely for distinguishing between similar elements and not for describing a particular sequential or chronological order, but it is to be understood that "first," "second," etc. may, where permissible, be interchanged with other specific sequences or orderings such that embodiments of the present invention described herein may be practiced otherwise than as specifically illustrated or described herein.

The above description is only for the specific embodiments of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present invention, and all the changes or substitutions should be covered within the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the appended claims.

Claims

1. A method for predicting network traffic, the method comprising:

acquiring historical service related data of a plurality of user equipment;

2. The method of claim 1, further comprising:

3. The method of claim 2, wherein the determining the similarity of the traffic-related characteristics between each two user equipments in the plurality of user equipments based on the traffic-related characteristics of each user equipment comprises:

4. The method of claim 3, wherein the determining the similarity of motion trajectories between each two user devices in the plurality of user devices comprises:

5. The method of claim 3, wherein the determining the stay area similarity between each two of the plurality of user equipments comprises:

6. The method of claim 3, wherein the determining the traffic flow similarity between each two of the plurality of user equipments comprises:

7. The method of claim 1, wherein the clustering the plurality of user equipments based on similarity of service related features between each two user equipments in the plurality of user equipments to obtain at least two user equipment groups comprises:

8. The method of claim 7, further comprising: normalizing the similarity matrix to obtain a normalized similarity matrix;

9. The method of claim 1, further comprising:

10. The method of claim 9, further comprising:

11. A method for predicting network traffic, the method comprising:

12. The method of claim 11, wherein the determining a total predicted network traffic value for the target location based on the target location of the target ue group at the target time and the predicted network traffic value of the target ue group at the target time comprises:

13. The method of claim 12, further comprising:

14. An apparatus for predicting network traffic, the apparatus comprising:

15. An apparatus for predicting network traffic, the apparatus comprising:

16. An apparatus for predicting network traffic, comprising: a first processor and a first memory for storing a computer program operable on the processor;

wherein the first processor is adapted to perform the steps of the method of any one of claims 1 to 10 when running the computer program.

17. An apparatus for predicting network traffic, comprising: a second processor and a second memory for storing a computer program operable on the processor;

wherein the second processor is adapted to perform the steps of the method of any of claims 11 to 13 when running the computer program.

18. A storage medium having stored thereon a computer program for performing the steps of the method of any one of claims 1 to 10 or for performing the steps of the method of any one of claims 11 to 13 when executed by a processor.