CN117411818A

CN117411818A - Network path determining method and device, electronic equipment and storage medium

Info

Publication number: CN117411818A
Application number: CN202210802912.5A
Authority: CN
Inventors: 刘波
Original assignee: China Telecom Corp Ltd
Current assignee: China Telecom Corp Ltd
Priority date: 2022-07-07
Filing date: 2022-07-07
Publication date: 2024-01-16

Abstract

The embodiment of the disclosure provides a network path determining method, a device, electronic equipment and a storage medium, and a first network element contained in each network layer between a target user end and a corresponding target resource end is determined. And acquiring network state data of the first network element aiming at the target type user side as network state data to be processed aiming at each first network element. The target type represents the type of the target client. And inputting the network state data to be processed into a service quality prediction model corresponding to the target type to obtain service quality data corresponding to the first network element, wherein the service quality data is used as the service quality data to be processed. For each network layer, a target network element is selected from the network layer based on the service quality data to be processed corresponding to the first network element in the network layer. And determining a network path formed by each target network element as a target network path between the target user end and the target resource end. Based on the method, the quality of network service when the target user terminal accesses the target resource terminal can be improved.

Description

Network path determining method and device, electronic equipment and storage medium

Technical Field

The disclosure relates to the technical field of big data, and in particular relates to a network path determining method, a network path determining device, electronic equipment and a storage medium.

Background

With the development of information technology, a user can access a resource end through a user end so as to acquire corresponding network resources from the resource end. For example, a user accesses a corresponding resource end through a user end of the video class to acquire video network resources from the resource end. Furthermore, the user side can play the obtained video network resources for the user to watch.

The network state of each network element in the network path between the user end and the resource end can affect the quality of network service. For example, when the network state of the network element is poor, situations such as blocking and delay occur when the user side plays the video resource, and the quality of the network service is reduced.

In the related art, when the network state of the network element is poor, the user end can only passively adapt to the network. For example, when the network state of the network element is poor, the user side may decrease the resolution of the played video network resource, so as to avoid situations such as jamming and delay. It can be seen that the ue can avoid the occurrence of the jamming, the delay, etc. by reducing the resolution, however, reducing the resolution also affects the quality of the network service, i.e. it cannot effectively improve the quality of the network service per se.

Therefore, a method is needed to effectively improve the quality of network traffic.

Disclosure of Invention

An object of an embodiment of the present disclosure is to provide a network path determining method, apparatus, electronic device, and storage medium, so as to improve quality of network service when a target user accesses a target resource. The specific technical scheme is as follows:

in order to achieve the above object, an embodiment of the present disclosure provides a network path determining method, including:

determining network elements contained in each network layer between a target user end and a corresponding target resource end as a first network element;

for each first network element, acquiring network state data of the first network element for a target type user terminal as network state data to be processed; wherein the target type represents the type of the target user terminal;

inputting the network state data to be processed into a service quality prediction model corresponding to the target type to obtain service quality data corresponding to the first network element, wherein the service quality data is used as the service quality data to be processed; the service quality prediction model is obtained by training sample network state data of a sample user terminal of the target type based on a sample network element and sample service quality data corresponding to the sample network state data; the sample network element is a network element in a network path between the sample user end and a corresponding resource end; the sample quality of service data represents: when the network state of the sample network element is the sample network state data, the quality of the network service provided by the sample user terminal;

For each network layer, selecting one network element from the network layer based on the service quality data to be processed corresponding to the first network element in the network layer as a target network element;

and determining a network path formed by each target network element as a target network path between the target user end and the target resource end.

In some embodiments, the training process of the quality of service prediction model includes the steps of:

acquiring sample network state data of a sample network element aiming at a sample user end of the target type and sample service quality data corresponding to the sample network state data;

inputting the sample network state data into a service quality prediction model of a preset structure to obtain service quality data corresponding to the sample network state data as predicted service quality data;

calculating a loss value based on the predicted quality of service data and the sample quality of service data;

and adjusting model parameters of the service quality prediction model of the preset structure based on the loss value, and continuing training until the service quality prediction model of the preset structure reaches convergence.

In some embodiments, the obtaining sample network state data of the sample network element for the sample user end of the target type and sample service quality data corresponding to the sample network state data includes:

For each sample time period, acquiring sample network state data of the sample network element for the sample user side of the target type in the sample time period;

and acquiring sample service quality data of the sample user terminal in the sample time period, wherein the sample service quality data corresponds to the sample network state data of the sample user terminal of the target type in the sample time period.

In some embodiments, the sample service quality data of the sample user end in each sample period is generated by the following steps:

for each sample time period, acquiring service state data of each sample user side in the sample time period;

based on cluster analysis of the service state data, obtaining service quality data represented by each service state data as first service quality data;

and exchanging the first service quality data corresponding to the two sample user terminals belonging to the same network segment aiming at the target sample time period to obtain the sample service quality data of the two sample user terminals in the target sample time period.

In some embodiments, the sample network state data of the sample network element for the sample user end of the target type in each sample period is generated by:

For each sample time period, acquiring network state data of a sample user side of the sample network element aiming at the target type in the sample time period as first network state data;

and carrying out differential processing based on the first network state data corresponding to the sample time period and the first network state data of the next sample time period of the sample time period to obtain sample network state data of the sample network element of the sample time period aiming at the sample user side of the target type.

In some embodiments, for each first network element, the acquiring network state data of the first network element for the target type of the user side as the network state data to be processed includes:

for each first network element, acquiring network state data of the first network element for a plurality of user terminals of a target type;

and obtaining the network state data to be processed based on the obtained network state data.

In a second aspect, to achieve the above object, an embodiment of the present disclosure provides a network path determining apparatus, including:

a first determining module, configured to determine network elements included in each network layer between a target user terminal and a corresponding target resource terminal, as a first network element;

The first acquisition module is used for acquiring network state data of the first network element aiming at the target type user side as network state data to be processed aiming at each first network element; wherein the target type represents the type of the target user terminal;

the first prediction module is used for inputting the network state data to be processed into a service quality prediction model corresponding to the target type to obtain service quality data corresponding to the first network element, wherein the service quality data is used as service quality data to be processed; the service quality prediction model is obtained by training sample network state data of a sample user terminal of the target type based on a sample network element and sample service quality data corresponding to the sample network state data; the sample network element is a network element in a network path between the sample user end and a corresponding resource end; the sample quality of service data represents: when the network state of the sample network element is the sample network state data, the quality of the network service provided by the sample user terminal;

a selecting module, configured to select, for each network layer, one network element from the network layers based on to-be-processed quality of service data corresponding to a first network element in the network layer, as a target network element;

And the second determining module is used for determining the network path formed by each target network element as a target network path between the target user end and the target resource end.

In some embodiments, the apparatus further comprises:

the second acquisition module is used for acquiring sample network state data of a sample user side of the sample network element aiming at the target type and sample service quality data corresponding to the sample network state data;

the second prediction module is used for inputting the sample network state data into a service quality prediction model of a preset structure, and obtaining service quality data corresponding to the sample network state data as predicted service quality data;

a third determining module for calculating a loss value based on the predicted quality of service data and the sample quality of service data;

and the training module is used for adjusting the model parameters of the service quality prediction model of the preset structure based on the loss value, and continuing training until the service quality prediction model of the preset structure reaches convergence.

In some embodiments, the second obtaining module is specifically configured to obtain, for each sample period, sample network state data of the sample network element for the sample user side of the target type in the sample period;

In some embodiments, the second obtaining module is specifically configured to obtain, for each sample period, service state data of each sample user terminal in the sample period;

In some embodiments, the second obtaining module is specifically configured to obtain, for each sample period, network state data of the sample network element for the sample user terminal of the target type in the sample period, as first network state data;

In some embodiments, the first obtaining module is specifically configured to obtain, for each first network element, network state data of the first network element for a plurality of clients of a target type;

The embodiment of the disclosure also provides electronic equipment, which comprises a processor, a communication interface, a memory and a communication bus, wherein the processor and the communication interface, and the memory are communicated with each other through the communication bus;

a memory for storing a computer program;

and the processor is used for realizing any of the network path determining method steps when executing the program stored in the memory.

The disclosed embodiments also provide a computer readable storage medium having a computer program stored therein, which when executed by a processor, implements any of the above described network path determination method steps.

The disclosed embodiments also provide a computer program product containing instructions which, when run on a computer, cause the computer to perform any of the network path determination methods described above.

The network path determining method provided by the implementation of the present disclosure determines network elements included in each network layer between a target user terminal and a corresponding target resource terminal as a first network element. And acquiring network state data of the first network element aiming at the target type user side as network state data to be processed aiming at each first network element. Wherein the target type represents the type of the target user terminal. And inputting the network state data to be processed into a service quality prediction model corresponding to the target type to obtain service quality data corresponding to the first network element, wherein the service quality data is used as the service quality data to be processed. The service quality prediction model is obtained by training sample network state data of a sample user side aiming at a target type based on a sample network element and sample service quality data corresponding to the sample network state data; the sample network element is a network element in a network path between the sample user end and the corresponding resource end; sample quality of service data representation: and when the network state of the sample network element is the sample network state data, the quality of the network service provided by the sample user terminal. For each network layer, selecting one network element from the network layers based on the service quality data to be processed corresponding to the first network element in the network layer as a target network element. And determining a network path formed by each target network element as a target network path between the target user end and the target resource end.

Based on the above processing, for each network element included in each network layer between the target user terminal and the corresponding target resource terminal, corresponding service quality data (i.e. to-be-processed service quality data) can be determined based on the network state data of the network element, where the to-be-processed service quality data can represent the quality of the network service provided by the corresponding user terminal. Furthermore, based on the service quality data to be processed corresponding to each network element, the network element with better quality of the corresponding network service can be selected, and the network path formed by each selected network element is determined as the target network path between the target user terminal and the target resource terminal, so that the quality of the network service when the target user terminal accesses the target resource terminal can be improved.

Drawings

In order to more clearly illustrate the embodiments of the present disclosure or the technical solutions in the prior art, the following description will briefly introduce the drawings that are required to be used in the embodiments or the description of the prior art, and it is apparent that the drawings in the following description are only some embodiments of the present disclosure, and other embodiments may be obtained according to these drawings to those of ordinary skill in the art.

Fig. 1 is a schematic diagram of a flow transmission provided in an embodiment of the present disclosure;

fig. 2 is a flowchart of a network path determining method according to an embodiment of the present disclosure;

FIG. 3 is a flowchart of another network path determination method provided by an embodiment of the present disclosure;

fig. 4 is a flowchart of a method for training a service quality prediction model according to an embodiment of the present disclosure;

FIG. 5 is a flowchart of another method for training a quality of service prediction model according to an embodiment of the present disclosure;

fig. 6 is a schematic diagram of a training method of a service quality prediction model according to an embodiment of the present disclosure;

fig. 7 is a block diagram of a network path determining apparatus according to an embodiment of the present disclosure;

fig. 8 is a block diagram of an electronic device according to an embodiment of the present disclosure.

Detailed Description

The following description of the technical solutions in the embodiments of the present disclosure will be made clearly and completely with reference to the accompanying drawings in the embodiments of the present disclosure, and it is apparent that the described embodiments are only some embodiments of the present disclosure, not all embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by one of ordinary skill in the art based on the present disclosure are within the scope of the present disclosure.

The user end can communicate with the resource end through each network element, and each network element forms a network path between the user end and the resource end. For example, referring to fig. 1, fig. 1 is a schematic diagram of a network path provided by an implementation of the present disclosure. The 5G-CPE (5G-Customer Premise Equipment ) is the customer premise equipment in the embodiments of the present disclosure. The base station, the bearer network, the UPF (User Plane Function ), and the gateway device are network elements included in each network layer between the user end and the corresponding resource end in the embodiment of the present disclosure. The application cloud is a resource end in the embodiment of the disclosure.

In the 5G service, the service flow of the user sequentially passes through the UPF, gateway device, bearer network 2, and other network elements of the base station, bearer network 1, 5GC (5 gcore,5G core network), and reaches the application cloud (i.e., the resource end). The base station, the bearer network 1, the UPF, the gateway device and the bearer network 2 form a network path between the user end and the resource end.

In the related art, when the network state of the network element in the network path is poor, the user terminal can only passively adapt to the network. For example, when the network state of the network element is poor, the user side may decrease the resolution of the played video network resource, so as to avoid situations such as jamming and delay. It can be seen that the ue can avoid the occurrence of the jamming, the delay, etc. by reducing the resolution, however, reducing the resolution also affects the quality of the network service, i.e. it cannot effectively improve the quality of the network service per se.

In order to solve the above-mentioned problems, the present disclosure provides a network path determining method, as shown in fig. 2, fig. 2 is a flowchart of network path determination provided in an embodiment of the present disclosure, where the method may include the following steps:

s201: and determining network elements contained in each network layer between the target user end and the corresponding target resource end as a first network element.

S202: and acquiring network state data of the first network element aiming at the target type user side as network state data to be processed aiming at each first network element.

Wherein the target type represents the type of the target user terminal.

S203: and inputting the network state data to be processed into a service quality prediction model corresponding to the target type to obtain service quality data corresponding to the first network element, wherein the service quality data is used as the service quality data to be processed.

The service quality prediction model is obtained by training sample network state data of a sample user side aiming at a target type based on a sample network element and sample service quality data corresponding to the sample network state data; the sample network element is a network element in a network path between the sample user end and the corresponding resource end; sample quality of service data representation: and when the network state of the sample network element is the sample network state data, the quality of the network service provided by the sample user terminal.

S204: for each network layer, selecting one network element from the network layers based on the service quality data to be processed corresponding to the first network element in the network layer as a target network element.

S205: and determining a network path formed by each target network element as a target network path between the target user end and the target resource end.

Based on the network path determining method provided by the embodiment of the present disclosure, for each network element included in each network layer between the target user terminal and the corresponding target resource terminal, corresponding service quality data (i.e., to-be-processed service quality data) may be determined based on the network state data of the network element, where the to-be-processed service quality data may represent the quality of the network service provided by the corresponding user terminal. Furthermore, based on the service quality data to be processed corresponding to each network element, the network element with better quality of the corresponding network service can be selected, and the network path formed by each selected network element is determined as the target network path between the target user terminal and the target resource terminal, so that the quality of the network service when the target user terminal accesses the target resource terminal can be improved.

For step S201, the target user side may be a user side where any one of application programs (may be referred to as a target application program) used by the user is located, the target application program may be a video application program, a web application program, or the like, and the target user side may be a smart phone, a tablet computer, a desktop computer, or the like. The target resource end corresponding to the target user end can be a server of the application program used by the user. A plurality of network layers are included between the target user end and the corresponding target resource end, each network layer includes a plurality of network elements, which may be referred to as a first network element.

When a user acquires a resource from a target resource end through a target user end, the target user end sends a resource access request, and the resource access request is transmitted to the target resource end through each network element in a network path. And after the target resource end receives the resource access request, the target resource end sends the network resource requested by the resource access request to the target user end. Based on the method provided by the embodiment of the disclosure, the target network path from the target user terminal to the target resource terminal with better quality of the corresponding network service can be determined, so that the quality of the network service when the target user terminal accesses the target resource terminal is improved.

For step S202, the target type represents the type of the target user terminal, the type of the target user terminal represents the type of the target application program, for example, the target application program is a video application program, and the target type of the target user terminal is a video type; or the target application program is a webpage application program, and the target user terminal is a webpage type; or the target application program is an audio application program, and the target user terminal is an audio type.

The network state data of the first network element for the target type user comprises: the network state data when the first network element transmits the network resource requested by the user terminal of the target type may include: total port traffic, port packet loss rate, packet error rate, user Session bandwidth, source IP (Internet Protocol ) address, destination IP address, session time, etc. User Session refers to a Session between a user end and a resource end. The source IP address is: the address of the user end, the destination IP address is: address of resource end.

The 5GC management platform can extract network state data such as total port flow, port packet loss rate, packet error rate, user Session bandwidth, source IP address, destination IP address, session time and the like from the base station and UPF. The network management platform can obtain the data such as the total flow of different ports, the port packet loss rate and the like on the bearing network and the gateway equipment.

When the network state data is acquired, the first network elements between the user end and the resource end are required to be connected through the connection relation of the network topology, in some cases, the bearing network, the gateway equipment, the base station and the ports of the UPF are possibly not in one-to-one correspondence relation, and the network state data are discarded. For example, if some ports of the base station are not connected to the bearer network, the total port traffic of the ports of the base station, the port packet loss rate, and other network state data may not be acquired.

In some embodiments, based on fig. 2, referring to fig. 3, step S202 may include the steps of:

s2021: and acquiring network state data of the first network element for a plurality of user terminals of the target type aiming at each first network element.

S2022: and obtaining the network state data to be processed based on the obtained network state data.

For each first network element, the electronic device may acquire network state data of a plurality of clients of the target type for the first network element in a period of time closest to the current time. Furthermore, the electronic device may calculate an average value of the network status data of the plurality of clients as the network status data to be processed. For example, an average value of port packet loss rates of the first network element for a plurality of user terminals of the target type is calculated, and an average value of packet error rates of the first network element for a plurality of user terminals of the target type is calculated, so as to obtain network state data to be processed.

Alternatively, the electronic device may calculate the weighted value of the network state data of the first network for the plurality of user terminals of the target type as the network state data to be processed.

For step S203, the quality of service prediction model may be a classification model, where different types of quality of service prediction models are used to predict the quality of service data for the first network element for different types. For example, if the target type is a video type, the service quality prediction model corresponding to the target type is used for predicting the service quality data of the first network element aiming at the video type; or if the target type is an audio type, the service quality prediction model corresponding to the target type is used for predicting the service quality data of the first network element aiming at the audio type.

The service quality data to be processed corresponding to the first network element is represented by: based on the to-be-processed network state data of the first network element, the larger the to-be-processed service quality data is, the better the quality of the network service provided by the corresponding target user terminal is. For example, the target type is a video type, and the service quality data to be processed corresponding to the first network element indicates: based on the network state data to be processed of the first network element, the quality of the video service provided by the corresponding target user terminal.

The electronic equipment inputs the acquired network state data to be processed of the first network element into a target type service quality prediction model, and service quality data corresponding to the first network element can be obtained.

For step S204, for each network layer, the electronic device may select, as the target network element, a network element with better quality of network service from the first network elements in the network layer based on the to-be-processed quality of service data corresponding to the first network elements in the network layer according to the actual requirement. For example, the network layer is the network layer where the base station is located, and the electronic device may select, from the base stations, a base station with the largest service quality data to be processed as the target network element.

For step S205, after determining each target network element between the target user end and the target resource end, for each target network element, the electronic device may determine whether communication connection is performed between the target network element and the next target network element, and if communication connection is not performed between the target network element and the next target network element, the electronic device may select, from other network elements in the network layer where the next target network element is located, one network element that is in communication connection with the target network element, and that has better quality of network service corresponding to the target user end, as the next target network element of the target network element. The next target network element of one target network element is: and among the network elements from the target user end to the target resource end, the target network element in the next network layer of the network layer where the target network element is located.

Furthermore, the electronic device may determine a network path formed by each target network element, which is a target network path between the target user end and the target resource end.

In some embodiments, the electronic device may further train the service quality prediction model of the preset structure based on the sample network state data of the sample user side of the target type and the sample service quality data corresponding to the sample network state data, so as to obtain the service quality prediction model corresponding to the target type.

Referring to fig. 4, fig. 4 is a flowchart of a method for generating a quality of service prediction model according to an embodiment of the disclosure, where the method may include the following steps:

s401: and acquiring sample network state data of a sample user terminal of the sample network element aiming at the target type and sample service quality data corresponding to the sample network state data.

S402: and inputting the sample network state data into a service quality prediction model of a preset structure to obtain service quality data corresponding to the sample network state data, wherein the service quality data is used as predicted service quality data.

S403: a loss value is calculated based on the predicted quality of service data and the sample quality of service data.

S404: and adjusting model parameters of the service quality prediction model of the preset structure based on the loss value, and continuing training until the service quality prediction model of the preset structure reaches convergence.

Sample network element sample network state data for sample user of target type: the sample network element transmits network state data of network resources requested by a sample user terminal of a target type, and sample service quality data corresponding to the sample network state data represents: based on the sample network state data of the sample network element, the quality of the network service provided by the corresponding sample user terminal.

For example, the target type is a video type, and the sample network state data of the sample user terminal of the sample network element for the target type may include: the sample network element transmits network state data of video resources requested by a sample user terminal of the video type, and sample service quality data corresponding to the sample network state data represents: based on the sample network state data of the sample network element, the quality of the video service provided by the corresponding sample user terminal.

The sample quality of service data corresponding to the sample network state data may be: the technician determines based on the service status data of each sample user terminal in the sample time period. The service status data of the sample user side can refer to the related description of the subsequent embodiments.

In some embodiments, based on fig. 4, referring to fig. 5, step S401 may include the steps of:

s4011: and acquiring sample network state data of a sample network element aiming at a sample user end of a target type in the sample time period aiming at each sample time period.

S4012: and acquiring sample service quality data of the sample user terminal in the sample time period, and taking the sample service quality data as sample service quality data corresponding to sample network state data of the sample user terminal of the sample network element aiming at the target type in the sample time period.

The duration of the sample time period can be set by a technician according to the requirement, and for each sample time period, the electronic device obtains sample network state data of the sample network element for the sample user terminal of the target type in the sample time period, and obtains sample service quality data of the sample user terminal in the sample time period, as sample service quality data corresponding to the sample network state data of the sample network element for the sample user terminal of the target type in the sample time period.

For example, if the sample period is 9:00 to 10:00, the electronic device obtains sample network state data of the sample network element for the sample user end of the target type in 9:00 to 10:00, and obtains sample service quality data of the sample user end in 9:00 to 10:00, so as to obtain sample service quality data corresponding to the sample network state data of the sample network element for the sample user end of the target type.

Based on the processing, the sample network state data of the sample network element aiming at the sample user end of the target type and the sample service quality data corresponding to the sample network state data of the sample network element aiming at the sample user end of the target type in the same sample time period can be obtained, further, the sample network state data and the sample service quality data corresponding to the sample network state data can be subjected to federal learning, the defect that the sample network state data can only be independently analyzed or the sample service quality data can only be independently analyzed in the prior art is overcome, the accuracy of the influence of the change of the network state data of the sensing network element on the change of the service quality data of the user end is improved, and the problem of sensing the quality of the network service corresponding to the user end by an application server is successfully solved, so that the application server can optimize the quality of the network service of the user end.

In some embodiments, step S4011 may comprise the steps of:

step 1: and acquiring network state data of a sample network element aiming at a sample user end of a target type in the sample time period as first network state data.

Step 2: and carrying out differential processing on the basis of the first network state data corresponding to the sample time period and the first network state data of the next sample time period of the sample time period to obtain sample network state data of a sample network element of the sample time period aiming at a sample user terminal of a target type.

The acquired first network state data may be expressed as a vector as follows: { I, T }: { A1: x0, x1, x2, … …, xn, A2: x0, X1, X2, … …, xn, an: x0, x1, x2, x3, … …, xn }.

Wherein I represents a source IP address and a destination IP address, T represents a sample period, an represents An nth sample network element, xn represents nth network state data of the sample network element.

After the first network state data corresponding to the sample time period and the first network state data of the next sample time period in the sample time period are obtained, differential processing can be performed, that is, a difference value between the first network state data of the next sample time period in the sample time period and the first network state data corresponding to the sample time period is calculated, so as to obtain sample network state data.

Based on the above processing, the security of the network state data of the sample network element can be improved by taking the first network state data after the differential processing as the sample network state data.

In some embodiments, step S4012 may comprise the steps of:

step one: and acquiring service state data of each sample user side in each sample time period according to each sample time period.

Step two: based on the clustering analysis of the service state data, service quality data represented by each service state data is obtained and used as first service quality data.

Step three: and exchanging the first service quality data corresponding to the two sample user terminals belonging to the same network segment aiming at the target sample time period to obtain the sample service quality data of the two sample user terminals in the target sample time period.

The target type is a video type, and the service state data of the sample user terminal may include: the total time of the user watching the video at the sample user terminal, the IP address of the sample user terminal, the total flow of video playing, the number of times of re-requesting the video resource by the sample user terminal, the number of times of clicking/switching by the user in playing, the interaction time delay from the sample user terminal to the corresponding resource terminal (which can be called as the sample resource terminal) and the like.

The target type is a game type, and the service state data of the sample user side may include: the time required by a user to log in a game at a sample user end, the time required by logging out of the game, the time of breaking in the game, the number of times of breaking, the interaction time delay from the sample user end to a sample resource end and the like.

The target type is a web page type, and the service state data of the sample user side may include: the first package time of the web page (namely, the time period from the initiation of the browsing request by the sample user terminal to the receipt of the response of the target resource terminal), the first screen time of the web page (namely, the time consumed by the sample user terminal for displaying the web page for the first time), the download rate of the web page, the success rate of opening the web page, and the like.

The clustering analysis can be performed on each service state data based on a preset clustering algorithm to obtain a plurality of service groups, wherein each service group comprises service state data of a plurality of sample user terminals. Further, a corresponding score may be assigned to each traffic packet by a technician based on the traffic state data in that traffic packet as traffic quality data for that traffic packet. The service quality data of each service packet, that is, the service quality data (i.e., the first service quality data) represented by the service status data in the service packet, corresponds to the same service quality data in the same packet. The preset clustering algorithm is selected as a KNN (k-Nearest Neighbor classification, k nearest neighbor clustering) algorithm.

The first quality of service data may be expressed, for example, in the form of the following vectors: first service packet: { I, T, W1}; a second service packet; { I, T, W2}; … …; nth traffic packet: { I, T, wn }.

I represents the IP address of the sample user end and the IP address of the sample resource end, T represents the sample time period, wn represents the service quality data of the nth service packet, and the service quality data of the nth service packet is the service quality data corresponding to each service state data in the nth service packet.

The target sample period may include all of the sample periods, or the target sample period may be a part of the sample periods in each sample period.

And in the target sample time period, exchanging the first service quality data corresponding to the two sample user terminals belonging to the same network segment to obtain the sample service quality data of the two sample user terminals in the target sample time period. Two sample clients belonging to the same network segment may include: two sample clients located in the same cell user.

For example, for two sample clients belonging to the same network segment, the first quality of service data of sample client 1 is expressed as: { I, T1, W1}, the first quality of service data of the sample client 2 is expressed as: { I, T, W2}. The first service quality data of the sample user terminal 1 is exchanged with the first service quality data of the sample user terminal 2, so that the sample service quality data of the sample user terminal 1 can be obtained and expressed as: { I, T1, W2}, the sample quality of service data for sample client 2 is expressed as: { I, T, W1}.

Based on the processing, the service quality data of different sample user terminals in the same network segment are exchanged, so that the privacy of the user data can be ensured, and the safety of the user data can be improved.

The electronic device may calculate a loss value representing a difference between the predicted quality of service data and the sample quality of service data based on the cross entropy loss function, and adjust model parameters of the quality of service prediction model of the preset structure according to a gradient descent manner based on the calculated loss value until the quality of service prediction model of the preset structure reaches convergence.

Illustratively, the weight vector containing model parameters of the quality of service prediction model may be noted as: w= (ωf, ω0, ω1, ω2, …, ωn), where ωn represents the nth model parameter of the quality of service prediction model. The electronic device obtains a user quality feature vector containing a plurality of sample network state data, and marks as: x= (xf, x0, x1, x2, …, xn), where xn represents the nth sample network state data, xf is a preset value, xf=1. The product of the user quality feature vector and the weight vector is calculated to obtain a feature variable, denoted as z=ω fxf +ω0x0+ω1x1+ω2x2+ … +ωnxn.

Then, a mapping function h (Z) is obtained, and the characteristic variable Z is mapped into a range from 0 to 1 through the mapping function h (Z), wherein the mapping function h (Z) is:

h (Z) represents a mapping function, Z represents a characteristic variable, and e represents a natural constant.

The user quality feature vector x is noted as: TX, let z=wtx, based on the above formula (1), the following formula (2) can be obtained:

h _ω (x) A mapping function with x and ω as variables is represented.

Let y denote the quality of service data, and correspondingly, if the quality of service data corresponding to the user quality feature vector is y, the confidence coefficient of the user quality feature vector for the quality of service data y is 1, and if the quality of service data corresponding to the user quality feature vector is not y, the confidence coefficient of the user quality feature vector for the quality of service data y is 0.

h _ω (x) The values of (2) represent: the confidence that the quality of service data corresponding to the user quality feature vector is y, therefore, for the user quality feature vector x, the confidence that the quality of service data corresponding to the user quality feature vector is y may be expressed as: p (y= 1|X; ω) =h _ω (x) The confidence that the quality of service data corresponding to the user quality feature vector is not y can be expressed as: p (y= 1|X; ω) =1-h _ω (x)。

Further, based on the maximum likelihood estimation algorithm, a loss function as shown in formula (3) is constructed:

j (ω) represents a loss function having ω as a variable; m represents the number of user quality feature vectors, y ⁽ⁱ⁾ Quality of service data representing an ith user quality feature vector; h is a _ω (x ⁽ⁱ⁾ ) A confidence level that the quality of service data representing the ith user quality feature vector is y; x is x ⁽ⁱ⁾ Representing the ith user quality feature vector.

The convergence condition of the service quality prediction model is as follows: finding ω when J (ω) takes the minimum value, each model parameter of the quality of service prediction model at convergence can be obtained. The electronic device may iteratively update ω based on the following equation (4) to determine model parameters of the quality of service prediction model at convergence.

ω _j+1 Model parameters of a service quality prediction model obtained by j+1st calculation are represented; omega _j Model parameters representing a service quality prediction model obtained by jth calculation; alpha represents a preset coefficient;representing the loss function J (ω) with respect to ω _j Is a partial derivative of (c). Wherein j is [0, k]K represents the number of iterations. Omega _j+1 Any one of the model parameters that may represent the quality of service prediction model may be iteratively updated according to the above equation (4).

Referring to fig. 6, fig. 6 is a schematic diagram of a training method of a quality of service prediction model according to an embodiment of the present disclosure. The central server is the electronic device in the embodiment of the present disclosure, the network data is the network status data in the embodiment of the present disclosure, and the application data includes: the service state data and the service quality data in the embodiment of the disclosure.

The sample network element may obtain network data of the sample network element, and perform local training on the network data to obtain a plurality of data subsets (i.e., sample network state data in the embodiment of the present disclosure). That is, the sample network element acquires network state data of the sample network element, and performs differential processing on the acquired network state data to obtain sample network state data. The sample network element then uploads the sample network state data to the central server in a distributed manner, i.e., the sample network element uploads the sample network state data to the central server in multiple times, e.g., the sample network element may upload sample network state data for one sample period at a time, or the sample network element may upload sample network state data for multiple sample periods at a time, i.e., the subset of data is the sample network state data uploaded each time.

After the sample user side obtains the application data of the target type, the application data is locally trained, namely the sample user side obtains service state data of the sample user side in a sample time period, clustering analysis is carried out on each service state data to obtain service quality data represented by each service state data as first service quality data, and then the sample user side exchanges the first service quality data corresponding to two sample user sides belonging to the same network section aiming at the target sample time period to obtain the sample service quality data of the two sample user sides in the target sample time period. Furthermore, the sample ue uploads the sample qos data to the central server, that is, the sample ue uploads the sample qos data to the central server multiple times, for example, the sample ue may upload the sample qos data in one sample period at a time, and the sample ue may upload the qos data in multiple sample periods at a time.

The central server performs subset model iteration according to the received sample network state data and sample service quality data, namely, the central server trains a preset structure service quality prediction model based on the sample network state data and the sample service quality data received each time, and a trained service quality prediction model is obtained.

Based on the processing, the service quality prediction model can be trained by a federal training method to remedy the problem of data defects existing in independent analysis of network data and application data, so that better perception of network and application quality is obtained through the existing data, and the privacy of the data can be protected by carrying out feature clustering and data confusion processing in the application data and the network data. And, the user session (i.e. the process of accessing the resource end by the user end) is used to associate the user end with the network element, so that each network element in the network path can be associated with the application quality of the user end, and the application quality of the user end can be optimized based on the network state data of each network element in the network path.

Corresponding to the method embodiment of fig. 2, referring to fig. 7, fig. 7 is a block diagram of a network path determining apparatus provided in an embodiment of the present disclosure, where the apparatus includes:

a first determining module 701, configured to determine network elements included in each network layer between a target user terminal and a corresponding target resource terminal as a first network element;

a first obtaining module 702, configured to obtain, for each first network element, network state data of a user side of the first network element for a target type, as network state data to be processed; wherein the target type represents the type of the target user terminal;

A first prediction module 703, configured to input the network state data to be processed into a service quality prediction model corresponding to the target type, to obtain service quality data corresponding to the first network element, as service quality data to be processed; the service quality prediction model is obtained by training sample network state data of a sample user terminal of the target type based on a sample network element and sample service quality data corresponding to the sample network state data; the sample network element is a network element in a network path between the sample user end and a corresponding resource end; the sample quality of service data represents: when the network state of the sample network element is the sample network state data, the quality of the network service provided by the sample user terminal;

a selecting module 704, configured to select, for each network layer, one network element from the network layers based on the to-be-processed quality of service data corresponding to the first network element in the network layer, as a target network element;

a second determining module 705, configured to determine a network path formed by each target network element as a target network path between the target ue and the target resource.

In some embodiments, the apparatus further comprises:

In some embodiments, the first obtaining module 702 is specifically configured to obtain, for each first network element, network state data of the first network element for a plurality of clients of a target type;

Based on the network path determining device provided by the embodiment of the present disclosure, for each network element included in each network layer between the target user terminal and the corresponding target resource terminal, corresponding service quality data (i.e., to-be-processed service quality data) may be determined based on the network state data of the network element, where the to-be-processed service quality data may represent the quality of the network service provided by the corresponding user terminal. Furthermore, based on the service quality data to be processed corresponding to each network element, the network element with better quality of the corresponding network service can be selected, and the network path formed by each selected network element is determined as the target network path between the target user terminal and the target resource terminal, so that the quality of the network service when the target user terminal accesses the target resource terminal can be improved.

The disclosed embodiment also provides an electronic device, as shown in fig. 8, comprising a processor 801, a communication interface 802, a memory 803, and a communication bus 804, wherein the processor 801, the communication interface 802, the memory 803 complete communication with each other through the communication bus 804,

A memory 803 for storing a computer program;

the processor 801, when executing the program stored in the memory 803, implements the following steps:

The communication bus mentioned above for the electronic devices may be a peripheral component interconnect standard (Peripheral Component Interconnect, PCI) bus or an extended industry standard architecture (Extended Industry Standard Architecture, EISA) bus, etc. The communication bus may be classified as an address bus, a data bus, a control bus, or the like. For ease of illustration, the figures are shown with only one bold line, but not with only one bus or one type of bus.

The communication interface is used for communication between the electronic device and other devices.

The Memory may include random access Memory (Random Access Memory, RAM) or may include Non-Volatile Memory (NVM), such as at least one disk Memory. Optionally, the memory may also be at least one memory device located remotely from the aforementioned processor.

The processor may be a general-purpose processor, including a central processing unit (Central Processing Unit, CPU), a network processor (Network Processor, NP), etc.; but also digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), field programmable gate arrays (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

In yet another embodiment provided by the present disclosure, there is also provided a computer readable storage medium having stored therein a computer program which when executed by a processor implements the steps of any of the network path determination methods described above.

In yet another embodiment provided by the present disclosure, there is also provided a computer program product containing instructions that, when run on a computer, cause the computer to perform any of the network path determination methods of the above embodiments.

In the above embodiments, it may be implemented in whole or in part 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, produces a flow or function in accordance with embodiments of the present disclosure, in whole or in part. The computer may be a general purpose computer, a special purpose computer, a computer network, or other programmable apparatus. The computer instructions may be stored in or transmitted from one computer-readable storage medium to another, for example, by wired (e.g., coaxial cable, optical fiber, digital Subscriber Line (DSL)), or wireless (e.g., infrared, wireless, microwave, etc.). The computer readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. that contains an integration of one or more available media. The usable medium may be a magnetic medium (e.g., floppy Disk, hard Disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid State Disk (SSD)), etc.

It is noted that relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.

In this specification, each embodiment is described in a related manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments. In particular, for apparatus, electronic devices, computer readable storage media and computer program product embodiments, the description is relatively simple as it is substantially similar to method embodiments, as relevant points are found in the partial description of method embodiments.

The foregoing description is only of the preferred embodiments of the present disclosure, and is not intended to limit the scope of the present disclosure. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present disclosure are included in the protection scope of the present disclosure.

Claims

1. A method of network path determination, the method comprising:

2. The method according to claim 1, wherein the training process of the quality of service prediction model comprises the steps of:

3. The method according to claim 2, wherein the obtaining sample network state data of the sample network element for the sample user side of the target type and sample service quality data corresponding to the sample network state data includes:

4. A method according to claim 3, wherein the sample quality of service data for the sample user side during each sample period is generated by:

5. A method according to claim 3, wherein the sample network state data of the sample network element for the sample user side of the target type in each sample period is generated by:

6. The method of claim 1, wherein the obtaining, for each first network element, network state data of the first network element for the target type of the user terminal as the network state data to be processed includes:

7. A network path determination apparatus, the apparatus comprising:

8. The apparatus of claim 7, wherein the apparatus further comprises:

9. The apparatus of claim 8, wherein the second obtaining module is specifically configured to obtain, for each sample period, sample network state data of the sample network element for the sample user side of the target type in the sample period;

10. The apparatus of claim 9, wherein the second obtaining module is specifically configured to obtain, for each sample period, service state data of each sample user terminal in the sample period;

11. The apparatus of claim 9, wherein the second obtaining module is specifically configured to obtain, for each sample period, network state data of the sample network element for the sample user terminal of the target type in the sample period as the first network state data;

12. The apparatus of claim 7, wherein the first obtaining module is specifically configured to obtain, for each first network element, network state data of the first network element for a plurality of clients of a target type;

13. The electronic equipment is characterized by comprising a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory are communicated with each other through the communication bus;

A memory for storing a computer program;

a processor for carrying out the method steps of any one of claims 1-6 when executing a program stored on a memory.

14. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored therein a computer program which, when executed by a processor, implements the method steps of any of claims 1-6.