Cluster distribution method and device, equipment, storage medium and program product

Abstract

The invention discloses a cluster allocation method and device, equipment, a storage medium and a program product, and relates to the technical field of computers. The method comprises the following steps: acquiring a plurality of pieces of communication quality data; grouping the plurality of pieces of communication quality data by integrating at least one analysis dimension and an edge cluster to obtain a plurality of communication quality data groups; performing data processing on at least one piece of communication quality data in the same communication quality data group to obtain quality prediction data corresponding to the plurality of communication quality data groups; and constructing a resource allocation data table through the quality prediction data and the at least one analysis dimension. Through the above mode, the communication service is provided for the same edge cluster by means of the at least one analysis dimension, and the quality prediction is carried out on the plurality of communication quality data groups with the same analysis dimension, so that the more appropriate edge cluster is selected for the terminal through the quality prediction data to provide the communication service, and the data processing efficiency is improved. The method can be applied to various scenes such as cloud technology, artificial intelligence and intelligent transportation.

Description

Technical Field

[0001] Embodiments of the present application relate to the field of computer technology, and in particular, to a cluster allocation method, device, equipment, storage medium, and program product. Background Art

[0002] With the development of computer technology, the communication between terminals and servers has become increasingly frequent. In order to further improve communication efficiency, during the deployment of servers, the concept of edge computing is introduced on the original centralized architecture. Edge computing is realized through edge clusters close to terminals, and the edge clusters include multiple edge nodes for data processing.

[0003] In related technologies, the server platform selects an edge cluster with better performance for the terminal to provide communication services according to the actual network conditions. For example: the terminal can obtain test addresses corresponding to multiple regions from the server platform for network speed testing. After that, the terminal returns the test results to the server platform. The server platform selects the region with a higher network speed according to the test results, and allocates a computing node from the edge cluster deployed in this region to provide communication services for the terminal.

[0004] In the above process, although it is highly likely to select an edge node with better service speed for the terminal according to the test results, the process of the terminal determining the test results in real time based on the test addresses may obtain inaccurate values due to short-term network fluctuations, thereby affecting the accuracy of edge cluster determination; in addition, the process of simply relying on the network speed test results to select an edge cluster also has strong limitations. Summary of the Invention

[0005] Embodiments of the present application provide a cluster allocation method, device, equipment, storage medium, and program product, which can perform quality prediction on multiple communication quality data groups that provide communication services for the same edge cluster and have the same analysis dimension by means of at least one analysis dimension, so as to select a more appropriate edge cluster for the terminal to provide communication services through the quality prediction data, thereby improving the processing efficiency when the terminal processes data. The technical solution is as follows.

[0006] On the one hand, a cluster allocation method is provided, and the method includes:

[0007] Obtain multiple communication quality data, where the communication quality data is data statistically obtained from at least one analysis dimension for the communication between the terminal and the edge cluster during a historical time period, and the edge cluster is a cluster that provides communication services for the terminal based on the at least one analysis dimension;

[0008] Group the multiple communication quality data by combining the at least one analysis dimension and the edge clusters, to obtain multiple communication quality data groups that provide the communication service by the same edge cluster and have the same analysis dimension;

[0009] Perform data processing on at least one communication quality data within the same communication quality data group, to obtain quality prediction data corresponding to the multiple communication quality data groups respectively, where there is an association relationship between the quality prediction data and the edge clusters;

[0010] Construct a resource allocation data table through the quality prediction data and the at least one analysis dimension; wherein, the resource allocation data table is used to allocate an edge cluster that provides the communication service for a prediction terminal according to at least one analysis dimension corresponding to the prediction terminal and the association relationship.

[0011] On the other hand, a cluster allocation device is provided, and the device includes:

[0012] An acquisition module, configured to acquire multiple communication quality data, where the communication quality data is data statistically obtained from at least one analysis dimension when a terminal communicates with an edge cluster within a historical time period, and the edge cluster is a cluster that provides a communication service for the terminal based on the at least one analysis dimension;

[0013] A grouping module, configured to group the multiple communication quality data by combining the at least one analysis dimension and the edge clusters, to obtain multiple communication quality data groups that provide the communication service by the same edge cluster and have the same analysis dimension;

[0014] A processing module, configured to perform data processing on at least one communication quality data within the same communication quality data group, to obtain quality prediction data corresponding to the multiple communication quality data groups respectively, where there is an association relationship between the quality prediction data and the edge clusters;

[0015] A construction module, configured to construct a resource allocation data table through the quality prediction data and the at least one analysis dimension; wherein, the resource allocation data table is used to allocate an edge cluster that provides the communication service for a prediction terminal according to at least one analysis dimension corresponding to the prediction terminal and the association relationship.

[0016] On the other hand, a computer device is provided, the computer device includes a processor and a memory, and at least one instruction, at least one program, a code set or an instruction set is stored in the memory, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by the processor to implement the cluster allocation method as described in any one of the above embodiments of the present application.

[0017] On the other hand, a computer-readable storage medium is provided. At least one instruction, at least one program, a code set, or an instruction set is stored in the storage medium. The at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by a processor to implement the cluster allocation method as described in any one of the embodiments of the present application above.

[0018] On the other hand, a computer program product or a computer program is provided. The computer program product or the computer program includes computer instructions. The computer instructions are stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the cluster allocation method as described in any one of the above embodiments.

[0019] The beneficial effects brought by the technical solutions provided in the embodiments of the present application at least include:

[0020] Group the obtained multiple communication quality data by integrating at least one analysis dimension and edge clusters. Then, perform data processing on the communication quality data within the same communication quality data group to obtain multiple quality prediction data. Construct a resource allocation data table through the quality prediction data and at least one analysis dimension. Thus, allocate an edge cluster that provides communication services for the prediction terminal according to at least one analysis dimension corresponding to the prediction terminal. Make full use of at least one analysis dimension to obtain multiple communication quality data groups that are provided with communication services by the same edge cluster and have the same analysis dimension. Furthermore, based on data processing, obtain quality prediction data that is convenient for predicting subsequent communication quality situations. By constructing a resource allocation data table, the corresponding quality prediction data can be determined according to at least one analysis dimension, so as to select a corresponding edge cluster to provide communication services under the numerical condition of the quality prediction data, improve the allocation accuracy of edge clusters by means of a multi-level analysis process, and facilitate more efficient data interaction of the terminal based on the edge cluster. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0022] Figure 1 is a schematic diagram of an implementation environment provided by an exemplary embodiment of the present application;

[0023] Figure 2 is a flowchart of a cluster allocation method provided by an exemplary embodiment of the present application;

[0024] Figure 3It is a flowchart of a cluster allocation method provided by another exemplary embodiment of the present application;

[0025] Figure 4 It is a schematic diagram of determining transmission delay provided by an exemplary embodiment of the present application;

[0026] Figure 5 It is a flowchart of a cluster allocation method provided by still another exemplary embodiment of the present application;

[0027] Figure 6 It is a flowchart of a cluster allocation method provided by yet another exemplary embodiment of the present application;

[0028] Figure 7 It is a schematic diagram of obtaining and applying a cluster selection model provided by an exemplary embodiment of the present application;

[0029] Figure 8 It is a schematic diagram of determining an edge cluster score based on a cluster selection model provided by an exemplary embodiment of the present application;

[0030] Figure 9 It is a flowchart of a cluster allocation method provided by another exemplary embodiment of the present application;

[0031] Figure 10 It is a schematic diagram of allocating a predicted terminal to an edge cluster based on a preset allocation probability provided by an exemplary embodiment of the present application;

[0032] Figure 11 It is a schematic diagram of related technology provided by an exemplary embodiment of the present application;

[0033] Figure 12 It is a schematic diagram of related technology provided by still another exemplary embodiment of the present application;

[0034] Figure 13 It is a schematic diagram of an allocation record of allocating an edge cluster to a terminal provided by an exemplary embodiment of the present application;

[0035] Figure 14 It is an overall flowchart of a cluster allocation method provided by an exemplary embodiment of the present application;

[0036] Figure 15 It is a schematic diagram of applying a cluster selection model provided by an exemplary embodiment of the present application;

[0037] Figure 16 It is a structural block diagram of a cluster allocation device provided by an exemplary embodiment of the present application;

[0038] Figure 17 It is a structural block diagram of a cluster allocation device provided by still another exemplary embodiment of the present application;

[0039] Figure 18 It is a structural block diagram of a server provided by an exemplary embodiment of the present application. Specific embodiments

[0040] To make the objectives, technical solutions and advantages of the present application clearer, the embodiments of the present application will be further described in detail below with reference to the accompanying drawings.

[0041] In the related art, the server platform selects an edge cluster with better performance for the terminal according to the actual network conditions. For example, the terminal can obtain test addresses corresponding to multiple regions from the server platform for network speed testing. After that, the terminal returns the test results to the server platform. The server platform selects the region with a higher network speed according to the test results, and allocates a computing node from the edge cluster deployed in that region to provide communication services for the terminal. In the above process, although it is highly probable to select an edge node with better service speed for the terminal according to the test results, the process of the terminal determining the test results in real time based on the test addresses may obtain inaccurate values due to short-term network fluctuations, thereby affecting the accuracy of the determination of the edge cluster. In addition, the process of simply relying on the network speed test results to select the edge cluster also has strong limitations.

[0042] In the embodiments of the present application, a cluster allocation method is introduced. It can perform quality prediction on multiple communication quality data groups that provide communication services for the same edge cluster and have the same analysis dimension by means of at least one analysis dimension, so as to select a more appropriate edge cluster for the terminal to provide communication services through the quality prediction data, thereby improving the processing efficiency when the terminal processes data. The cluster allocation method can be applied to multiple usage scenarios such as the Internet of Things field, the intelligent retail field, the industrial automation field, and the intelligent transportation field. The embodiments of the present application do not limit this.

[0043] It should be noted that the information (including but not limited to user equipment information, user personal information, etc.), data (including but not limited to data for analysis, stored data, displayed data, etc.) and signals involved in the present application are all authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data need to comply with the relevant laws, regulations and standards of relevant countries and regions. For example, the communication quality data and other content involved in the present application are obtained under full authorization.

[0044] Secondly, the implementation environment involved in the embodiments of the present application is described. The cluster allocation method provided by the embodiments of the present application can be implemented by the terminal alone, or by the server, or by data interaction between the terminal and the server. The embodiments of the present application do not limit this. Optionally, taking the interaction between the terminal and the server to execute the cluster allocation method as an example for illustration.

[0045] For illustration, please refer to Figure 1 The implementation environment involves a terminal 110 and a server 120 , and the terminal 110 and the server 120 are connected via a communication network 130 .

[0046] In some embodiments, the terminal 110 has a data collection function, and the terminal 110 can obtain communication quality data based on the data collection function; or, the terminal 110 can obtain communication quality data by classifying and aggregating the data collected by the data collection function.

[0047] The communication quality data is data obtained from statistics when the terminal communicates with the edge cluster in a historical time period from at least one analysis dimension; the edge cluster is a cluster that provides communication services to the terminal based on at least one analysis dimension.

[0048] Schematically, when the terminal is running an application, it will interact with the server corresponding to the application. The server includes a central cluster and an edge cluster. The central cluster has the ability to coordinate and manage, and can process data interaction requests sent by multiple terminals in a decentralized manner. For example, the central cluster can use the deployed edge cluster to decentralizedly process data interaction requests corresponding to multiple terminals. When the terminal communicates with the edge cluster, communication quality data can be statistically obtained from at least one analysis dimension.

[0049] In some embodiments, the terminal 110 sends the communication quality data to the server 120 through the communication network 130. The server 120 groups the multiple communication quality data according to at least one analysis dimension and the edge cluster to obtain multiple communication quality data groups provided with communication services by the same edge cluster and having the same analysis dimension.

[0050] Illustratively, the server 120 is a device in the above-mentioned central cluster. In addition to comprehensively determining the edge cluster for the terminal, the server 120 can also group multiple communication quality data obtained in a historical time period to obtain multiple communication quality data groups.

[0051] The communication quality data in each communication quality data group corresponds to the same terminal, the same edge cluster and has the same analysis dimension. That is, after grouping, multiple communication quality data groups with communication services provided by the same edge cluster and the same analysis dimension are obtained.

[0052] In some embodiments, the server 120 further performs data processing on at least one communication quality data in the same communication quality data group to obtain quality prediction data corresponding to the plurality of communication quality data groups.

[0053] Among them, there is a correlation between quality prediction data and edge clusters.

[0054] Schematically, the data processing is implemented as at least one of multiple data processing methods such as mean processing, median processing, variance processing, etc. After obtaining multiple communication quality data groups, each communication quality data group is analyzed separately, and quality prediction data is obtained based on at least one communication quality data within the same communication quality data group, so as to obtain quality prediction data corresponding to multiple communication quality data groups respectively.

[0055] Based on the fact that the communication quality data within the communication quality data group belongs to the same edge cluster, there is an association relationship between the quality prediction data determined based on the same communication quality data group and the edge cluster.

[0056] In some embodiments, the server 120 also constructs a resource allocation data table through the quality prediction data and at least one analysis dimension.

[0057] Among them, the resource allocation data table is used to allocate an edge cluster that provides communication services for the prediction terminal according to at least one analysis dimension corresponding to the prediction terminal and the association relationship.

[0058] Schematically, the communication quality data within the communication quality data group also has the same analysis dimension. Therefore, a resource allocation data table can be constructed through the quality prediction data and at least one analysis dimension, and the quality prediction data in the resource allocation data table is mapped to the edge cluster through the association relationship.

[0059] With the constructed resource allocation data table, a query process can be carried out. For example: currently, the terminal that needs to allocate an edge cluster is a prediction terminal, and the prediction terminal corresponds to at least one analysis dimension. The resource allocation data table is queried according to at least one analysis dimension corresponding to the prediction terminal to obtain the quality prediction data corresponding to multiple edge clusters respectively, and then an edge cluster that provides communication services for the prediction terminal is allocated according to the quality prediction data.

[0060] Optionally, the edge cluster includes multiple edge nodes that provide computing resources, and each edge node corresponds to at least one computing device. When allocating an edge cluster for the prediction terminal, at least one edge node within the edge cluster is allocated to the prediction terminal to process data interaction requests corresponding to the prediction terminal through at least one edge node, etc.

[0061] Optionally, the server 120 sends the cluster identifier corresponding to the edge cluster to the terminal 110 through the communication network 120, so that the terminal 110 can send the data to be processed to the edge nodes within the edge cluster for data processing, and the server 120 can send the data processing result to the terminal 110 through the communication network 130, etc.

[0062] It should be noted that the above-mentioned terminals include, but are not limited to, mobile terminals such as mobile phones, tablet computers, portable laptops, intelligent voice interaction devices, intelligent household appliances, vehicle-mounted terminals, etc., and can also be implemented as desktop computers, etc.; the above-mentioned servers can be independent physical servers, or a server cluster or distributed system composed of multiple physical servers, or can also be cloud servers that provide basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, Content Delivery Network (CDN), as well as big data and artificial intelligence platforms.

[0063] Among them, cloud technology refers to a hosting technology that unifies a series of resources such as hardware, application programs, and networks within a wide area network or a local area network to achieve data calculation, storage, processing, and sharing. Cloud technology is the general term for network technology, information technology, integration technology, management platform technology, application technology, etc. based on the cloud computing business model, and can form a resource pool, which can be used on demand and is flexible and convenient.

[0064] In some embodiments, the above-mentioned server can also be implemented as a node in a blockchain system.

[0065] Combined with the above noun introduction and application scenarios, the cluster allocation method provided by this application will be described. Taking this method applied to a server as an example, as Figure 2 shown, this method includes the following steps 210 to step 240.

[0066] Step 210, obtain multiple communication quality data.

[0067] Schematically, communication quality data is data used to evaluate and describe performance indicators in the communication process. Optionally, the communication quality data is implemented as at least one of multiple data such as transmission delay, packet loss rate, jitter rate, signal quality, etc.

[0068] Among them, the communication quality data is data statistically obtained from at least one analysis dimension when the terminal communicates with the edge cluster during a historical time period.

[0069] Optionally, the analysis dimension is used to describe the dimension that affects the communication quality data. Statistical analysis through at least one analysis dimension is used to indicate that the size of the communication quality data can be determined through a single dimension or multiple dimensions.

[0070] Schematically, the analysis dimension is implemented as at least one of the terminal attribute dimension and the program running dimension. The terminal attribute dimension is used to indicate the dimension that describes terminal-related data, and the program running dimension is used to indicate the dimension that describes the relevant data generated by the terminal during the running of the application program.

[0071] The terminal attribute dimension includes at least one of the terminal location, terminal operator, terminal network type, and terminal model. The terminal location is used to describe the region where the terminal is located (e.g., the terminal is located in P); the terminal operator is used to describe the platform that provides mobile communication services to the terminal (e.g., the terminal uses the y card provided by the Y platform to access the Internet, and the Y platform is the terminal operator); the terminal network type is used to describe the type of communication network used by the terminal during network communication, such as: wireless network (Wireless Fidelity, Wi-Fi), fifth generation mobile communication technology (5G), fourth generation mobile communication technology (4G), wireless local area network (WLAN), cellular network (Cellular Network), satellite network (Satellite Network), etc.; the terminal model is used to describe the specific model or model name of the terminal, such as: B brand b1 model mobile phone, O brand o3 model computer, etc.

[0072] The program running dimension is the information described in the process of running the application on the terminal. For example, when the terminal opens the installed application A to run the application A, when the user operates the application A based on the terminal, an operation request may be generated to the application background of the application A, and at least one of the contents such as the number of online users of the application corresponding to the application and the compatibility between the application and the terminal is used as the program running dimension.

[0073] Optionally, the communication quality data is data of performance indicators of a communication process between the terminal and the edge cluster, and the edge cluster is a cluster that provides communication services to the terminal based on at least one analysis dimension.

[0074] Schematically, the terminal runs an application, and the application runs the background program through a distributed cluster. The distributed cluster includes a central cluster for overall management and multiple edge clusters. The central cluster is used to process data or determine the edge cluster for processing data requests after the terminal initiates a data request through the application, and execute the data processing process through the edge cluster.

[0075] The central cluster includes at least one server, and the at least one server includes a server executing the embodiment of the present application; therefore, the central cluster can obtain the communication quality data corresponding to the terminal.

[0076] For example, the terminal sends a data request to the central cluster, and the central cluster can allocate an edge cluster for the terminal from multiple edge clusters according to the communication quality data corresponding to the terminal. This part of the content will be introduced in Nie Rong as follows.

[0077] During a historical time period, the edge cluster providing communication services for the terminal is known. For example, at time 1 in the historical time period, the edge cluster a1 deployed in area A provides communication services for the terminal. Then, the communication quality data corresponding to the terminal at time 1 characterizes the communication performance between the terminal and the edge cluster a1, etc.

[0078] In some embodiments, multiple communication quality data are data corresponding to at least one terminal.

[0079] For example, multiple communication quality data include communication quality data 1, communication quality data 2, and communication quality data 3. Communication quality data 1, communication quality data 2, and communication quality data 3 all correspond to terminal A; or, communication quality data 1 and communication quality data 2 correspond to terminal A, and communication quality data 3 corresponds to terminal B, etc.

[0080] Schematically, based on the fact that the communication quality data is data statistically obtained from at least one analysis dimension when the terminal communicates with the edge cluster during the historical time period, the communication quality data is not only related to the edge cluster but also related to at least one analysis dimension.

[0081] For example, communication quality data 1, communication quality data 2, and communication quality data 3 all correspond to terminal A, describing the situation in different sub-time periods within the historical time period.

[0082] As shown in Table 1 below, communication quality data 1 characterizes that at sub-time period 1, the terminal location is in area P, the terminal operator is platform Y, and the terminal network type is (three analysis dimensions). Terminal A is provided with communication services by the edge cluster p1 in area P, and the value of communication quality data 1 is 0.23 ms, characterizing the transmission delay; communication quality data 2 characterizes that at sub-time period 2, the terminal location is in area P, the terminal operator is platform Y, and the terminal network type is 4G (5G is switched to 4G). Terminal A is provided with communication services by the edge cluster w1 in area W, and the value of communication quality data 2 is 0.44 ms; communication quality data 3 characterizes that at sub-time period 3, the terminal location is in area P, the terminal operator is platform D, and the terminal network type is WiFi (switched to the wireless network provided by other platforms). Terminal A is provided with communication services by the edge cluster p2 in area P, and the value of communication quality data 3 is 0.32 ms, etc.

[0083] Table 1

[0084]

[0085]

[0086] Similarly, when multiple communication quality data correspond to at least one terminal, the value of the communication quality data, the corresponding analysis dimension, and the represented edge cluster can also be characterized by referring to the above method. As shown in the example in Table 2, Terminal A communicates using the 5G network of Platform Y in the first 10 minutes of the application running, and the edge cluster providing communication services for Terminal A is p1. In the next 15 minutes of the application running, it communicates using the WiFi network of Platform L, and the edge cluster providing communication services for Terminal A is s1. Similarly, Terminal B and Terminal C also have corresponding characterization contents such as communication quality data, which will not be elaborated here.

[0087] Table 2

[0088] Terminal Communication quality data Terminal location Terminal operator Terminal network type Edge cluster A 1: 0.23ms P region Y platform 5G p1 A 2: 0.44ms P region L platform WiFi s1 B 3: 0.29ms P region Y platform 5G p1 C 4: 0.32ms S region L platform WiFi w2 …… …… …… …… …… ……

[0089] It should be noted that the implementation of the above at least one analysis dimension as the terminal location, terminal operator, and terminal network type is only a schematic example, and the embodiments of the present application are not limited thereto.

[0090] Step 220: Group multiple communication quality data based on at least one analysis dimension and the edge cluster to obtain multiple communication quality data groups that are provided with communication services by the same edge cluster and have the same analysis dimension.

[0091] Schematically, each communication quality data corresponds to at least one analysis dimension and the edge cluster. The multiple communication quality data come from multiple terminals. Group the multiple communication quality data based on at least one analysis dimension and the edge cluster to summarize at least one communication quality data that is provided with communication services by the same edge cluster and has the same analysis dimension into one communication quality data group.

[0092] As shown in Table 2, the communication quality data 1 corresponding to Terminal A and the communication quality data 2 corresponding to Terminal B are provided with services by the same edge cluster (edge cluster p1) and have the same analysis dimension (both corresponding to Region P, Platform Y, 5G). Therefore, the communication quality data 1 and the communication quality data 2 can be summarized into one communication quality data group. Similarly, the grouping and summarization process can be performed on other communication quality data in Table 2 to obtain multiple communication quality data groups.

[0093] Among them, each communication quality data group includes at least one communication quality data. The at least one communication quality data within the group is provided with communication services by the same edge cluster and has the same analysis dimension, and the analysis situations corresponding to the analysis dimension are also the same.

[0094] Step 230: Process at least one communication quality data within the same communication quality data group to obtain quality prediction data corresponding to multiple communication quality data groups respectively.

[0095] Schematically, after obtaining multiple communication quality data groups, analyze the multiple communication quality data groups respectively. During the analysis process, process at least one communication quality data within the same communication quality data group.

[0096] Among them, the data processing is implemented as at least one of multiple processing methods such as mean processing, median processing, 70th percentile value processing, 90th percentile value processing, variance processing, etc.

[0097] Optionally, taking mean processing as an example of the data processing method, perform mean processing on at least one communication quality data within the same communication quality data group to obtain mean data corresponding to multiple communication quality data groups respectively as quality prediction data.

[0098] For example: As shown in Table 2, the communication quality data 1 corresponding to terminal A and the communication quality data 2 corresponding to terminal B are provided by the same edge cluster and have the same analysis dimension. Therefore, the communication quality data 1 and the communication quality data 2 can be classified into one communication quality data group. When processing at least one communication quality data within this communication quality data group, perform a mean operation on the communication quality data 1 and the communication quality data 2 to obtain mean data 0.26 = (0.23 + 0.29) / 2 as quality prediction data.

[0099] Similarly, data processing can be performed on other communication quality data groups respectively, so as to obtain quality prediction data based on at least one communication quality data within the same communication quality data group, that is, obtain multiple quality prediction data, and there is a one-to-one correspondence between the multiple quality prediction data and the multiple communication quality data groups.

[0100] Based on the correspondence between one communication quality data group, one edge cluster, and the same analysis dimension, this quality prediction data also corresponds to this edge cluster and the same analysis dimension. That is: there is an association relationship between the quality prediction data and the edge cluster.

[0101] For example: Based on Table 2, the quality prediction data 0.26 corresponds to the edge cluster p1 and three identical analysis dimensions, expressed as: quality prediction data 0.26 - edge cluster p1 - P region - Y platform - 5G; similarly, based on the communication quality data 4 corresponding to terminal C in Table 2, it is determined that the quality prediction data 0.32 corresponds to the edge cluster w2 and three analysis dimensions, expressed as: quality prediction data 0.32 - edge cluster w2 - S region - L platform - WiFi, etc. This is only a schematic example here.

[0102] Through the process of obtaining quality prediction data, it is possible to comprehensively analyze the multiple communication quality data of multiple terminals obtained within a historical time period under the limitation of at least one analysis dimension and an edge cluster, improve the prediction accuracy of the quality prediction data, and facilitate a more comprehensive understanding of the group association relationship with the same analysis dimension and the same edge cluster.

[0103] Step 240: Construct a resource allocation data table based on the quality prediction data and at least one analysis dimension.

[0104] Illustratively, after obtaining the quality prediction data, a resource allocation data table is constructed by integrating the quality prediction data and at least one analysis dimension, where at least one analysis dimension is used to assist the terminal in making a resource allocation judgment. For example: Taking at least one analysis dimension including terminal location, terminal operator, terminal network type, and quality prediction data as an example, a resource allocation data table as shown in Table 3 below is established by integrating multiple quality prediction data and the corresponding at least one analysis dimension.

[0105] Table 3

[0106] Terminal location Terminal operator Terminal network type Quality prediction data P region Y platform 5G 1：0.26 P region Y platform 5G 2：0.48 P region Y platform 5G 3：0.21 S region L platform WiFi 4：0.32 …… …… …… ……

[0107] Among them, in the case where the terminal location is region P, the terminal operator is platform Y, and the terminal network type is 5G, there are 3 quality prediction data, namely quality prediction data 1, quality prediction data 2, and quality prediction data 3. Different quality prediction data have an association relationship with different edge clusters. For example: There is an association relationship between quality prediction data 1 and edge cluster p1, an association relationship between quality prediction data 2 and edge cluster p2, an association relationship between quality prediction data 3 and edge cluster w1, etc.

[0108] That is to say, when constructing the resource allocation data table, the association relationship between the quality prediction data and the edge cluster can also be used as a construction basis. That is: A resource allocation data table is constructed through the quality prediction data, at least one analysis dimension, and the edge cluster.

[0109] Illustratively, as shown in Table 4 below, it is another expression of the resource allocation data table, where the association relationship between the quality prediction data and the edge cluster is more intuitively reflected in the resource allocation data table.

[0110] Table 4

[0111]

[0112]

[0113] Among them, the resource allocation data table is used to allocate an edge cluster that provides communication services for a prediction terminal according to at least one analysis dimension corresponding to the prediction terminal and the association relationship.

[0114] Schematically, the prediction terminal is any terminal. The resource allocation data table, as a data table summarized based on communication quality data within a historical time period, can to a certain extent reflect the effect of allocating edge clusters to terminals through at least one analysis dimension within the historical time period.

[0115] Therefore, when it is necessary to allocate an edge cluster for a prediction terminal, at least one analysis dimension corresponding to the prediction terminal can be first determined, so as to query the resource allocation data table with at least one dimension, and determine at least one quality prediction data that meets at least one dimension, and determine the edge clusters that are respectively associated with at least one quality prediction data, so as to determine at least one edge cluster, and at least one edge cluster corresponds to at least one quality prediction data one by one.

[0116] Schematically, at least one analysis dimension corresponding to the terminal is obtained based on the Internet Protocol (IP) address of the terminal; or, the terminal sends at least one analysis dimension to a central cluster and the like in the form of data.

[0117] In some embodiments, an edge cluster allocated for the prediction terminal is selected from at least one edge cluster based on at least one quality prediction data.

[0118] Schematically, if the terminal location of the prediction terminal is region P, the terminal operator is platform Y, and the terminal network type is 5G, the resource allocation data table can be queried based on these three analysis dimensions, and the quality prediction data that meets the three analysis dimensions determined therefrom includes quality prediction data 1, quality prediction data 2, and quality prediction data 3, so as to determine at least one edge cluster as edge cluster p1, edge cluster p2, and edge cluster w1.

[0119] Optionally, the edge cluster associated with the smallest quality prediction data is used as the edge cluster allocated for the prediction terminal.

[0120] For example: Based on the values corresponding to quality prediction data 1, quality prediction data 2, and quality prediction data 3 respectively, edge cluster w2 associated with quality prediction data 3 is used as the edge cluster allocated for the prediction terminal.

[0121] Optionally, the edge cluster includes multiple edge nodes, and each edge node consists of at least one computing device. The computing device / edge node is used to provide computing resources for the terminal to perform the data processing process. When allocating an edge cluster to the prediction terminal, an edge node is selected from the edge cluster and allocated to the prediction terminal, so as to perform the data processing process on the data that the prediction terminal needs to process through this edge node.

[0122] It should be noted that the above is only a schematic example, and the embodiments of the present application are not limited thereto.

[0123] In summary, by fully leveraging at least one analysis dimension, multiple communication quality data groups provided by the same edge cluster and having the same analysis dimension are obtained. Furthermore, quality prediction data that is convenient for predicting the subsequent communication quality situation is obtained based on data processing. By constructing a resource allocation data table, the corresponding quality prediction data can be determined according to at least one analysis dimension, so as to select the corresponding edge cluster to provide communication services in the case of the numerical value of the quality prediction data, and the allocation accuracy of the edge cluster is improved by means of a multi-level analysis process, which is convenient for the terminal to perform more efficient data interaction based on the edge cluster.

[0124] In an optional embodiment, the terminal collects network quality data and feeds it back to the central cluster. The central cluster processes the network quality data based on at least one analysis dimension and the edge cluster to obtain communication quality data. Schematically, as Figure 3 shown, step 210 shown above can also be implemented as steps 310 to 330 as follows. Figure 2

[0125] Step 310, obtain multiple network quality data.

[0126] Among them, the network quality data is data collected from at least one analysis dimension when the terminal communicates with the edge cluster at a historical moment; the historical moment is a moment within a historical time period.

[0127] Schematically, the network quality data is the data collected, and the communication quality data is the data obtained after processing the network quality data.

[0128] In some embodiments, taking the network quality data being implemented as the transmission delay as an example.

[0129] The transmission delay can also be called the Round-Trip Time (RTT). The transmission delay can be calculated according to the time information carried in the Sender Report corresponding to the sending end and the Receive Report corresponding to the receiving end.

[0130] ​Schematically, such as Figure 4 shown, is a schematic diagram for determining the transmission delay, which is realized through data interaction between the sending end 410 and the receiving end 420. The sending end 410 is implemented as a terminal, and the receiving end 420 is implemented as an edge cluster that provides communication services for the terminal. The sending end 410 sends a Sender Report to the receiving end 420 at time T0, and the receiving end 420 receives the Sender Report at time t0; the receiving end 420 sends a Receive Report to the sending end 410 at time t1, and the sending end 410 receives the Receive Report at time T1. RTT = T1 - T0 - (t1 - t0), where t1 - t0 is the time interval between the receiving end 420 receiving the Sender Report and sending the Receiver Report, which needs to be subtracted.

[0131] Optionally, the terminal obtains the transmission delay based on data interaction with the edge cluster; each terminal can obtain at least one transmission delay after data interaction with at least one edge cluster based on the process such as Figure 4 shown.

[0132] In some embodiments, multiple terminals respectively collect network quality data and send the network quality data to the central cluster, so that the central cluster obtains multiple network quality data.

[0133] Schematically, the central cluster is used to overall manage multiple edge clusters, and the central cluster is a cluster that docks with multiple terminals in a distributed architecture. For example, the central cluster is implemented as a background program corresponding to an application installed on the terminal, etc.

[0134] Among them, each of the multiple network quality data corresponds to an instance identifier, and the instance identifier is used to represent the identifier of the edge cluster providing communication services for the terminal based on at least one analysis dimension.

[0135] Schematically, the network quality data sent by the terminal is received based on a preset duration. That is: the network quality data is data that the terminal periodically sends to the central cluster. For example: the preset duration is at intervals of 1 second, 1 minute, etc.

[0136] Each network quality data corresponds to an instance identifier, and the network quality data of the same terminal, the same analysis dimension, and provided by the same edge cluster for communication services has the same instance identifier. That is: the instance identifier can be used to reflect the allocation situation of the instance, and the instance is used to represent the computing resources provided for the terminal based on at least one analysis dimension, and the computing resources are the content provided by the edge cluster for the terminal.

[0137] As shown in Table 5 below, it is an example of some network quality data collected by a terminal.

[0138] Table 5

[0139]

[0140] Among them, the timestamp is used to represent the moment when the network quality data is collected, and the timing is carried out in the way of Coordinated Universal Time (UTC), that is, the number of seconds elapsed since 00:00:00 on January 1, 1970. The timestamp 1692075318 represents 02:15:18 on January 13, 2023, the timestamp 1692075328 represents 02:15:28 on January 13, 2023, the timestamp 1692075338 represents 02:15:38 on January 13, 2023, and the timestamp 1692075338 represents 02:15:48 on January 13, 2023. During the period from 02:15:18 to 02:15:38, the analysis dimension remains unchanged, and the terminal is always provided with communication services by the edge cluster p1. Therefore, the first three network quality data correspond to an instance identifier id001; after 02:15:48, the analysis dimension corresponding to the terminal changes, and the edge cluster r2 is changed to provide communication services for the terminal. Therefore, this network quality data corresponds to different instance identifiers id002, etc.

[0141] It should be noted that the above Table 5 is only a schematic example. When at least one analysis dimension changes, the edge cluster providing communication services may or may not change; even if at least one analysis dimension remains unchanged, the edge cluster providing communication services may also change, such as detecting that the load of the current edge cluster is large, or other edge clusters are idle, etc., which are not limited here.

[0142] Step 320, group multiple network quality data based on the instance identifier to obtain multiple groups of network quality data.

[0143] Schematically, multiple network quality data can be grouped based on the instance identifier, and multiple network quality data corresponding to the same instance identifier are divided into one group, so as to obtain multiple groups of network quality data. That is: at least one network quality data in the network quality data group corresponds to the same instance identifier.

[0144] As shown in Table 5, the first three network quality data are divided into one group, and then a network quality data group includes network quality data 1, network quality data 2, and network quality data 3.

[0145] Analyze multiple network quality data based on the instance identifier, so the obtained network quality data group corresponds to an instance identifier. That is: the network quality data group corresponds to an instance identifier.

[0146] Step 330: Perform data analysis on multiple network quality data groups respectively to obtain communication quality data corresponding to the multiple network quality data groups respectively.

[0147] Illustratively, the data analysis is used to perform statistical analysis by integrating at least one network quality data within a network quality data group to obtain communication quality data corresponding to the network quality data group.

[0148] In an optional embodiment, perform mean processing on at least one network quality data within the same network quality data group to obtain mean quality data corresponding to the multiple network quality data groups respectively; use the mean quality data corresponding to the network quality data groups respectively as communication quality data to obtain multiple communication quality data respectively marked with instance identifiers.

[0149] Illustratively, as shown in Table 5, a network quality data group is composed of network quality data 1, network quality data 2, and network quality data 3. Perform mean processing on network quality data 1, network quality data 2, and network quality data 3 to obtain the mean quality data corresponding to this network quality data group as 25.2 = (23.5 + 24.5 + 27.6), and use the mean quality data 25.2 as the communication quality data corresponding to this network quality data group.

[0150] Illustratively, based on the fact that the communication quality data is obtained from the network quality data divided according to the instance identifier, there is a corresponding relationship between the communication quality data and the instance identifier.

[0151] Among them, the communication quality data is obtained based on at least one network quality data for the same terminal, using the same analysis dimension and provided with communication services by the same edge cluster. Therefore, the communication quality data can more accurately reflect the relative data interaction situation between the terminal and the edge cluster under the same analysis dimension. For example, under the same analysis dimension, the communication quality data between terminal 1 and edge cluster A is 22.1, and the communication quality data between terminal 1 and edge cluster B is 26.7, which means that the communication quality between terminal 1 and edge cluster A is higher and the data interaction efficiency is better, etc.

[0152] It should be noted that the above is only an illustrative example, and the embodiments of the present application are not limited thereto.

[0153] By fully leveraging at least one analysis dimension, multiple communication quality data groups that are provided with communication services by the same edge cluster and have the same analysis dimension are obtained. Furthermore, quality prediction data that facilitates the prediction of subsequent communication quality conditions is obtained based on data processing. By constructing a resource allocation data table, the corresponding quality prediction data can be determined according to at least one analysis dimension, so as to select the corresponding edge cluster to provide communication services in the case of the numerical value of the quality prediction data, and the allocation accuracy of the edge cluster is improved through a multi-level analysis process, facilitating more efficient data interaction by the terminal based on the edge cluster.

[0154] In the embodiment of the present application, the content of processing multiple network quality data according to the instance identifier to obtain communication quality data is introduced. By collecting multiple network quality data, each network quality data corresponds to an instance identifier according to the terminal, the edge cluster that provides communication services for the terminal, and at least one analysis dimension. Thus, the multiple network quality data are grouped according to the instance identifier, and at least one network quality data is processed within the same network quality data group to obtain communication quality data corresponding to the multiple network quality data groups respectively. Considering the situation where multiple terminals are processed by different edge clusters respectively with instance identifiers, the communication quality data that can reflect the corresponding relationship between the terminal and the edge cluster is obtained, facilitating subsequent comprehensive analysis of multiple terminals based on the edge cluster under the analysis dimension and refining the analysis level.

[0155] In an optional embodiment, at least one analysis dimension is used as the data table index, and the quality prediction data corresponding to at least one analysis dimension and the edge cluster with an associated relationship are used as the data table index values, thereby constructing a resource allocation data table. Schematically, as Figure 5 shown, the above Figure 2 shown embodiment can also be implemented as the following steps 510 to step 540; among them, step 240 can also be implemented as the following step 540.

[0156] Step 510, obtain multiple communication quality data.

[0157] Among them, the communication quality data is data statistically obtained from at least one analysis dimension when the terminal communicates with the edge cluster during the historical time period, and the edge cluster is a cluster that provides communication services for the terminal based on at least one analysis dimension.

[0158] The content of step 510 has been introduced in the above steps 210, 310 to 330, and will not be elaborated here.

[0159] Step 520, group the multiple communication quality data by comprehensively considering at least one analysis dimension and the edge cluster, to obtain multiple communication quality data groups that are provided with communication services by the same edge cluster and have the same analysis dimension.

[0160] Schematically, each communication quality data corresponds to at least one analysis dimension and an edge cluster. Multiple communication quality data come from multiple terminals. Group the multiple communication quality data by integrating at least one analysis dimension and the edge cluster, so as to summarize at least one communication quality data that provides communication services by the same edge cluster and has the same analysis dimension into one communication quality data group.

[0161] Among them, each communication quality data group includes at least one communication quality data. At least one communication quality data within the group is provided with communication services by the same edge cluster and has the same analysis dimension, and the analysis situations corresponding to the analysis dimensions are also the same.

[0162] Step 530: Perform data processing on at least one communication quality data within the same communication quality data group to obtain quality prediction data corresponding to multiple communication quality data groups respectively.

[0163] Among them, there is an association relationship between the quality prediction data and the edge cluster.

[0164] Optionally, the data processing is implemented as at least one of multiple processing methods such as mean processing, median processing, seventieth percentile value processing, ninetieth percentile value processing, variance processing, etc.

[0165] In an optional embodiment, perform mean processing on at least one communication quality data within the same communication quality data group to obtain mean data corresponding to multiple communication quality data groups respectively as the quality prediction data.

[0166] Schematically, as shown in Table 2, based on the communication quality data 1 corresponding to terminal A and the communication quality data 2 corresponding to terminal B, which are provided with services by the same edge cluster and have the same analysis dimension, the communication quality data 1 and the communication quality data 2 can be summarized into one communication quality data group. Perform mean processing on the communication quality data 1: 0.23ms and the communication quality data 2: 0.44ms to obtain the quality prediction data of 0.335ms.

[0167] That is: there is a corresponding relationship between the quality prediction data and the edge cluster and at least one analysis dimension used when determining the quality prediction data. For example: the corresponding relationship is expressed by "Region P - Platform Y - 5G - Edge Cluster p1 - 0.335".

[0168] In an optional embodiment, perform numerical sorting on at least one communication quality data within the same communication quality data group to obtain a quality data sequence corresponding to multiple communication quality data groups respectively; based on the data distribution of the communication quality data within the quality data sequence, obtain the quality prediction data corresponding to multiple communication quality data groups respectively.

[0169] Among them, the data distribution is used to characterize the numerical changes among multiple communication quality data.

[0170] Schematically, the data distribution is implemented as at least one of multiple distribution situations such as normal distribution, skewed distribution, uniform distribution, discrete distribution, etc., which helps to select appropriate models and methods in data analysis to comprehensively analyze multiple communication quality data to predict data changes in future time periods.

[0171] Optionally, when the data distribution corresponding to the communication quality data group is a normal distribution, the pre-selected data processing model 1 is used to process multiple communication quality data within the communication quality data group; when the data distribution corresponding to the communication quality data group is a skewed distribution, the pre-selected data processing model 2 is used to process multiple communication quality data within the communication quality data group, etc.

[0172] Or, when the data distribution corresponding to the communication quality data group is a normal distribution, the variance corresponding to the communication quality data group is determined as the quality prediction data, etc.

[0173] It should be noted that the above is only a schematic example, and the embodiments of the present application are not limited thereto.

[0174] Step 540, using at least one analysis dimension as the data table index, and using the quality prediction data corresponding to at least one analysis dimension and the edge clusters having an association relationship with the quality prediction data as the data table index values, to construct a resource allocation data table.

[0175] Schematically, as shown in Table 4, using the three analysis dimensions of terminal location, terminal operator, and terminal network type as the data table index, and using the corresponding quality prediction data and the edge clusters having an association relationship with the quality prediction data as the data table index values, so as to determine the quality prediction data according to at least one analysis dimension and determine at least one edge cluster that can be optimally allocated.

[0176] Among them, the resource allocation data table is used to allocate an edge cluster that provides communication services for the prediction terminal according to at least one analysis dimension corresponding to the prediction terminal and the association relationship.

[0177] Schematically, the prediction terminal is any terminal, and the resource allocation data table is a data table summarized based on communication quality data within a historical time period, and can determine at least one edge cluster that can provide good communication services for the prediction terminal from multiple edge clusters according to at least one analysis dimension corresponding to the prediction terminal.

[0178] Optionally, obtain the terminal information corresponding to the prediction terminal, where the terminal information includes at least one analysis dimension.

[0179] For example: as shown in Table 4, if the terminal information of the predicted terminal represents that the terminal location of the predicted terminal is P region, the terminal operator is Y platform, and the terminal network type is 5G, then the resource allocation data table can be queried based on these three analysis dimensions, and the quality prediction data that meets the three analysis dimensions can be determined to include quality prediction data 1, quality prediction data 2, and quality prediction data 3, thereby determining at least one edge cluster including edge cluster p1, edge cluster p2, and edge cluster w1.

[0180] It is worth noting that the above are merely illustrative examples and are not limited to the embodiments of the present application.

[0181] By making full use of at least one analysis dimension, multiple communication quality data groups with the same analysis dimension and provided by the same edge cluster for communication services are obtained, and then quality prediction data that is convenient for predicting subsequent communication quality conditions is obtained based on data processing; by constructing a resource allocation data table, the corresponding quality prediction data can be determined according to at least one analysis dimension, so that the corresponding edge cluster can be selected to provide communication services under the numerical value of the quality prediction data, and the allocation accuracy of the edge cluster is improved with the help of a multi-level analysis process, so that the terminal can perform more efficient data interaction based on the edge cluster.

[0182] In an embodiment of the present application, the content of a resource allocation data table constructed based on multiple quality prediction data is introduced. At least one analysis dimension is used as a data table index, and the quality prediction data corresponding to the at least one analysis dimension and the edge cluster with an associated relationship are used as the data table index value, so that after knowing at least one analysis dimension of the prediction terminal, the resource allocation data table can be queried to determine multiple quality prediction data and corresponding edge clusters that meet the at least one analysis dimension, and then the multiple edge clusters are sorted and scored according to the numerical values of the multiple quality prediction data, so as to allocate an edge cluster with better performance to the prediction terminal, thereby making full use of the communication quality data in the historical time period to improve the accuracy of edge cluster allocation.

[0183] In an optional embodiment, the edge cluster allocated to the predicted terminal can be determined by means of the constructed resource allocation data table. Figure 6 As shown above Figure 2 The illustrated step 240 may be followed by the following steps 610 to 630 .

[0184] Step 610: Acquire terminal information corresponding to the predicted terminal.

[0185] The terminal information includes at least one analysis dimension corresponding to the predicted terminal.

[0186] Schematically, the terminal information is implemented as various basic information of the terminal. For example, after predicting that the terminal is connected to the server (server side, central cluster, etc.), the prediction terminal sends terminal information to the server, such as: IP address, terminal identifier, etc. Based on the IP address, the terminal location, terminal operator, and terminal network type can be determined. Based on the terminal identifier, the terminal model, etc. can be determined.

[0187] Optionally, the information included in the terminal information is used as at least one analysis dimension for analysis; or, at least one analysis dimension is refined from the terminal information based on a preset analysis dimension list, etc.

[0188] Step 620, query the resource allocation data table with at least one analysis dimension in the terminal information, and determine at least one quality prediction data that matches the at least one analysis dimension.

[0189] Schematically, querying the resource allocation data table with at least one analysis dimension in the terminal information means querying at least one quality prediction data that meets the at least one analysis dimension from the resource allocation data table; based on the association relationship between the quality prediction data and the edge cluster, at least one edge cluster corresponding to the at least one quality prediction data will also be determined, that is, at least one edge cluster is obtained. The at least one edge cluster is the edge cluster that can be allocated to the prediction terminal and matches the prediction terminal.

[0190] As shown in Table 4, if the terminal location of the prediction terminal is region P, the terminal operator is platform Y, and the terminal network type is 5G, then quality prediction data 1, quality prediction data 2, and quality prediction data 3 can be determined from the resource allocation data table based on these three analysis dimensions, so as to determine that at least one edge cluster includes edge cluster p1, edge cluster p2, and edge cluster w1, that is: edge cluster p1, edge cluster p2, and edge cluster w1 match the prediction terminal.

[0191] Step 630, synthesize at least one quality prediction data, and determine the target edge cluster allocated to the prediction terminal.

[0192] Among them, the target edge cluster is used to allocate computing resources for the prediction terminal to execute communication services.

[0193] In an optional embodiment, based on at least one quality prediction data and the association relationship, at least one edge cluster is sorted to obtain a cluster sorting result.

[0194] Among them, the cluster sorting result is the content sorted according to the numerical values of the quality prediction data associated with the edge cluster, and can reflect the performance of at least one edge cluster when communicating with multiple terminals respectively in the historical time period.

[0195] For example, as shown in Table 4, based on Quality Prediction Data 1, Quality Prediction Data 2, and Quality Prediction Data 3, the edge clusters p1, p2, and w1 are sorted to obtain the cluster sorting result: Edge Cluster w1 > Edge Cluster p1 > Edge Cluster p2, where ">" means "superior to", indicating that Edge Cluster w1 performs better in the analysis dimension of Region P - Platform Y - 5G in terms of communication services with the terminal and is more suitable for being recommended to the prediction terminal to provide communication services, etc.

[0196] In an alternative embodiment, in the cluster sorting result, the edge cluster with the smallest quality prediction data is used as the target edge cluster.

[0197] Among them, the value of the quality prediction data has a negative correlation with the quality of the communication service. That is: the larger the value of the quality prediction data, the worse the quality of the communication service provided by the terminal through this edge cluster in the current at least one analysis dimension; the smaller the value of the quality prediction data, the better the quality of the communication service provided by the terminal through this edge cluster in the current at least one analysis dimension.

[0198] Illustratively, based on the cluster sorting result of "Edge Cluster w1 > Edge Cluster p1 > Edge Cluster p2", the edge cluster w1 with the smallest quality prediction data is used as the target edge cluster.

[0199] Among them, the target edge cluster is used to provide computing resources for the prediction terminal. For example: Edge Node 1 is allocated from the target edge cluster to the prediction terminal to process the data corresponding to the prediction terminal, etc.

[0200] Optionally, the value of the quality prediction data can also have a positive correlation with the quality of the communication service. That is: the smaller the value of the quality prediction data, the worse the quality of the communication service provided by the terminal through this edge cluster in the current at least one analysis dimension; the larger the value of the quality prediction data, the better the quality of the communication service provided by the terminal through this edge cluster in the current at least one analysis dimension.

[0201] That is: the relationship between the quality prediction data and the communication service quality is related to the type of the selected quality prediction data, which is not limited here.

[0202] In some embodiments, the above resource allocation data table is referred to as a cluster selection model, as Figure 7 shown, is a schematic diagram for obtaining the cluster selection model 710 and using the cluster selection model 710 to determine the target edge cluster for the prediction terminal 720.

[0203] Schematically, first, each instance reports the transmission delay (such as the above-mentioned network quality data) every 10 seconds; then, the average transmission delay (such as the above-mentioned communication quality data) is aggregated according to the instance dimension; and then, aggregation is performed according to dimensions such as the terminal location, terminal operator, terminal network type, and edge cluster, to obtain the cluster selection model 710.

[0204] When applied, the terminal information corresponding to the prediction terminal 720 is used as the input of the cluster selection model 710, and the terminal information includes the terminal location, terminal operator, and terminal network type corresponding to the prediction terminal 720; the output of the cluster selection model 710 is the cluster ranking (the above-mentioned cluster sorting result).

[0205] Schematically, if the cluster ranking is obtained by sorting in descending order based on the quality prediction data, then as Figure 7 shown, the edge cluster 2 is the target edge cluster; or, if the cluster ranking is obtained by sorting in ascending order based on the quality prediction data, then as Figure 7 shown, the edge cluster 4 is the target edge cluster, etc.

[0206] In some embodiments, the above-mentioned resource allocation data table is referred to as a cluster selection model, and the target edge cluster can be determined according to the cluster score. As Figure 8 shown, it is a schematic diagram for determining the target edge cluster based on the cluster score.

[0207] Among them, the prediction terminal 810 scores each of the multiple edge clusters respectively based on the quality prediction data corresponding to each of the multiple edge clusters, to obtain the cluster scores corresponding to each of the multiple edge clusters.

[0208] Optionally, the cluster score is determined based on the numerical range of the quality prediction data; or, the cluster score is determined based on the sorting among multiple quality prediction data; or, the cluster score is assigned based on the numerical magnitude among multiple quality prediction data, etc. Among them, the quality prediction data of the edge cluster 821 is the smallest, and the cluster score corresponding to the edge cluster 821 is the highest, which is 10 points; the quality prediction data of the edge cluster 822 is the second smallest, and the cluster score corresponding to the edge cluster 822 is 8 points; the quality prediction data of the edge cluster 823 is the third smallest, and the cluster score corresponding to the edge cluster 823 is 7 points; the quality prediction data of the edge cluster 824 is the largest, and the cluster score corresponding to the edge cluster 824 is the lowest, which is 5 points, etc.

[0209] It should be noted that the above is only a schematic example, and the embodiments of the present application are not limited thereto.

[0210] In summary, by fully leveraging at least one analysis dimension, multiple communication quality data groups provided by the same edge cluster and having the same analysis dimension are obtained. Furthermore, quality prediction data facilitating the prediction of subsequent communication quality conditions is obtained based on data processing. By constructing a resource allocation data table, the corresponding quality prediction data can be determined according to at least one analysis dimension, so as to select the corresponding edge cluster to provide communication services under the numerical condition of the quality prediction data, improving the allocation accuracy of the edge cluster through a multi-level analysis process and facilitating more efficient data interaction of the terminal based on the edge cluster.

[0211] In an embodiment of the present application, the content of allocating a target edge cluster to a prediction terminal through a resource allocation data table is introduced. Query the resource allocation data table with at least one analysis dimension in the terminal information, so as to obtain at least one quality prediction data corresponding to at least one analysis dimension. Furthermore, sort the edge clusters according to the quality prediction data, and use the edge cluster with the best performance as the target edge cluster to be allocated to the prediction terminal, and process the data of the prediction terminal through the computing resources of the edge nodes in the target edge cluster, improving the efficiency of the terminal to obtain the data processing result.

[0212] In an optional embodiment, the process of obtaining the resource allocation data table according to the communication quality data can be implemented as a real-time update process, that is, after obtaining the resource allocation data table, the resource allocation data table can be updated in real time according to the subsequently obtained communication quality data, so as to increase the amount of sample data for matching between the terminal and the edge cluster, thereby improving the accuracy of matching the edge cluster for the terminal. The number of edge clusters that can be allocated to the terminal may change in this process. For example, new edge clusters are added to better disperse data processing. The newly added edge clusters cannot provide communication services for the terminal based on the currently constructed resource allocation data table, because the method of querying the resource allocation table cannot allocate the newly added edge clusters to the terminal. Therefore, when there are newly added edge clusters, the above cluster allocation method can be appropriately optimized.

[0213] Schematically, as Figure 9 shown, Figure 2 After step 240 shown, it can also be implemented as steps 910 to 920 as follows.

[0214] Step 910, based on the terminals having a communication service relationship with multiple edge clusters, determine the number of terminals corresponding to each of the multiple edge clusters.

[0215] Among them, the communication service relationship is used to represent the relationship that the terminal is provided with communication services by the edge cluster.

[0216] Optionally, the communication service relationship can be limited to a preset time period. That is: the edge cluster that is determined to provide communication services to the terminal within the preset time period, there is a communication service relationship between the terminal and this edge cluster. Even if the edge cluster has provided communication services to the terminal outside the preset time period, there is no communication service relationship.

[0217] For example: within the past 20 minutes, the edge cluster that provides communication services to the terminal is edge cluster A, then there is a communication service relationship between the terminal and edge cluster A; even if the edge cluster that provided communication services to the terminal 25 minutes ago was edge cluster B, it is not considered that there is a communication service relationship between the terminal and edge cluster B.

[0218] Optionally, the communication service relationship can be set to all past time periods. That is: for the edge clusters that are determined to have provided communication services to the terminal within all past time periods, it can be considered that there is a communication service relationship between the terminal and the edge clusters.

[0219] For example: within all past time periods, terminal 1 has been provided with communication services by edge cluster A and also by edge cluster B, then it can be expressed that there is a communication service relationship between terminal 1 and edge cluster A, and there is also a communication service relationship between terminal 1 and edge cluster B, etc.

[0220] In some embodiments, taking the communication service relationship being set to all past time periods as an example, it is determined that there is a communication service relationship with multiple edge clusters, so that the terminal data corresponding to multiple edge clusters can be determined respectively.

[0221] Among them, the number of terminals is used to represent the number of terminals that have a communication service relationship with the edge cluster.

[0222] Schematically, if edge cluster A has provided communication services to 105 terminals, then the number of terminals corresponding to edge cluster A is 105; if edge cluster B has provided communication services to 39 terminals, then the number of terminals corresponding to edge cluster A is 39, etc.

[0223] Step 920, in response to the first number of terminals corresponding to the first edge cluster being less than the preset number threshold, allocate an edge cluster that provides communication services to the predicted terminal by integrating the preset allocation probability and the resource allocation data table.

[0224] Among them, the first edge cluster is any one of the multiple edge clusters, and the multiple edge clusters are edge clusters that can provide communication services to the terminal.

[0225] Optionally, multiple edge clusters are distributed in multiple regions. Each region may or may not deploy at least one edge cluster. Based on regional development, new edge clusters can be established in regions without edge clusters, or additional edge clusters can be added in regions where edge clusters have already been deployed, etc.

[0226] Schematically, determine the number of first terminals corresponding to the first edge cluster, that is, determine the number of terminals having a communication service relationship with the first edge cluster.

[0227] Among them, the preset quantity threshold is a preset quantity threshold used to measure the situation of the number of terminals having a communication service relationship with the edge cluster. For example: the preset quantity thresholds are values such as 10, 20, 25, etc.

[0228] Optionally, compare the number of first terminals with the preset quantity threshold. When the number of first terminals is less than the preset quantity threshold, when allocating an edge cluster to a predicted terminal through the resource allocation data table, additionally add the preset allocation probability to the allocation consideration factors, and determine the edge cluster allocated to the predicted terminal by combining the preset allocation probability and the resource allocation table.

[0229] Among them, the preset allocation probability is a preset allocation probability used to represent the probability of allocating the first edge cluster to the predicted terminal. For example: the preset allocation probabilities are 1%, 0.5%, etc.

[0230] Schematically, if the number of first terminals corresponding to the first edge cluster is less than the preset quantity threshold, it is considered that in the historical time period, the situation of allocating terminals to the first edge cluster to provide communication services is less, resulting in less communication quality data characterizing the communication effect between the first edge cluster and the terminals. Therefore, the problem of ignoring the first edge cluster is likely to occur. Therefore, by setting the preset allocation probability, the first edge cluster can be allocated to the predicted terminal with a small probability, so as to increase the probability of allocating the first edge cluster to the terminal and appropriately increase the presence of the first edge cluster among multiple edge clusters.

[0231] In some embodiments, in the process of allocating an edge cluster to a predicted terminal through the resource allocation data table, allocate the first edge cluster to the predicted terminal with the preset allocation probability, and determine the second edge cluster allocated to the predicted terminal.

[0232] Among them, the second edge cluster may be the first edge cluster or other edge clusters among multiple edge clusters.

[0233] Schematically, at least one edge cluster matching the prediction terminal can be determined with the help of a resource allocation data table, and at least one edge cluster is sorted according to the quality prediction data respectively corresponding to the at least one edge cluster. The reliability of this process comes from the fact that there is more terminal data corresponding to multiple edge clusters (i.e., a larger sample size), and the reliability of edge clusters with a relatively small sample size is relatively low. For example, in the case where a new edge cluster is put on the shelf (deployed), this edge cluster lacks samples very much (the number of terminals is small or even none).

[0234] In some embodiments, in the process of allocating an edge cluster for a prediction terminal through a resource allocation data table, an associated edge cluster having an associated relationship with the minimum quality prediction data is determined.

[0235] Among them, the value of the quality prediction data has a negative correlation with the quality of the communication service.

[0236] Optionally, the cluster location and the cluster operator corresponding to the associated edge cluster are determined.

[0237] Among them, the cluster location is used to characterize the region where the associated edge cluster is located, and the cluster operator is used to characterize the operator deploying the associated edge cluster.

[0238] Schematically, each edge cluster has a regional attribute and an operator attribute. Generally speaking, the transmission delay from edge clusters in the same region and of the same operator to the same terminal is close.

[0239] Optionally, at least one candidate edge cluster belonging to the same cluster operator is determined within the cluster location corresponding to the associated edge cluster. The at least one candidate edge cluster includes a first edge cluster; in the process of allocating the at least one candidate edge cluster to the prediction terminal, the first edge cluster is allocated to the prediction terminal with a preset allocation probability, and a second edge cluster allocated to the prediction terminal is determined.

[0240] Schematically, first, a sample quantity threshold (i.e., the above-mentioned preset quantity threshold) is set. If it is lower than this sample quantity threshold, it is considered a cluster lacking samples (i.e., the first edge cluster), and this edge cluster is marked as a cluster lacking samples.

[0241] When allocating an edge cluster for a prediction terminal, at least one edge cluster with better communication effect (i.e., smaller prediction quality data) is determined through the resource allocation data table, and it is retrieved whether there is a cluster lacking samples in the same region and of the same operator. If so, this cluster lacking samples is selected with a relatively small probability (preset allocation probability); this process can be called a cluster detection mechanism, which can supplement samples for clusters with a small sample size.

[0242] For example: Detect clusters lacking samples in the same region and of the same operator with a preset allocation probability of 1%. The detection process is as followsFigure 10 As shown in the figure, for the full amount of data during the three-day time window statistics period, due to the existence of the detection mechanism, 1% of the prediction terminals among all the prediction terminals will be assigned to the sample-deficient cluster 1010, which enables the sample-deficient cluster 1010 to have a certain amount of sample data so that the sample data of the sample-deficient cluster 1010 can be accumulated.

[0243] It should be noted that the above is only a schematic example, and the embodiments of the present application do not limit this.

[0244] In summary, by fully leveraging at least one analysis dimension, multiple communication quality data groups provided by the same edge cluster and having the same analysis dimension are obtained. Then, based on data processing, quality prediction data that is convenient for predicting the subsequent communication quality situation is obtained. By constructing a resource allocation data table, the corresponding quality prediction data can be determined according to at least one analysis dimension, so as to select the corresponding edge cluster to provide communication services in the case of the numerical value of the quality prediction data, improving the allocation accuracy of the edge cluster through a multi-level analysis process, and facilitating more efficient data interaction of the terminal based on the edge cluster.

[0245] In the embodiments of the present application, considering that there may be newly established edge clusters or edge clusters with communication service relationships with a small number of terminals, a method for jointly determining the edge cluster assigned to the terminal by combining a preset allocation probability and a resource allocation data table is introduced. According to the comparison between the number of terminals corresponding to each of the multiple edge clusters and a preset number threshold, it is determined whether to execute the cluster allocation process using the preset allocation probability. Thus, when the number of the first terminals is less than the preset number threshold, the terminals can be allocated to the first edge cluster at a preset allocation ratio, so as to avoid the situation that the first edge cluster always corresponds to a small number of terminals, which affects the process of data processing by the first edge cluster using computing resources, avoiding the generalization problem of cluster allocation, and overcoming the limitation situation.

[0246] In an alternative embodiment, the content of allocating edge clusters to terminals by network speed measurement in the related art is introduced. As Figure 11 shown, it is a schematic diagram of allocating edge clusters.

[0247] (1) Speed measurement service sites are built in multiple regions respectively. For example: In region S1, speed measurement sites of operator Y, operator L, and operator D are built; in region S2, speed measurement sites of operator Y, operator L, and operator D are also built. Among them, each speed measurement site provides a speed measurement address so that other devices can know the speed measurement results.

[0248] (2) During the operation of the application, the terminal 1110 can obtain a list of speed measurement addresses from the program background 1120 (such as the cloud game background), and perform speed measurement on multiple speed measurement addresses in the list of speed measurement addresses respectively.

[0249] Among them, at least one edge cluster is deployed in some areas of multiple regions. The edge cluster is related to the region and the operator. For example, operator L has built edge cluster a in region S1, operator L has built edge cluster b in region S2, and operator Y has built edge cluster c in region S1, etc. Therefore, the network status of the edge clusters located in the same region and deployed by the same operator can be determined with reference to the speed measurement results corresponding to the operator in this region. For example, operator L has built edge cluster a in region S1, and the speed measurement result of operator L in region S1 is 23.5 ms, which can be indirectly used as the speed measurement result corresponding to edge cluster a.

[0250] (3) The terminal returns the speed measurement results to the program background.

[0251] Schematically, the speed measurement results include at least one of multiple types of information such as IP address, RTT, and packet loss rate.

[0252] (4) The program background obtains the regions corresponding to the sites with better results based on the speed measurement results.

[0253] Schematically, the program background selects the test sites with better test results based on the speed measurement results, and determines the regions where the test sites are located. For example, the test site with the best test result is the test site deployed by operator L in region S2.

[0254] (5) The program background allocates a device from the edge clusters in the better regions to the terminal.

[0255] Schematically, if the better region is region S2 and the edge cluster is edge cluster b deployed by operator L in region S2, then a device can be allocated from edge cluster b to the terminal to process data requests of the terminal, etc.

[0256] However, it should be noted that the above related technologies have at least the following technical problems.

[0257] (1) The insufficient coverage of speed measurement sites limits the range of optional edge clusters

[0258] Schematically, as Figure 12 shown, the identifier 1210 represents speed measurement sites, and these speed measurement sites need to be independently deployed; when the speed measurement sites do not cover all regions containing edge clusters, it will affect the range of optional clusters.

[0259] In addition, for terminals in different regions and of different operators, the speed test list is a subset of all speed test sites, but this subset needs to be manually maintained, resulting in relatively low efficiency.

[0260] (2) Real-time speed tests may obtain inaccurate values due to short-term network fluctuations.

[0261] (3) The speed test site is not the same server as the server to which the terminal is connected.

[0262] Illustratively, taking the program background as the cloud game background as an example, there is audio and video transmission between the terminal and the cloud game background. The audio and video stream transmission of the cloud game is based on the peer-to-peer long connection between the webrtc client in the terminal and the webrtc service endpoint in the cloud instance. This server is not the same as the server used during speed tests, so the speed test results do not directly reflect the situation during the actual use of the terminal.

[0263] (4) Real-time speed tests need to be performed every time a terminal is allocated, greatly increasing the waiting time.

[0264] Illustratively, when performing speed tests on each speed test address in the speed test list, whether it is parallel speed testing or serial speed testing, it takes a large amount of time, such as in the range of several hundred milliseconds to one or two seconds. This process is likely to greatly increase the user's waiting time, resulting in a large loss of users.

[0265] Based on this, by using the above cluster allocation method, the communication quality data within a historical time period is summarized, and with the help of the resource allocation data table, the allocation process of the edge clusters is more coordinated. Thus, through big data statistics, the actual experience during the use of the terminal is statistically analyzed using the existing allocation records as the basis for sorting each edge cluster.

[0266] In an optional embodiment, the above cluster allocation method is referred to as the "data-driven computing resource allocation method", where the computing resources are the computing resources provided by the edge nodes within the edge clusters.

[0267] Optionally, taking the application of the cluster allocation method to the cloud game scenario as an example, an application program for cloud gaming is installed on the terminal. Based on the operations within the application program, the terminal sends a data processing request to the cloud game platform (or program background, central cluster, etc.) so that the cloud game platform can perform data processing by itself or select an edge cluster to provide communication services. Usually, the process involves the rendering and transmission of the picture frames during the cloud game process.

[0268] Schematically, for cloud games, a frame of picture is transmitted from the cloud server to the terminal and seen by the user. The transmission process is as follows: cloud server image rendering, video encoding, server receiving encoded frames, network transmission, user receiving encoded frames, video decoding, and terminal image rendering process.

[0269] For cloud gaming platforms, in order to achieve the goal of allocating edge clusters nearby, the main focus is on network transmission. By allocating users to edge clusters with low network transmission latency (roughly speaking, edge clusters that are closer to the terminal) to reduce network transmission latency, this process will directly reduce the delay in users seeing the game screen. And because of the selection of high-quality network transmission links, it also has a positive effect on reducing packet loss rate and freeze rate.

[0270] During use, the terminal will periodically report data related to network quality, such as transmission delay, packet loss rate, etc. By statistically analyzing the data reported by all terminals over a period of time, the network quality from the terminal to each edge cluster in each region and operator can be obtained.

[0271] like Figure 13 As shown, the thick line 1310 is used to divide each region, and each thin line 1320 represents an allocation record. Since the distribution between computing resources and terminals in the edge cluster is not completely matched, the allocation result is not fixed to a certain edge cluster, but dynamic. Each allocation may have multiple allocation results. By regularly aggregating and counting the network quality data reported by these terminals during use, an optimal allocation model (i.e., the above-mentioned resource allocation data table) is obtained; when a terminal enters the platform online, a cluster with better network quality is selected for the terminal based on this model.

[0272] However, the above records cannot guarantee that all edge clusters can be covered, and when a new edge cluster is launched, the terminal cannot be assigned to the new edge cluster without the data of this edge cluster. Therefore, an auxiliary detection mechanism is also needed to improve the optimal allocation model. Therefore, the following five aspects are used to illustrate the solution.

[0273] (1) Reporting of transmission delay

[0274] Indicatively, the attribute values of the terminal and the edge cluster are first formulated, and then the network quality data such as transmission delay and packet loss rate are reported and counted, and then the statistical results are used in the allocation stage.

[0275] The attribute value is the above-mentioned analysis dimension, which can be roughly divided into two aspects: region and network. Region can be refined into provinces and cities; network can be divided into operators, network types, etc.

[0276] Among them, for edge clusters, the attribute values mainly involve provinces, cities, operators, etc.; for terminals, the attribute values include provinces, cities, operators, network types, etc.

[0277] The reason for considering the above attribute values is that the transmission delay of the audio and video network during the use of the terminal is related to the above-mentioned factors. For example: In City S, for the terminals in the edge cluster assigned to Operator D, when using the 4G network, the network transmission delay is about 50 milliseconds, but if using the Wi-Fi network, it is only about 30 milliseconds; or, for the terminals using Wi-Fi for cloud gaming in City S, for the edge cluster of Operator Y, the transmission delay is only about 20 milliseconds. This is the reason for identifying the terminal and the edge cluster with the above attribute values.

[0278] In some embodiments, the cloud gaming server will establish a long connection with the terminal to transmit the encoded audio and video data in real time. As Figure 4 shown, what needs to be recorded is the round-trip delay (RTT) to obtain the network quality data. Based on the above process, the cloud gaming server can calculate the round-trip delay periodically (for example: every 10 seconds), and then report it to the central data warehouse, which is used to represent the memory for caching data in the central cluster, etc.

[0279] (2) Statistics of data aggregation

[0280] Schematically, analyze and statistically process the data stored in the central data warehouse; aggregate the data within the historical time period according to the region where the terminal belongs (terminal location), the operator to which the terminal belongs (terminal operator), and the edge cluster, and take the average of the network transmission delay of the audio and video to obtain the communication quality data.

[0281] Optionally, during the process of obtaining the communication quality data from the network quality data, some network quality data can also be screened. For example: For some terminal-to-edge cluster latency data with too few samples, the credibility of this part of the content is not high enough and can be removed. For example, if there is only one data interaction between the terminal and the edge cluster, the credibility of the network quality data collected this time is relatively low and can be filtered; or, if there is a lack of terminal information or cluster information during reporting, it can also be filtered.

[0282] As Figure 7 shown, after collecting the network quality data, aggregate according to the instance dimension (such as the above instance identifier) to obtain the average transmission delay as the communication quality data; and then aggregate at least one analysis dimension such as terminal location, terminal operator, terminal network type, etc. and the edge cluster to obtain the resource allocation data table, which is the cluster selection model.

[0283] (3) Cluster detection mechanism

[0284] Schematically, at least one analysis dimension corresponding to the prediction terminal can be used as input, and an edge cluster can be assigned to the prediction terminal through a cluster selection model, that is: determine the quality prediction data corresponding to multiple edge clusters under at least one analysis dimension according to the cluster selection model, so as to rank the experience quality of multiple edge clusters.

[0285] However, the reliability of the above method depends on the size of the sample quantity, and the reliability of edge clusters with a small sample quantity is relatively low; however, in the case of a new edge cluster being put on the shelves, this edge cluster lacks samples. To address the problem of some edge clusters lacking samples, a detection mechanism is needed to assign terminals to clusters lacking samples with a certain probability.

[0286] As Figure 10 shown, under the condition of applying the cluster selection model, there is a 1% probability of assigning the prediction terminal to the cluster lacking samples 1010.

[0287] (4) Apply the cluster selection model

[0288] Schematically, the cluster selection model is a model that describes the experience quality of terminals from a certain operator in a certain region to each edge cluster. For a prediction terminal, the scores and rankings of each edge cluster can be obtained through this cluster selection model. Thus, there is a basis for allocation.

[0289] As Figure 14 shown, it is an interactive schematic diagram for cluster allocation through the above four links.

[0290] Among them, during the process of the user using cloud games based on the terminal 1410, the terminal 1410 uploads network quality data to the data platform 1420 (the above-mentioned central cluster, cloud game server, etc.). The data platform 1420 calculates the cluster selection model 1430 through offline analysis. When the resource allocation system makes an allocation, it obtains the edge cluster 1440 with the best terminal experience according to the cluster selection model 1430, and allocates the computing resources 1441 in the edge cluster 1440 to the terminal 1410, and the user can perform data processing based on the computing resources 1441.

[0291] Schematically, the process of assigning an edge cluster to a terminal is described as follows.

[0292] Because there are multiple edge clusters distributed in various regions, computing and data processing tasks will be processed on edge nodes closer to the terminal; the goal of edge computing is to provide low latency, high bandwidth, and better user experience. Therefore, if it is necessary to assign the data processing task of the terminal to a certain edge cluster, a cluster selection model that can rank each edge cluster is required; thus, the cluster selection model is applied to the allocation decision of the computing task by the cluster allocation platform.

[0293] Schematic, such as Figure 15 As shown, when multiple terminals perform edge cluster allocation through a cluster allocation platform, the cluster allocation platform needs to allocate a computing resource to the terminal to complete the tasks required by the user. When allocating computing resources, the cluster selection model 1510 ranks each edge cluster based on terminal information (such as: terminal location, terminal operator, etc.), and evaluates which edge cluster can provide a better user experience.

[0294] For example, based on the cluster selection model 1510, edge cluster 2 is allocated to terminal 1521, and based on the cluster selection model 1510, edge cluster 4 is allocated to terminal 1522, etc.

[0295] It should be noted that the above is only a schematic example, and the embodiments of the present application are not limited thereto.

[0296] In summary, by fully leveraging at least one analysis dimension, multiple communication quality data groups provided by the same edge cluster and having the same analysis dimension are obtained, and then quality prediction data convenient for predicting subsequent communication quality conditions is obtained based on data processing; by constructing a resource allocation data table, the corresponding quality prediction data can be determined according to at least one analysis dimension, so as to select the corresponding edge cluster to provide communication services under the numerical conditions of the quality prediction data, and the allocation accuracy of the edge cluster is improved through a multi-level analysis process, facilitating more efficient data interaction of the terminal based on the edge cluster.

[0297] Figure 16 is a structural block diagram of a cluster allocation device provided by an exemplary embodiment of the present application, such as Figure 16 As shown, the device includes the following parts:

[0298] An acquisition module 1610, configured to acquire multiple communication quality data, where the communication quality data is data statistically obtained from at least one analysis dimension when the terminal communicates with an edge cluster during a historical time period, and the edge cluster is a cluster that provides communication services for the terminal based on the at least one analysis dimension;

[0299] A grouping module 1620, configured to group the multiple communication quality data by integrating the at least one analysis dimension and the edge cluster, to obtain multiple communication quality data groups provided by the same edge cluster and having the same analysis dimension;

[0300] A processing module 1630, configured to perform data processing on at least one communication quality data within the same communication quality data group, to obtain quality prediction data corresponding to the multiple communication quality data groups respectively, and there is an association relationship between the quality prediction data and the edge cluster;

[0301] A building block 1640 for constructing a resource allocation data table through the quality prediction data and the at least one analysis dimension; wherein the resource allocation data table is used to allocate an edge cluster that provides the communication service for the prediction terminal according to at least one analysis dimension corresponding to the prediction terminal and the association relationship.

[0302] In an optional embodiment, the obtaining module 1610 is further configured to obtain a plurality of network quality data, where the network quality data is data collected from at least one analysis dimension when the terminal communicates with the edge cluster at a historical moment, and the historical moment is a moment within the historical time period; each of the plurality of network quality data corresponds to an instance identifier, and the instance identifier is used to represent an identifier of the edge cluster providing the communication service for the terminal based on the at least one analysis dimension; grouping the plurality of network quality data based on the instance identifier to obtain a plurality of network quality data groups, where at least one network quality data within the network quality data group corresponds to the same instance identifier; performing data analysis on the plurality of network quality data groups respectively to obtain communication quality data corresponding to the plurality of network quality data groups respectively.

[0303] In an optional embodiment, the obtaining module 1610 is further configured to perform a mean value process on at least one network quality data within the same network quality data group to obtain mean quality data corresponding to the plurality of network quality data groups respectively; using the mean quality data corresponding to the network quality data groups respectively as the communication quality data to obtain the plurality of communication quality data respectively labeled with the instance identifier.

[0304] In an optional embodiment, the processing module 1630 is further configured to perform a mean value process on at least one communication quality data within the same communication quality data group to obtain mean data corresponding to the plurality of communication quality data groups respectively as the quality prediction data.

[0305] In an optional embodiment, the processing module 1630 is further configured to perform a numerical sorting on at least one communication quality data within the same communication quality data group to obtain a quality data sequence corresponding to the plurality of communication quality data groups respectively; based on the data distribution of the communication quality data within the quality data sequence, obtaining the quality prediction data corresponding to the plurality of communication quality data groups respectively, where the data distribution is used to represent the numerical change between the plurality of communication quality data.

[0306] In an optional embodiment, the building block 1640 is further configured to use the at least one analysis dimension as a data table index, and use the quality prediction data corresponding to the at least one analysis dimension and the edge clusters having the association relationship with the quality prediction data as data table index values to construct the resource allocation data table.

[0307] As Figure 17 shown, in an optional embodiment, the apparatus further includes:

[0308] An allocation module 1650, configured to obtain terminal information corresponding to a prediction terminal, where the terminal information includes at least one analysis dimension corresponding to the prediction terminal; query the resource allocation data table with the at least one analysis dimension in the terminal information to determine at least one quality prediction data matching the at least one analysis dimension; and synthesize the at least one quality prediction data to determine a target edge cluster allocated to the prediction terminal, where the target edge cluster is used to allocate computing resources for executing the communication service to the prediction terminal.

[0309] In an optional embodiment, the allocation module 1750 is further configured to sort the at least one edge cluster based on the at least one quality prediction data and the association relationship to obtain a cluster sorting result; and use the edge cluster with the smallest quality prediction data in the cluster sorting result as the target edge cluster, where the value of the quality prediction data has a negative correlation with the quality of the communication service.

[0310] In an optional embodiment, the allocation module 1750 is further configured to determine the number of terminals corresponding to the multiple edge clusters based on the terminals having a communication service relationship with the multiple edge clusters, where the communication service relationship is used to represent the relationship that the edge cluster provides communication services for the terminal, and the number of terminals is used to represent the number of terminals having the communication service relationship with the edge cluster; in response to the number of first terminals corresponding to a first edge cluster being less than a preset number threshold, synthesize a preset allocation probability and the resource allocation data table to allocate an edge cluster providing the communication service for the prediction terminal, where the preset allocation probability is used to characterize the probability of allocating the first edge cluster to the prediction terminal.

[0311] In an optional embodiment, the allocation module 1750 is further configured to, in the process of allocating the edge cluster for the prediction terminal through the resource allocation data table, allocate the first edge cluster to the prediction terminal with the preset allocation probability to determine a second edge cluster allocated to the prediction terminal.

[0312] In an optional embodiment, the allocation module 1750 is further configured to determine, during the process of allocating the edge cluster to the prediction terminal through the resource allocation data table, an associated edge cluster associated with the minimum quality prediction data, where the value of the quality prediction data has a negative correlation with the quality of the communication service; determine the cluster location and the cluster operator corresponding to the associated edge cluster, where the cluster location is used to represent the region where the associated edge cluster is located, and the cluster operator is used to represent the operator deploying the associated edge cluster; determine at least one candidate edge cluster belonging to the same cluster operator within the cluster location corresponding to the associated edge cluster, where the at least one candidate edge cluster includes the first edge cluster; during the process of allocating the at least one candidate edge cluster to the prediction terminal, allocate the first edge cluster to the prediction terminal with the preset allocation probability, and determine the second edge cluster allocated to the prediction terminal.

[0313] In summary, by fully leveraging at least one analysis dimension, multiple communication quality data groups provided by the same edge cluster and having the same analysis dimension are obtained, and then quality prediction data facilitating the prediction of subsequent communication quality conditions is obtained based on data processing; by constructing a resource allocation data table, the corresponding quality prediction data can be determined according to at least one analysis dimension, so as to select the corresponding edge cluster to provide communication services in the case of the value of the quality prediction data, and the allocation accuracy of the edge cluster is improved by means of a multi-level analysis process, facilitating more efficient data interaction of the terminal based on the edge cluster.

[0314] It should be noted that: for the cluster allocation device provided in the above embodiment, only the above division of each functional module is used for illustration. In practical applications, the above functions can be allocated to different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the cluster allocation device provided in the above embodiment and the embodiment of the cluster allocation method belong to the same concept, and the specific implementation process can be seen in the method embodiment, which will not be elaborated here.

[0315] Figure 18The structural schematic diagram of a server provided by an exemplary embodiment of the present application is shown. The server 1800 includes a Central Processing Unit (CPU) 1801, a system memory 1804 including a Random Access Memory (RAM) 1802 and a Read Only Memory (ROM) 1803, and a system bus 1805 connecting the system memory 1804 and the central processing unit 1801. The server 1800 further includes a mass storage device 1806 for storing an operating system 1813, application programs 1814, and other program modules 1815.

[0316] The mass storage device 1806 is connected to the central processing unit 1801 through a mass storage controller (not shown) connected to the system bus 1805. The mass storage device 1806 and its associated computer-readable medium provide non-volatile storage for the server 1800. That is to say, the mass storage device 1806 may include a computer-readable medium (not shown) such as a hard disk or a Compact Disc Read Only Memory (CD-ROM) drive.

[0317] Without loss of generality, computer-readable media may include computer storage media and communication media. Computer storage media includes volatile and non-volatile, removable and non-removable media implemented by any method or technology for storing information such as computer-readable instructions, data structures, program modules, or other data. The above-mentioned system memory 1804 and mass storage device 1806 may be collectively referred to as memory.

[0318] According to various embodiments of the present application, the server 1800 may also be run by connecting to a remote computer on the network through a network such as the Internet. That is, the server 1800 may be connected to the network 1812 through a network interface unit 1811 connected to the system bus 1805, or in other words, the network interface unit 1811 may also be used to connect to other types of networks or remote computer systems (not shown).

[0319] The above-mentioned memory further includes one or more programs, and one or more programs are stored in the memory and configured to be executed by the CPU.

[0320] An embodiment of the present application further provides a computer device, which includes a processor and a memory. At least one instruction, at least one segment of program, code set, or instruction set is stored in the memory, and at least one instruction, at least one segment of program, code set, or instruction set is loaded and executed by the processor to implement the cluster allocation method provided by the above-mentioned method embodiments.

[0321] An embodiment of the present application further provides a computer-readable storage medium, on which at least one instruction, at least one program segment, a code set or an instruction set is stored, and the at least one instruction, at least one program segment, the code set or the instruction set is loaded and executed by a processor to implement the cluster allocation method provided by each of the above method embodiments.

[0322] An embodiment of the present application further provides a computer program product or a computer program, which includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the cluster allocation method described in any one of the above embodiments.

[0323] The foregoing are only optional embodiments of the present application and are not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included within the protection scope of the present application.

Claims

1. A cluster allocation method, characterized in that, The method includes: Obtaining a plurality of communication quality data, where the communication quality data is data statistically obtained from at least one analysis dimension when the terminal communicates with the edge cluster within a historical time period, and the edge cluster is a cluster that provides communication services for the terminal based on the at least one analysis dimension; Grouping the plurality of communication quality data based on the at least one analysis dimension and the edge cluster to obtain a plurality of communication quality data groups with the same communication services provided by the same edge cluster and having the same analysis dimension; Performing data processing on at least one communication quality data within the same communication quality data group to obtain quality prediction data corresponding to the plurality of communication quality data groups respectively, and there is an association relationship between the quality prediction data and the edge cluster; Constructing a resource allocation data table through the quality prediction data and the at least one analysis dimension; wherein, the resource allocation data table is used to allocate an edge cluster that provides communication services for the prediction terminal according to at least one analysis dimension corresponding to the prediction terminal and the association relationship.

2. The method according to claim 1, characterized in that The obtaining of the plurality of communication quality data includes: Obtaining a plurality of network quality data, where the network quality data is data collected from at least one analysis dimension when the terminal communicates with the edge cluster at a historical moment, and the historical moment is a moment within the historical time period; each of the plurality of network quality data corresponds to an instance identifier, and the instance identifier is used to represent an identifier of the edge cluster that provides communication services for the terminal based on the at least one analysis dimension; Grouping the plurality of network quality data based on the instance identifier to obtain a plurality of network quality data groups, and at least one network quality data within the network quality data group corresponds to the same instance identifier; Performing data analysis on the plurality of network quality data groups respectively to obtain communication quality data corresponding to the plurality of network quality data groups respectively.

3. The method according to claim 2, wherein The performing of data analysis on the plurality of network quality data groups respectively to obtain communication quality data corresponding to the plurality of network quality data groups respectively includes: Performing mean processing on at least one network quality data within the same network quality data group to obtain mean quality data corresponding to the plurality of network quality data groups respectively; Using the mean quality data corresponding to the network quality data groups respectively as the communication quality data to obtain the plurality of communication quality data respectively labeled with the instance identifier.

4. The method according to any one of claims 1 to 3, characterized in that, The performing of data processing on at least one communication quality data within the same communication quality data group to obtain quality prediction data corresponding to the plurality of communication quality data groups respectively includes: Performing mean processing on at least one communication quality data within the same communication quality data group to obtain mean data corresponding to the plurality of communication quality data groups respectively as the quality prediction data.

5. The method according to any one of claims 1 to 3, characterized in that, The performing of data processing on at least one communication quality data within the same communication quality data group to obtain quality prediction data corresponding to the plurality of communication quality data groups respectively includes: Numerically sort at least one communication quality data within the same communication quality data group to obtain quality data sequences respectively corresponding to the multiple communication quality data groups; Based on the data distribution of the communication quality data within the quality data sequence, obtain the quality prediction data respectively corresponding to the multiple communication quality data groups, where the data distribution is used to characterize the numerical change among the multiple communication quality data.

6. The method according to any one of claims 1 to 3, characterized in that, The constructing a resource allocation data table through the quality prediction data and the at least one analysis dimension includes: Using the at least one analysis dimension as a data table index, and using the quality prediction data corresponding to the at least one analysis dimension and the edge cluster having the association relationship with the quality prediction data as data table index values to construct the resource allocation data table.

7. The method according to any one of claims 1 to 3, characterized in that, After constructing the resource allocation data table through the quality prediction data and the at least one analysis dimension, it further includes: Obtain the terminal information corresponding to the prediction terminal, where the terminal information includes at least one analysis dimension corresponding to the prediction terminal; Query the resource allocation data table with the at least one analysis dimension in the terminal information to determine at least one quality prediction data matching the at least one analysis dimension; Based on the at least one quality prediction data, determine the target edge cluster allocated to the prediction terminal, where the target edge cluster is used to allocate computing resources for the prediction terminal to execute the communication service.

8. The method according to claim 7, wherein The determining the target edge cluster allocated to the prediction terminal by integrating the at least one quality prediction data includes: Based on the at least one quality prediction data and the association relationship, sort the at least one edge cluster to obtain a cluster sorting result; In the cluster sorting result, use the edge cluster with the smallest quality prediction data as the target edge cluster, and the value of the quality prediction data has a negative correlation with the quality of the communication service.

9. The method according to any one of claims 1 to 3, characterized in that The method further includes: Based on the terminals having a communication service relationship with the multiple edge clusters, determine the number of terminals respectively corresponding to the multiple edge clusters, where the communication service relationship is used to represent the relationship that the edge cluster provides communication services for the terminal, and the number of terminals is used to represent the number of terminals having the communication service relationship with the edge cluster; In response to the number of the first terminals corresponding to the first edge cluster being less than a preset number threshold, integrate the preset allocation probability and the resource allocation data table to allocate an edge cluster providing the communication service for the prediction terminal, where the preset allocation probability is used to characterize the probability of allocating the first edge cluster to the prediction terminal.

10. The method according to claim 9, characterized in that, The integrating the preset allocation probability and the resource allocation data table to allocate an edge set providing the communication service for the prediction terminal includes: During the process of allocating the edge cluster for the prediction terminal through the resource allocation data table, allocate the first edge cluster to the prediction terminal with the preset allocation probability to determine the second edge cluster allocated to the prediction terminal.

11. The method according to claim 10, wherein In the process of allocating the edge cluster to the prediction terminal through the resource allocation data table, the first edge cluster is allocated to the prediction terminal with the preset allocation probability, and determining the second edge cluster allocated to the prediction terminal includes: In the process of allocating the edge cluster to the prediction terminal through the resource allocation data table, determining the associated edge cluster associated with the minimum quality prediction data, where the value of the quality prediction data has a negative correlation with the quality of the communication service; Determining the cluster location and the cluster operator corresponding to the associated edge cluster, where the cluster location is used to represent the region where the associated edge cluster is located, and the cluster operator is used to represent the operator deploying the associated edge cluster; Determining at least one candidate edge cluster belonging to the same cluster operator within the cluster location corresponding to the associated edge cluster, where the at least one candidate edge cluster includes the first edge cluster; In the process of allocating the at least one candidate edge cluster to the prediction terminal, the first edge cluster is allocated to the prediction terminal with the preset allocation probability, and determining the second edge cluster allocated to the prediction terminal.

12. A cluster allocation device, characterized in that, The device includes: An acquisition module, configured to acquire a plurality of communication quality data, where the communication quality data is data statistically obtained from at least one analysis dimension when the terminal communicates with the edge cluster within a historical time period, and the edge cluster is a cluster providing communication services for the terminal based on the at least one analysis dimension; A grouping module, configured to group the plurality of communication quality data by integrating the at least one analysis dimension and the edge cluster, to obtain a plurality of communication quality data groups provided with the communication services by the same edge cluster and having the same analysis dimension; A processing module, configured to perform data processing on at least one communication quality data within the same communication quality data group, to obtain quality prediction data corresponding to the plurality of communication quality data groups respectively, where the quality prediction data has an association relationship with the edge cluster; A construction module, configured to construct a resource allocation data table through the quality prediction data and the at least one analysis dimension; where the resource allocation data table is used to allocate the edge cluster providing the communication service to the prediction terminal according to at least one analysis dimension corresponding to the prediction terminal and the association relationship.

13. A computer device, characterized in that, The computer device includes a processor and a memory, where at least one program is stored in the memory, and the at least one program is loaded and executed by the processor to implement the cluster allocation method according to any one of claims 1 to 11.

14. A computer-readable storage medium, characterized in that, At least one program is stored in the storage medium, and the at least one program is loaded and executed by the processor to implement the cluster allocation method according to any one of claims 1 to 11.

15. A computer program product, characterized in that, Including a computer program or instruction, where the computer program or instruction, when executed by a processor, implements the cluster allocation method according to any one of claims 1 to 11.

Images