Method and system for deploying service function chains based on mec

Abstract

The application discloses a service function chain deployment method and system based on MEC, and relates to the technical field of communication, and the method comprises the following steps: calculating the access experience probability of user requests at each access edge node; deploying a static SFC on an edge node close to the access edge node with the maximum access experience probability; deploying a dynamic SFC on edge nodes other than the edge node close to the access edge node with the maximum access experience probability; receiving a batch of user requests, performing similarity checking on the batch of user requests; according to the similarity checking result, classifying and aggregating the batch of user requests to obtain K request classes; and respectively assigning the user requests in the K request classes to edge nodes with corresponding VNFs for processing.

Description

Technical Field

[0001] This invention relates to the field of communication technology, and more specifically to a service function chain deployment method and system based on MEC. Background Technology

[0002] Mobile edge computing (MEC) is a novel distributed computing approach based on mobile communication networks. By building a cloud service environment on the radio access network side, it enables certain network services and functions to be decoupled from the core network, thereby achieving goals such as cost savings, reduced latency and round-trip time, optimized traffic, enhanced physical security, and improved caching efficiency.

[0003] Network function virtualization (NFV) facilitates the application of mobile edge computing (MEC). NFV is used to replace dedicated hardware devices, which not only reduces costs but also allows for flexible adjustment of services to respond to rapidly changing user requests, without the need to deploy and maintain physical infrastructure. Each network function is implemented as software in a virtual machine, which is called a virtual network function (VNF).

[0004] As user request types become more diverse, in order to reliably, securely, and scalably transmit user data, this user data is usually transmitted in a predetermined order through a series of virtual network functions (VNFs), known as a Service Function Chain (SFC). Network Function Virtualization (NFV) can create a Service Function Chain (SFC) consisting of a set of ordered virtual network functions (VNFs) to provide services for user requests.

[0005] Mobile edge computing (MEC) deploys Service Function Chains (SFCs) for each mobile user through Network Function Virtualization (NFV). Deploying SFCs in a mobile edge computing (MEC) environment has the following drawbacks:

[0006] (1) Most mobile edge computing (MEC) applications are latency-sensitive in network applications;

[0007] (2) The computing resources and physical link bandwidth resources of the edge server deployed in Mobile Edge Computing (MEC) are limited. Multiple user requests to access the edge server through the Service Function Chain (SFC) at the same time will cause congestion. Summary of the Invention

[0008] This invention provides a service function chain deployment method and system based on MEC, with the aim of solving the problems of network latency and congestion.

[0009] This invention is achieved through the following technical solution:

[0010] The first aspect provides a service function chain deployment method based on MEC, including the following steps:

[0011] Calculate the empirical probability of a user request accessing each access edge node;

[0012] Deploy static SFCs on edge nodes that are close to the access edge nodes with the highest probability of access experience;

[0013] Deploy dynamic SFCs on edge nodes other than the edge nodes that are closest to the access edge node with the highest probability of access experience.

[0014] In the MEC network, identify the edge node with the highest probability of user access experience among the edge nodes. Deploy a static SFC on the edge node near the edge node with the highest probability of access experience, and deploy multiple dynamic SFCs on the edge nodes near the edge node with a relatively high probability of access experience. This is to reduce the latency caused by the user requests moving to other edge nodes and alleviate the network congestion caused by multiple user requests simultaneously accessing the edge server through the edge node with the highest probability of access experience.

[0015] Furthermore, calculating the access experience probability of a user request at the access edge node includes the following steps:

[0016] Obtain the user's set of access edge nodes and the user's requested dwell time at each access edge node;

[0017] The empirical probability of this user request being accessed at the access edge node is calculated using the following formula:

[0018]

[0019] in, N represents the access experience probability of user u at access edge node i; u This represents the set of access edge nodes for user u; This represents the time user u spends at the access edge node i.

[0020] Furthermore, after calculating the access experience probability of a user request at each access edge node, it is necessary to compare the access experience probabilities of each access edge node to obtain the access edge node with the highest access experience probability.

[0021] Deploy a static SFC on an edge node near the access edge node with the highest probability of access experience. This will allow user requests to access the static SFC first, reducing the deployment of dynamic SFCs. Furthermore, the resource capacity of the access edge node can meet the resource requirements of the VNF on the SFC.

[0022] Furthermore, before deploying dynamic SFC on edge nodes other than those closest to the edge node with the highest probability of access experience, it is necessary to determine whether the access edge node requested by the user has moved, or whether the resource capacity of the access edge node with the highest probability of access experience is sufficient.

[0023] If the user's access edge node moves, or the resource capacity of the access edge node with the highest access experience probability is insufficient, then deploy dynamic SFC on edge nodes other than the edge node closest to the access edge node with the highest access experience probability.

[0024] If the user's access edge node has not been moved, or if the access edge node with the highest access experience probability has sufficient resource capacity, then there is no need to deploy dynamic SFC on edge nodes other than the edge node closest to the access edge node with the highest access experience probability.

[0025] Deploy multiple dynamic SFC instances on edge nodes near the access edge node with the highest probability of access experience. This reduces the latency caused by user requests moving to other access edge nodes and alleviates congestion caused by multiple user requests simultaneously accessing the access edge node with the highest probability of access experience. If the resource capacity of the access edge node can meet the resource requirements of user requests, there is no need to deploy dynamic SFC, thus reducing the deployment of dynamic SFC.

[0026] Furthermore, after determining whether the user's access edge node has moved, or whether the resource capacity of the access edge node with the highest probability of access is sufficient, and completing the dynamic SFC deployment,

[0027] Upon receiving a batch of user requests, perform a similarity check on the batch of user requests;

[0028] Based on the similarity check results, the user requests in this batch are classified and aggregated to obtain K request classes;

[0029] The user requests from the K request classes are assigned to the edge nodes with the corresponding VNFs for processing.

[0030] Furthermore, the cosine similarity value is used to perform a similarity check on this batch of user requests. The specific steps are as follows:

[0031] Based on the probability of the u-th user making a user request and the conditional probability of the u-th user request accessing the v-th VNF, calculate the joint probability of all users making user requests accessing the v-th VNF.

[0032] Based on the joint probability and the conditional probability of the u-th user requesting access to the v-th VNF, calculate the similarity cosine value among the requests of this batch of users.

[0033] The larger the cosine value of the similarity, the greater the similarity of the user requests.

[0034] Furthermore, based on the similarity check results, the user requests in this batch are classified and aggregated. The specific steps are as follows:

[0035] Check the resource capacity of the edge node that is closest to the edge node with the highest probability of access experience, and determine whether the resource capacity of the edge node is less than the resource requirement requested by the user.

[0036] If the resource capacity of the edge node is not less than the resource requirement of the user request, then the user requests whose similarity cosine value reaches the threshold are generated into a request class, all user requests in the request class are aggregated, and the corresponding VNF is deployed on the edge node.

[0037] If the resource capacity of the edge node is less than the resource requirement requested by the user, then the resource capacity of the edge nodes other than the edge node that is closest to the access edge node with the highest probability of access experience is randomly checked, and it is further determined whether the resource capacity of the edge node is less than the resource requirement requested by the user.

[0038] All user requests on the aforementioned edge nodes are divided into multiple request classes. User requests with high similarity are grouped into the same request class, which facilitates subsequent transmission and reduces transmission latency. The resource capacity of the edge nodes is checked to ensure that the aforementioned edge nodes have sufficient resource capacity to meet the resource requirements of the VNF.

[0039] Furthermore, the user requests from the K request classes are distributed to the edge nodes with the corresponding VNFs for processing, including the following steps:

[0040] The priority of user requests in a request class is determined based on the shortest processing time of user requests in that request class on the edge node where the corresponding VNF is deployed.

[0041] User requests are scheduled according to their priority within the request class, and then distributed to edge nodes with corresponding VNFs deployed.

[0042] A priority-based scheduling strategy is adopted to schedule user requests, reducing processing latency. User requests are assigned to edge nodes with corresponding VNFs for processing, and no user request data packets are dropped during transmission, ensuring the smooth execution of requests on edge nodes.

[0043] The second aspect is to provide a service function chain system based on MEC, which uses the above-mentioned MEC-based service function chain deployment method to deploy the service function chain;

[0044] The aforementioned service function chain system includes:

[0045] A MEC network consisting of multiple access edge nodes and edge nodes;

[0046] Each edge node in the aforementioned MEC network is equipped with either a static SFC or a dynamic SFC.

[0047] The deployed static and dynamic SFCs cover all access edge nodes of the MEC network, which reduces the latency caused by user requests moving to other access edge nodes and alleviates the network congestion caused by multiple user requests simultaneously accessing the edge server through the access edge node with the highest probability of access experience.

[0048] Furthermore, each of the aforementioned edge nodes is equipped with a controller and multiple servers.

[0049] The controller described above is used to receive user requests in batches, perform similarity checks on the batch of user requests, classify and aggregate the batch of user requests based on the similarity check results, obtain K request classes, and then distribute the user requests in the K request classes to the servers with the corresponding VNFs for processing.

[0050] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0051] In the MEC network, identify the edge node with the highest probability of user access experience among the edge nodes. Deploy a static SFC on the edge node near the edge node with the highest probability of access experience, and deploy multiple dynamic SFCs on the edge nodes near the edge node with a relatively high probability of access experience. This is to reduce the latency caused by the user requests moving to other edge nodes and alleviate the network congestion caused by multiple user requests simultaneously accessing the edge server through the edge node with the highest probability of access experience. Attached Figure Description

[0052] To more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be considered as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort. In the drawings:

[0053] Figure 1 The main flowchart provided for Example 1;

[0054] Figure 2 The flowchart provided in Example 2 illustrates the similarity check, classification, aggregation, and allocation scheduling of user requests within a request class.

[0055] Figure 3 The diagram shows the MEC network construction provided in Example 3. Detailed Implementation

[0056] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0057] Example 1

[0058] This embodiment 1 provides a service function chain deployment method based on MEC, such as Figure 1 As shown, it includes the following steps:

[0059] S(1) Calculate the empirical probability of a user request at each access edge node;

[0060] S(2) Deploy static SFCs on edge nodes that are close to the edge nodes with the highest access experience probability;

[0061] S(3) Deploy dynamic SFCs on edge nodes other than the edge nodes that are closest to the edge nodes with the highest access experience probability.

[0062] In the MEC network, identify the edge node with the highest probability of user access experience among the edge nodes. Deploy a static SFC on the edge node near the edge node with the highest probability of access experience, and deploy multiple dynamic SFCs on the edge nodes near the edge node with a relatively high probability of access experience. This is to reduce the latency caused by the user requests moving to other edge nodes and alleviate the network congestion caused by multiple user requests simultaneously accessing the edge server through the edge node with the highest probability of access experience.

[0063] In a specific implementation, calculating the probability of a user's access experience at the access edge node includes the following steps:

[0064] Obtain the user's set of access edge nodes and the user's requested dwell time at each access edge node;

[0065] The empirical probability of this user request being accessed at the access edge node is calculated using the following formula:

[0066]

[0067] in, N represents the access experience probability of user u at access edge node i; u This represents the set of access edge nodes for user u; This represents the time user u spends at the access edge node i.

[0068] In a specific implementation, after calculating the access experience probability of a user request at each access edge node, it is also necessary to...

[0069] S(1.1) compares the access experience probabilities of each access edge node to obtain the access edge node with the highest access experience probability.

[0070] Deploy a static SFC on an edge node near the access edge node with the highest probability of access experience. This will allow user requests to access the static SFC first, reducing the deployment of dynamic SFCs. Furthermore, the resource capacity of the access edge node can meet the resource requirements of the VNF on the SFC.

[0071] In a specific implementation, before deploying dynamic SFC on edge nodes other than those closest to the access edge node with the highest probability of access experience, it is also necessary to...

[0072] S(2.1) determines whether the access edge node requested by the user has moved, or whether the resource capacity of the access edge node with the highest access experience probability is sufficient;

[0073] If the user's access edge node moves, or the resource capacity of the access edge node with the highest access experience probability is insufficient, then deploy dynamic SFC on edge nodes other than the edge node closest to the access edge node with the highest access experience probability.

[0074] If the user's access edge node has not been moved, or if the access edge node with the highest access experience probability has sufficient resource capacity, then there is no need to deploy dynamic SFC on edge nodes other than the edge node closest to the access edge node with the highest access experience probability.

[0075] Deploy multiple dynamic SFC instances on edge nodes near the access edge node with the highest probability of access experience. This reduces the latency caused by user requests moving to other access edge nodes and alleviates congestion caused by multiple user requests simultaneously accessing the access edge node with the highest probability of access experience. If the resource capacity of the access edge node can meet the resource requirements of user requests, there is no need to deploy dynamic SFC, thus reducing the deployment of dynamic SFC.

[0076] In a specific implementation, after determining whether the user's access edge node has moved, or whether the resource capacity of the access edge node with the highest access probability is sufficient, and completing the dynamic SFC deployment,

[0077] S(4) receives a batch of user requests and performs a similarity check on the batch of user requests;

[0078] S(5) Based on the similarity check results, classify and aggregate the user requests in this batch to obtain K request classes;

[0079] S(6) assigns the user requests from the K request classes to the edge nodes with the corresponding VNFs for processing.

[0080] Example 2

[0081] like Figure 2 As shown, the cosine similarity value is used to perform a similarity check on this batch of user requests. The specific steps are as follows:

[0082] S(4.1) Calculate the joint probability of all users requesting to access the v-th VNF based on the probability of the u-th user issuing a user request and the conditional probability of the u-th user requesting to access the v-th VNF, as shown in the following formula:

[0083]

[0084] in, q represents the joint probability that all user requests will access the v-th VNF; u,v ρ(u) represents the conditional probability that a user request from the u-th user will access the v-th VNF; n ρ(u) represents the probability of a user request sent by the u-th user. n )∈[0,1].

[0085] S(4.2) Based on the joint probability mentioned above and the conditional probability of the u-th user requesting to access the v-th VNF, calculate the similarity cosine value among the requests of this batch of users, as follows:

[0086]

[0087] Here, the feature vector represents the joint request probability of a user to a VNF. The larger the similarity cosine value mentioned above, the greater the similarity of the user requests.

[0088] In a specific implementation, based on the similarity check results, the batch of user requests is classified and aggregated. The specific steps are as follows:

[0089] S(5.1) Check the resource capacity of the edge node that is closest to the access edge node with the highest access experience probability, and determine whether the resource capacity of the edge node is less than the resource requirement requested by the user.

[0090] S(5.2) If the resource capacity of the edge node is not less than the resource requirement of the user's request, and the resource capacity of the physical link is less than the resource requirement of the virtual link, then generate a request class for the batch of user requests whose similarity cosine value reaches the threshold (i.e., the magnitude of the cosine value differs by 0 to 0.1). Aggregate all user requests of this request class; and deploy the corresponding VNF on this edge node;

[0091] S(5.3) If the resource capacity of the edge node is less than the resource requirement requested by the user, then randomly check the resource capacity of the edge nodes other than the edge node that is closest to the access edge node with the highest access experience probability, and continue to determine whether the resource capacity of the edge node is less than the resource requirement requested by the user.

[0092] All user requests on the aforementioned edge nodes are divided into multiple request classes. User requests with high similarity are grouped into the same request class, which facilitates subsequent transmission and reduces transmission latency. The resource capacity of the edge nodes is checked to ensure that the aforementioned edge nodes have sufficient resource capacity to meet the resource requirements of the VNF.

[0093] In a specific implementation, user requests from the K request classes are distributed to edge nodes with corresponding VNFs for processing, including the following steps:

[0094] S(6.1) determines the priority of user requests in a request class based on the shortest processing time of user requests in a request class on the edge node where the corresponding VNF is deployed;

[0095] S(6.2) schedules user requests in order of priority according to the user requests in the request class, and assigns the user requests in the request class to the edge nodes with the corresponding VNFs deployed.

[0096] A priority-based scheduling strategy is adopted to schedule user requests, reducing processing latency. User requests are assigned to edge nodes with corresponding VNFs for processing, and no user request data packets are dropped during transmission, ensuring the smooth execution of requests on edge nodes.

[0097] Example 3

[0098] This embodiment 3 provides a service function chain system based on MEC, such as Figure 3 As shown, the system uses the above-mentioned MEC-based service function chain deployment method to deploy the service function chain;

[0099] The aforementioned service function chain system includes:

[0100] A MEC network consisting of multiple access edge nodes and edge nodes;

[0101] Each edge node in the aforementioned MEC network is equipped with either a static SFC or a dynamic SFC.

[0102] The deployed static and dynamic SFCs cover all access edge nodes of the MEC network, which reduces the latency caused by user requests moving to other access edge nodes and alleviates the network congestion caused by multiple user requests simultaneously accessing the edge server through the access edge node with the highest probability of access experience.

[0103] In a specific embodiment, each of the aforementioned edge nodes is equipped with a controller and multiple servers.

[0104] The controller described above is used to receive user requests in batches, perform similarity checks on the batch of user requests, classify and aggregate the batch of user requests based on the similarity check results, obtain K request classes, and then distribute the user requests in the K request classes to the servers with the corresponding VNFs for processing.

[0105] Example 4

[0106] After deploying a static SFC and multiple dynamic SFCs in the MEC network, when the resource demand requested by a user is less than the resource capacity of the access edge node, a group of user requests will access the static SFC through the access edge node; when the resource demand requested by a user is greater than the resource capacity of the access edge node, or when the user requests to move to another access edge node, the dynamic SFC instance will be accessed through the other access edge node.

[0107] In a specific implementation, multiple dynamic SFCs are deployed on edge nodes within the range of access edge nodes with a high access experience probability. (The access experience probability of all access edge nodes is calculated according to the access experience probability formula. Then, the access edge node with the highest access experience probability is excluded. The access experience probabilities of the remaining access edge nodes are compared and sorted in descending order of access experience probability. Dynamic SFCs are then deployed on nearby edge nodes in sequence.) This is mainly because the location requested by the user may change or the resource capacity of the access edge node accessing the static SFC may be insufficient. Under the constraints of edge node resource capacity and link resource capacity, the QoS of user-sensitive services is improved.

[0108] A specific implementation of a method for deploying static and dynamic SFC instances in an MEC network to reduce network downtime and latency and improve user QoS includes the following steps:

[0109] Step 101: Construct multiple access edge nodes and edge nodes in the MEC network, with each node having a certain resource capacity. User requests are routed through the access edge nodes to access either static or dynamic SFC instances, with each SFC instance located on one edge node.

[0110] Step 102: Find the node with the highest access experience probability in the MEC network, which is calculated by the previous dwell time of each user request on each access edge node;

[0111] Step 103: Deploy a static SFC instance on an edge node near the access edge node with the highest empirical probability, and this static SFC instance can only be accessed by the access edge node with the highest access empirical probability.

[0112] Step 104: When the location requested by the user moves or the resource capacity of the access edge node with the highest access experience probability is insufficient, deploy a dynamic SFC instance on the edge nodes within the range of other access edge nodes with a higher access experience probability. A dynamic SFC instance can be accessed by multiple access edge nodes.

[0113] Step 105: Static SFC and dynamic SFC instances must cover all access edge nodes in the MEC, and each SFC instance is composed according to a certain VNF arrangement order.

[0114] In a specific implementation, when a user requests access to a static or dynamic SFC instance from an access edge node, the controller on the edge node performs similarity checks, classifications, and aggregations on the incoming user requests, and schedules them to the appropriate edge node to reduce processing latency and further improve user QoS. The specific steps are as follows:

[0115] Step 201: On the controller of the edge node, the cosine similarity representation is used to check the similarity of incoming user requests. If the cosine similarity value between two user requests is high, they can be classified into the same request class.

[0116] Step 202: Divide all user requests into K request classes, and then use a priority sorting strategy to aggregate and schedule the request classes. Here, the priority sorting is based on the shortest processing time on the controller.

[0117] Step 203: Use the M / M / 1 queuing method to distribute request classes to the edge nodes where the corresponding VNFs are placed for processing; where the average request arrival rate of the request class is greater than the average service rate of the edge nodes (i.e., in, μ represents the average request arrival rate of request class K at edge node n. n (This represents the average service rate of edge node n), ensuring that no user request data packets are dropped during transmission, and that user requests can be successfully executed on the edge node.

[0118] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A service function chain deployment method based on MEC, characterized in that, Includes the following steps: Obtain the user's set of access edge nodes and the time the user's request stays at each access edge node, and calculate the user's access experience probability at each access edge node using the following formula: ,in, This represents the access experience probability of user u at access edge node i; This represents the set of access edge nodes for user u; This represents the time user u spends at access edge node i; Deploy static SFCs on edge nodes that are close to the access edge nodes with the highest probability of access experience; Deploy dynamic SFCs on edge nodes other than those that are closest to the access edge node with the highest probability of access experience. Before deploying dynamic SFC on edge nodes other than those closest to the edge node with the highest probability of access experience, it is necessary to determine whether the access edge node requested by the user has moved, or whether the resource capacity of the access edge node with the highest probability of access experience is sufficient. If the user's access edge node moves, or the resource capacity of the access edge node with the highest access experience probability is insufficient, then deploy dynamic SFC on edge nodes other than the edge node closest to the access edge node with the highest access experience probability. If the user's access edge node has not been moved, or if the access edge node with the highest access experience probability has sufficient resource capacity, then there is no need to deploy dynamic SFC on edge nodes other than the edge node close to the access edge node with the highest access experience probability. After determining whether the user's access edge node has moved, or whether the resource capacity of the access edge node with the highest access probability is sufficient, and completing the dynamic SFC deployment, a batch of user requests are received. A similarity check is performed on the batch of user requests. Based on the similarity check results, the batch of user requests is classified and aggregated to obtain K request classes. The user requests in the K request classes are then assigned to the edge nodes with the corresponding VNFs for processing.

2. The service function chain deployment method based on MEC according to claim 1, characterized in that, After calculating the access experience probability of a user request at each access edge node, it is necessary to compare the access experience probabilities of each access edge node to obtain the access edge node with the highest access experience probability.

3. The service function chain deployment method based on MEC according to claim 1, characterized in that, The cosine similarity score is used to perform a similarity check on this batch of user requests. The specific steps are as follows: Based on the probability of the u-th user making a user request and the conditional probability of the u-th user request accessing the v-th VNF, calculate the joint probability of all users making user requests accessing the v-th VNF. Based on the joint probability and the conditional probability of the u-th user requesting access to the v-th VNF, calculate the similarity cosine value among the requests of this batch of users.

4. The service function chain deployment method based on MEC according to claim 3, characterized in that, Based on the similarity check results, the user requests in this batch are classified and aggregated. The specific steps are as follows: Check the resource capacity of the edge node that is closest to the edge node with the highest probability of access experience, and determine whether the resource capacity of the edge node is less than the resource requirement requested by the user. If the resource capacity of the edge node is not less than the resource requirement of the user request, then the user requests whose similarity cosine value reaches the threshold are generated into a request class, all user requests in the request class are aggregated, and the corresponding VNF is deployed on the edge node. If the resource capacity of the edge node is less than the resource requirement requested by the user, then the resource capacity of the edge nodes other than the edge node that is closest to the access edge node with the highest probability of access experience is randomly checked, and it is further determined whether the resource capacity of the edge node is less than the resource requirement requested by the user.

5. The service function chain deployment method based on MEC according to claim 1, characterized in that, The user requests from the K request classes are distributed to the edge nodes with the corresponding VNFs for processing, including the following steps: The priority of a user request is determined based on the shortest processing time of the user request in a request class on the edge node where the corresponding VNF is deployed. User requests are scheduled according to their priority within the request class, and then distributed to edge nodes with corresponding VNFs deployed.

6. A service function chain system based on MEC, characterized in that, The system deploys the service function chain using the MEC-based service function chain deployment method described in any one of claims 1 to 5; The service function chain system includes: A MEC network consisting of multiple access edge nodes and edge nodes; Each edge node in the MEC network is equipped with either a static SFC or a dynamic SFC.

7. The MEC-based service function chain system according to claim 6, characterized in that, Each of the aforementioned edge nodes is equipped with a controller and multiple servers; The controller is used to receive user requests in batches, perform similarity checks on the batch of user requests, classify and aggregate the batch of user requests based on the similarity check results to obtain K request classes, and then distribute the user requests in the K request classes to the servers with the corresponding VNFs for processing.

Images