Network function on-demand deployment method and device applied to unmanned cluster platform

By building a deep reinforcement learning framework and graph convolution network on the unmanned cluster platform, combining with the GRU encoder to dynamically allocate resources, the dynamic problem of network function orchestration in the unmanned cluster platform is solved, and efficient network status and resources are achieved manage.

CN119095058APending Publication Date: 2024-12-06BEIJING UNIV OF POSTS & TELECOMM
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202411196267.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-29
Publication Date
2024-12-06

AI Technical Summary

Technical Problem

The existing service function chain orchestration mechanism cannot be applied to dynamic nodes in unmanned cluster platforms, and it cannot effectively deploy network functions on demand, especially when network topology and resource states change rapidly.

Method used

By pre-configuring a variety of constraints, including resource quantity, bandwidth and physical node placement constraints, an adaptive service function chain placement framework for deep reinforcement learning is built, and combined with graph convolutional networks and GRU encoders, unmanned cluster network resources are dynamically allocated. Generate service function chain orchestration information to optimize the deployment of virtual network functions.

Benefits of technology

It significantly improves the accuracy and efficiency of service function chain orchestration, can quickly make optimization decisions, adapt to changes in network topology and resource state, and provide more accurate network state information and resource demand prediction.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119095058A_ABST
    Figure CN119095058A_ABST
Patent Text Reader

Abstract

The invention provides a network function on-demand deployment method and device applied to an unmanned cluster platform, and relates to the technical field of unmanned aerial vehicle communication, and the method comprises the steps: pre-configuring a plurality of constraint conditions, and carrying out the placement constraint of a virtual network function through the plurality of constraint conditions; if one or more constraint conditions of the multiple constraint conditions are met, responding to a service function chain placement request initiated by an unmanned platform base station, matching unmanned cluster network resources with the service function chain placement request, dynamically allocating the unmanned cluster network resources based on the service function chain placement request, and generating service function chain arrangement information; and deploying the virtual network function according to the service function chain arrangement information. According to the method and the device, the accuracy and the efficiency of SFC arrangement are remarkably improved, and particularly, an optimization decision can be quickly made when the network topology and the resource state are quickly changed.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of unmanned aerial vehicle communication technology, and in particular to a method and device for on-demand deployment of network functions for an unmanned cluster platform. Background Art

[0002] The current service function chain orchestration mechanism can roughly include mathematical optimization, heuristic research, and reinforcement learning. However, the current method is only applicable to static resources in physical servers and does not consider the dynamic changes of nodes in the physical network. Therefore, the current service function chain orchestration method is not applicable to the dynamic nodes in the unmanned cluster platform, and the network function deployment of dynamically changing unmanned network nodes cannot be performed by applying static nodes. Summary of the invention

[0003] The purpose of this application is to provide a method and device for on-demand deployment of network functions applied to an unmanned cluster platform, so as to alleviate the above-mentioned technical problems existing in the prior art.

[0004] In a first aspect, the present invention provides a method for on-demand deployment of network functions applied to an unmanned cluster platform, the method comprising:

[0005] Pre-configure multiple constraints to constrain the placement of virtual network functions;

[0006] If one or more of the multiple constraints are met, responding to the service function chain placement request initiated by the unmanned platform base station, matching the unmanned cluster network resources with the service function chain placement request, and dynamically allocating the unmanned cluster network resources based on the service function chain placement request to generate service function chain orchestration information;

[0007] Deploy virtual network functions based on service function chain orchestration information.

[0008] In an optional implementation, the constraint conditions include at least resource quantity constraint, bandwidth constraint, and physical node placement constraint, wherein:

[0009] The resource quantity constraint is that the total amount of resources required for service function chain placement requests mapped to the virtual node does not exceed the remaining resources;

[0010] The bandwidth constraint is that the total bandwidth of the service function chain placement requests mapped to the virtual link does not exceed the remaining bandwidth of the virtual link;

[0011] The physical node placement constraint is that each virtual node can only be placed on one physical node.

[0012] In an optional implementation, in response to a service function chain placement request initiated by an unmanned platform base station, unmanned cluster network resources are matched with the service function chain placement request, and unmanned cluster network resources are dynamically allocated based on the service function chain placement request to generate service function chain orchestration information, including:

[0013] Constructed a deep reinforcement learning-based adaptive service function chain placement framework;

[0014] A graph convolutional network that extracts network features of unmanned platforms and captures the ordered information of service function chain placement requests to generate a sequence-to-sequence model for placement strategies;

[0015] The service function chain orchestration information is generated by processing the sequence-to-sequence model of the placement strategy to maximize the long-term average benefit and minimize the communication delay.

[0016] In an optional embodiment, the method further comprises:

[0017] Determine the deployment cost based on the remaining CPU ratio, remaining memory ratio, and remaining bandwidth ratio on the virtual node when deploying the service function chain;

[0018] The candidate deployment paths are sorted according to the deployment costs to obtain the target path for deploying the service function chain in the network.

[0019] In an optional implementation, in response to a service function chain placement request initiated by an unmanned platform base station, unmanned cluster network resources are matched with the service function chain placement request and unmanned cluster network resources are dynamically allocated based on the service function chain placement request to generate service function chain orchestration information, including:

[0020] The first current state of the unmanned cluster network is input into the graph neural network, and the nearest neighbor relationship between nodes is calculated by dynamically updating the graph structure to obtain the state information of the unmanned cluster network; wherein the first current state includes the resource usage, location information and network topology state of each drone node;

[0021] Input the second current state of the service function chaining request into the GRU encoder to obtain the hidden state of the current time step;

[0022] Generate a placement action decision set according to the last hidden state of the encoder, the current state of the GRU decoder and the deployment action; the placement action decision set is used to characterize the deployment order and deployment location of the service function chain in the physical network;

[0023] The correlation between the input sequence and the set of placement action decisions is calculated based on the context vector to generate service function chain orchestration information.

[0024] In an optional embodiment, the method further comprises:

[0025] Determine whether to dynamically expand the capacity of virtual network function instances in a service function chain based on communication delay.

[0026] In an optional implementation, determining whether to dynamically expand the capacity of a virtual network function instance in a service function chain based on communication delay includes:

[0027] Input a service function chaining orchestration request and a weighted topology graph corresponding to the service function chaining orchestration request;

[0028] Update the communication delay of the service function chain orchestration request according to the maximum number of iterations, calculate the deployment cost of the CPU of the virtual node when deploying the service function chain, the memory of the virtual node when deploying the service function chain, and the bandwidth on the virtual link when deploying the service function chain;

[0029] Determine whether the communication delay does not exceed the constraint of the maximum tolerable delay;

[0030] If yes, no adjustment is needed, end;

[0031] If not, then for the service function chaining request, follow Arrange the nodes in ascending order; where, Indicates the CPU of the virtual node when deploying the service function chain; Indicates the memory of the virtual node when deploying the service function chain;

[0032] judge Whether it does not exceed the preset threshold;

[0033] If yes, increase the CPU core allocation of the virtual node virtual network function and iteratively update the communication delay. If no, end.

[0034] Determine whether the updated communication delay does not exceed the constraint of the maximum tolerable delay;

[0035] If the requirement is met, the virtual network function instance corresponding to the communication delay requirement is expanded, and the new CPU core is used to allocate and iteratively calculate the new communication delay;

[0036] If not, expand the capacity of the VNF instance with the largest number of CPU cores.

[0037] In a second aspect, the present invention provides a device for on-demand deployment of network functions applied to an unmanned cluster platform, the device comprising:

[0038] A placement constraint module is used to pre-configure multiple constraints and to constrain the placement of virtual network functions through multiple constraints;

[0039] A placement request response module, configured to respond to a service function chain placement request initiated by an unmanned platform base station if one or more of the plurality of constraints are met, match unmanned cluster network resources with the service function chain placement request, dynamically allocate unmanned cluster network resources based on the service function chain placement request, and generate service function chain orchestration information;

[0040] The virtual network function deployment module is used to deploy the virtual network function according to the service function chain orchestration information.

[0041] In a third aspect, the present invention provides an electronic device comprising a processor and a memory, wherein the memory stores computer-executable instructions that can be executed by the processor, and the processor executes the computer-executable instructions to implement the method for on-demand deployment of network functions applied to an unmanned cluster platform of any of the aforementioned implementation modes.

[0042] In a fourth aspect, the present invention provides a computer-readable storage medium, which stores computer-executable instructions. When the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to implement the method for on-demand deployment of network functions applied to an unmanned cluster platform of any of the aforementioned implementation modes.

[0043] The method and device for on-demand deployment of network functions for unmanned cluster platforms provided by the present application have the following beneficial effects:

[0044] The embodiment of the present application performs placement constraints by pre-configuring multiple constraints, matches unmanned cluster network resources with the service function chain placement request, and dynamically allocates unmanned cluster network resources based on the service function chain placement request, generates service function chain orchestration information (SFC orchestration information), and then deploys virtual network functions according to the SFC orchestration information, thereby providing more accurate network status information and resource demand prediction. This method significantly improves the accuracy and efficiency of SFC orchestration, especially when facing rapid changes in network topology and resource status, and can quickly make optimization decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the specific implementation methods of the present application or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0046] Figure 1 A schematic diagram of an SFC coding framework provided in an embodiment of the present application;

[0047] Figure 2 A flowchart of a method for on-demand deployment of network functions applied to an unmanned cluster platform provided in an embodiment of the present application;

[0048] Figure 3 A schematic diagram of a weighted directed graph of an SFC request provided in an embodiment of the present application;

[0049] Figure 4 A schematic diagram of the SFC placement principle provided in an embodiment of the present application;

[0050] Figure 5 A flowchart of the expansion process of a virtual network function in a service function chain provided in an embodiment of the present application;

[0051] Figure 6 A structural diagram of a device for on-demand deployment of network functions for an unmanned cluster platform provided in an embodiment of the present application;

[0052] Figure 7 A structural diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations.

[0054] Therefore, the following detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the present application for which protection is sought, but merely represents selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.

[0055] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0056] The embodiment of the present application provides a method for on-demand deployment of network functions for an unmanned cluster platform, which can be applied to a variety of unmanned application scenarios, such as environmental monitoring, disaster relief, and intelligent transportation. In the embodiment of the present application, when arranging a service function chain (SFC), the SFC coding framework used is as follows: Figure 1 As shown, the framework mainly includes application layer, service layer, virtualization layer and physical layer.

[0057] 1) Application layer, which is used to publish different applications according to different scenarios, such as environmental monitoring applications, disaster relief applications, smart transportation applications, etc.

[0058] 2) A service layer, which is used to convert the application into a corresponding SFC request, wherein the SFC request includes various non-embedded virtual network function requests arranged in a specified order, and each application corresponds to a corresponding SFC request. In addition, when performing SFC arrangement, it can be assumed that each SFC request has a source node and a target node, the source node can be the first base station that issues the SFC request, and the target node can be the second base station that receives the SFC encoding information after SFC arrangement.

[0059] 3) Virtualization layer, used to match network resources with SFC requests. In the embodiment of the present application, the unmanned platform group network is abstracted into a resource pool, in which virtual resources can be dynamically allocated to meet the resource requirements of the SFC.

[0060] 4) Physical layer, corresponding to the unmanned platform cluster network environment composed of multiple unmanned platform nodes and base stations. As an edge node, the base station drives the SFC request orchestration process. Some base stations (the first base station mentioned above) can be selected as source nodes to publish SFC requests to their neighboring unmanned aerial vehicle (UAV) nodes, while other base stations other than the first base station (the second base station mentioned above) can be used as target nodes to receive SFC orchestration information from the last virtual network function embedded node.

[0061] Applied to the above SFC orchestration architecture, the present application embodiment provides a method for on-demand deployment of network functions applied to an unmanned cluster platform, see Figure 2 As shown, the method mainly includes the following steps:

[0062] Step S110, pre-configure a plurality of constraint conditions, and perform placement constraints on virtual network functions through the plurality of constraint conditions.

[0063] In one embodiment, to achieve on-demand deployment of SFC, the SFC request orchestration process corresponding to each application can be modeled. In one embodiment, for any application, the SFC request corresponding to it can be modeled as a weighted directed graph G i :

[0064] G i =(N i ,L i ,D i max ), where N irepresents the set of virtual network functions (VNFs), L i represents the set of virtual links, D i max is the maximum tolerable delay. i ∈N i The requested resource is passed express.

[0065] See also Figure 3 As shown in the figure, for the sake of convenience, four virtual network functions are taken as an example. 1 To N 4 is the virtual network function, L 1 To L 3 is a virtual link, the number of virtual network functions in each virtual network function (for example, N 1 10, 20, 5) represent the resources requested by the VNF, and the number corresponding to each virtual link (such as L 1 20 below, L 2 10 below, L 3 The 20) below are the maximum tolerable delays of the corresponding virtual links.

[0066] In one embodiment, two binary variables can be introduced pass Indicates the placement result of virtual nodes to physical nodes, through Indicates the placement result of the virtual link to the physical link. If n i (l i ) is mapped to n′(l′), which is the virtual node n i and virtual link l i The mapped physical node n' and physical link (l') are: 1 if both are true, otherwise 0.

[0067] When mapping nodes and links, the above constraints include at least resource quantity constraints, bandwidth constraints, and physical node placement constraints, where:

[0068] 1. The resource constraint is mapped to virtual node n i The total amount of resources k required by SFC request i on the scalar does not exceed the remaining resources, that is:

[0069] 2. Bandwidth constraints are mapped to virtual link l i The total bandwidth of SFC request i on Does not exceed its remaining bandwidth, that is:

[0070] 3. Physical node placement constraints for each virtual node VNFn iIt can only be placed on one physical node n′, that is:

[0071] If SFC request i is accepted, the path mapped in the physical network should traverse the VNFs in the order specified in the request, that is, satisfy Denote l(n′) and O(n′) as the set of incoming and outgoing links of physical node n′, and Indicates the source VNF and target VNF ​​of the virtual link. That is, when the above constraints are met, the path mapped in the physical network should traverse the virtual network functions VNF in the order specified in the SFC request.

[0072] Step S120: If one or more of the multiple constraints are met, respond to the service function chain placement request initiated by the unmanned platform base station, match the unmanned cluster network resources with the service function chain placement request, and dynamically allocate the unmanned cluster network resources based on the service function chain placement request to generate service function chain orchestration information.

[0073] In one implementation, in response to a service function chain placement request initiated by an unmanned platform base station, unmanned cluster network resources are matched with the service function chain placement request, and unmanned cluster network resources are dynamically allocated based on the service function chain placement request to generate service function chain orchestration information. In specific implementation, the following steps 1-1 to 1-3 may be included:

[0074] Step 1-1, constructing a deep reinforcement learning adaptive service function chain placement framework;

[0075] Step 1-2, extract the graph convolutional network of the unmanned platform group network features and capture the ordered information of the service function chain placement request to generate a sequence-to-sequence model of the placement strategy;

[0076] Steps 1-3 generate service function chain orchestration information by processing the sequence-to-sequence model of the placement strategy to maximize the long-term average benefit and minimize the communication delay.

[0077] In practical applications, when performing service function chain (SFC) orchestration, this application constructs an adaptive SFC placement framework based on deep reinforcement learning, combines a graph convolutional network that extracts network features of unmanned platform groups and captures ordered information of SFC requests to generate a sequence-to-sequence model for placement strategies, thereby maximizing long-term average benefits, minimizing latency, and determining the final SFC orchestration information.

[0078] In a specific implementation, the above-mentioned response to the service function chain placement request initiated by the unmanned platform base station, matching the unmanned cluster network resources with the service function chain placement request and dynamically allocating the unmanned cluster network resources based on the service function chain placement request, and generating the service function chain orchestration information may include the following steps 2-1 to 2-4:

[0079] Step 2-1, inputting the first current state of the unmanned cluster network into the graph neural network, and obtaining the state information of the unmanned cluster network by dynamically updating the graph structure to calculate the nearest neighbor relationship between nodes; wherein the first current state includes the resource usage, location information and network topology state of each drone node;

[0080] Step 2-2, inputting the second current state of the service function chaining request into the GRU encoder to obtain the hidden state of the current time step;

[0081] Step 2-3, generating a placement action decision set according to the last hidden state of the encoder, the current state of the GRU decoder and the deployment action; the placement action decision set is used to characterize the deployment order and deployment location of the service function chain in the physical network;

[0082] Step 2-4, calculate the correlation between the input sequence and the placement action decision set based on the context vector, and generate service function chain orchestration information.

[0083] In the deep reinforcement learning (DRL) framework, the interaction between the agent and the environment is defined as a Markov decision process (MDP). Applied to the above-mentioned adaptive SFC placement framework of deep reinforcement learning, the embodiment of the present application converts the placement of each SFC request i into an MDP within a limited range. In the decision-making process, the agent continuously selects consecutive physical nodes to place N i VNFs until the horizontal line T = N is reached i .

[0084] In the specific implementation, for the three key elements of Markov decision-making, namely, state, action and reward,

[0085] 1) The status includes the current unmanned cluster network status and SFC request requirements. The status is defined as in, The current status of the unmanned swarm network, such as the resource usage, location information, and network topology status of each drone node; is the current state of the SFC request, the amount of resources requested by the tth VNF It consists of the bandwidth required for the t-th virtual link and the number of remaining VNFs to be placed.

[0086] 2) The action of the agent selects node n0 from the candidate physical servers whose available resources exceed the requirements of the VNF request. Otherwise, the action forces a transition to the final state.

[0087] 3) The reward signal aims to encourage the agent to place SFCs to maximize the long-term average revenue. To this end, when the SFC request i is accepted at time step t = T (T is a time constant), the agent receives a reward where δ and μ are weight coefficients. In intermediate steps (i.e., t < T), when the resource constraints are met, the agent receives a small reward Otherwise it is negative, where Ψ is the reward coefficient.

[0088] To capture the ordered requirements of the SFC requests, in the specific implementation, refer to Figure 4 As shown, using the encoder of the Seq2Seq model implemented by a gated recurrent unit (GRU) network, a series of inputs are taken and a hidden state e t is generated. Specifically, for a given time step t, the GRU cell takes the current state of the current input unmanned cluster network and the hidden state e t-1 from the previous timestamp as inputs, and then outputs the hidden state e t for the current time step, that is:

[0089]

[0090] Furthermore, an attention decoder of the Seq2Seq model is used to generate appropriate actions. The decoder obtains the last state e T from the encoder, and based on the current state d t of the decoder and the action a t-1 generates an output a of size T = (a1; …; a T ), where:

[0091]

[0092] The above actions refer to selecting which node to embed. Specifically, each action (such as a1, a2, …… a _t-1 ) represents selecting a physical node at a given moment to deploy the current virtual network function (VNF).

[0093] a refers to the set of actions from 1 to T, which is the result of generating the seq2seq model. These action sets (a1, a2, ..., aT) represent the placement strategy of the entire service function chain (SFC), specifically including the embedding location of each virtual network function (VNF) in the physical network. Unlike a single action, a represents the set of all decisions in the entire placement process. It directly determines the deployment order and location of the SFC in the physical network. Generate a = (a1; ...; a T ) is mainly achieved by combining deep reinforcement learning and dynamic graph convolutional network (DGGCN). The following are more detailed generation steps:

[0094] 1. State representation: The state of the current physical network is represented as the input of the dynamic graph convolutional network (DGGCN) to extract node features.

[0095] 2. SFC request encoding: Use gated recurrent unit (GRU) to encode SFC request into hidden state.

[0096] 3. Action generation: Based on the current hidden state and physical network state, the attention mechanism decoder of the Seq2Seq model generates actions, that is, selecting appropriate physical nodes to place VNFs.

[0097] 4. Action selection: The softmax function is used to calculate the probability distribution of the action and select a physical node for VNF deployment.

[0098] Furthermore, when processing the reasonable placement order of inferring SFC requests, the correlation between the input sequence and the output sequence can be calculated based on the context-based attention mechanism. The context vector is calculated as e j The weighted sum of is given by:

[0099]

[0100] Among them, T is the horizontal line T = N i , j is a substitution, indicating that j can be 1 to T. Since the correlation between the input sequence and the output sequence is calculated, α t,j is the attention weight, which represents the decoder's hidden state d at time step t to the encoder at time step j j The importance of each part of the input sequence is used to characterize the importance of the current decoding step.

[0101] These weights are normalized using the softmax function to ensure that their sum is 1, thus forming a probability distribution.

[0102] Step S130: deploy the virtual network function according to the service function chain orchestration information.

[0103] In one implementation, a physical node may be selected based on conditional probability to accommodate the current virtual network function VNF, and the virtual network function may be deployed.

[0104] After obtaining the parameters calculated in the above steps, in a specific implementation, in order to obtain the probability distribution of the candidate actions, the current state of the decoder e t , current context vector c t and the flattened output z of the DGCNN layer t Connect them and then convert them into a fully connected layer so that the final output matches the number of physical network nodes. Candidate actions refer to the set of physical nodes that the agent can choose at a specific time t. Specifically, candidate actions include all physical nodes whose current resources are sufficient to meet the needs of virtual network functions (VNFs). Each action represents placing the VNF that currently needs to be placed on a candidate physical node.

[0105] Dynamic graph convolutional network (DGCNN) has significant advantages in dealing with dynamic changes in unmanned cluster networks. The specific operation is as follows:

[0106] 1. Input processing: The current state of the unmanned cluster network Input to DGCNN. These states include resource usage, location information, and network topology status of each drone node.

[0107] 2. Feature extraction: DGCNN can capture the temporal dependencies and spatial structures in the network through its structure. It uses a dynamically updated graph structure so that each layer of the network can recalculate the nearest neighbor relationship between nodes, thereby accurately capturing local and global topological changes.

[0108] 3. Output generation: After being processed by DGCNN, the output Zt is used as a feature vector, which contains the comprehensive information of the unmanned cluster network, providing more accurate network status information and resource demand prediction for subsequent SFC orchestration.

[0109] Through dynamic graph updates and convolution operations, DGCNN can capture more detailed dependencies between nodes, significantly improving the accuracy of modeling network status. When faced with rapid changes in network topology and resource status, DGCNN can quickly make optimization decisions and improve the accuracy and efficiency of SFC orchestration.

[0110] Further, the following is an explanation of the agent-related content in deep learning:

[0111] 1. Agent: In the deep reinforcement learning framework, an agent is an intelligent entity that takes actions based on the observed state in a given environment to maximize its cumulative reward. In the service function chain (SFC) deployment, the agent places virtual network functions (VNFs) by selecting physical nodes and adjusts its strategy based on feedback. The agent selects an action (such as selecting a physical node) based on the current state at each time step, and then interacts with the environment to obtain rewards and the next state. Through continuous interaction and learning, the agent is able to optimize its decision-making strategy.

[0112] 2. Worker Agent:

[0113] Roles: Multiple worker agents perform environment exploration and experience gathering in parallel.

[0114] Functionality: Each worker agent consists of two networks: an actor network and a critic network. The actor network generates actions, and the critic network evaluates the value of actions.

[0115] Relationship: Workers synchronize parameters from the master agent, interact with the environment independently, collect experience data, and then send this data to the master agent for global parameter updates.

[0116] 3. Master Agent:

[0117] Role: The master agent coordinates the training process of multiple worker agents.

[0118] Function: The master agent stores and updates the global shared parameters and distributes the updated parameters to each worker agent.

[0119] Relationship: The master agent and worker agents work together through parameter synchronization and experience data sharing. The master agent is responsible for global parameter optimization, while the worker agent is responsible for parallel environment exploration and strategy improvement.

[0120] The agent performs a deploy VNF operation and selects a physical node to accommodate the current virtual network function VNF according to the conditional probability, which is:

[0121]

[0122] Under the conditional probability calculation formula, the probability distribution of candidate physical nodes is generated, indicating the possibility of each node being selected, so that a physical node can be selected to accommodate the current VNF ​​according to the calculated probability distribution.

[0123] The embodiment of the present application innovatively adopts a dynamic graph convolutional network (Dynamic Graph CNN, DGCNN), which can handle the dynamic change problem in the unmanned platform cluster network. DGCNN can effectively capture the time dependency and spatial structure in the network, and provide more accurate network status information and resource demand prediction for SFC scheduling. This method significantly improves the accuracy and efficiency of SFC scheduling, especially when facing rapid changes in network topology and resource status, and can quickly make optimization decisions.

[0124] By combining deep reinforcement learning (DRL) and the service function chain orchestration strategy of dynamic graph neural network, it is possible to intelligently perform SFC orchestration based on the dynamic changes of the unmanned platform cluster network and the specific needs of SFC requests. By maximizing the long-term average benefit and reducing the overall delay, the purpose of multi-objective optimization is achieved, thereby meeting the requirements of service quality QoS.

[0125] The above long-term average return is calculated as follows:

[0126]

[0127]

[0128] Among them, revenue(i) is the income, μk is the weight of node resources in the income, α is the weight of link resources in the income, R(π) is the long-term average revenue, and τ is time.

[0129] Furthermore, the above method also includes: determining the deployment cost based on the remaining CPU ratio, remaining memory ratio and remaining bandwidth ratio on the virtual node when deploying the service function chain; sorting the deployment candidate paths according to the deployment cost to obtain the target path for deploying the service function chain in the network.

[0130] In order to obtain the best path for deploying SFC in the network, the embodiment of the present application defines "deployment cost" for physical links and nodes, which can be used to indicate overloaded nodes, thereby balancing network load and ultimately reducing resource bottlenecks. In one embodiment, the deployment cost of path l when deploying SFC request i is defined as:

[0131]

[0132] Among them, N l represents the nodes along the selected path l; ε l represents the links along the selected path l, Indicates the CPU of virtual node n when deploying SFCi; Indicates the memory of virtual node n when SFCi is deployed; It represents the deployment cost of the bandwidth on the virtual link (u, v) when SFC i is deployed. The specific calculation method is as follows:

[0133]

[0134]

[0135]

[0136] in, Indicates the remaining CPU ratio on node n; Indicates the remaining memory ratio on node n; represents the remaining bandwidth ratio on the link (u, v), When the network load is low, The value of is relatively small and increases slowly, whereas if resource consumption is compared to resource capacity, the value will be very large and grow rapidly.

[0137] Based on the above parameters, it can be seen that if the deployment cost R i,l is small, there is no bottleneck at the edge node, and the link ε on the selected path l There is no bottleneck. On the contrary, if the deployment cost of the selected path is R i,l If the cost is very large, there must be a bottleneck at the edge node or some overloaded links along the path l. Therefore, in order to improve the deployment stability, we can first select other paths with lower costs to deploy SFC i, and then i,l The candidate paths are sorted in ascending order to obtain the best path for deploying SFC in the network.

[0138] The present application also provides a specific implementation of GRU network training. In an optional implementation, A3C (a "master-slave" parallel training architecture) can be used for network training, which consists of multiple working agents and a master agent to speed up the training and improve the robustness of the model. Each agent contains two networks: i) Actor network: maintains a parameterized placement strategy π θ (a t |s t ), and generates an action based on the current state; ii) Critic network: through V ω (s t , a t) estimates the Q-value to evaluate whether the action is good or bad, and Vω(st,at) represents the value function, which is used to evaluate the quality or value of a given state-action pair (st,at). The value function estimates the expected return (accumulated future reward) of taking action at from the current state st and then following the policy. After obtaining the global shared parameters from the master agent, each worker agent u∈U initializes its own parameters using θ and ω, explores the environment to collect experience, and then sends them to the master agent to update the global shared parameters, where θ usually represents the set of parameters of the policy network (or actor network). The policy network uses these parameters to generate a probability distribution of choosing different actions in a given state. In this algorithm, the actor network updates its policy parameters θ by maximizing the cumulative reward. These parameters control how the policy network chooses the best action based on the current state. ω usually represents the set of parameters of the value function (or critic network). The critic network uses these parameters to estimate the value of a given state or state-action pair. In this algorithm, the critic network updates the parameters ω by minimizing the square of the temporal difference error (TD error). These parameters are used to estimate the long-term value of taking an action in a specific state.

[0139] At each time step t, each agent follows a random policy Select an action t and interact with the environment to obtain rewards r t and the next state s t+1 The critic network computes the temporal difference error (TD error) to evaluate the new state, which is given as follows:

[0140] A(a t ,s t )=r t +γV ω (s t+1 )-V ω (s t )

[0141] Among them, A(a t ,s t ) is the time difference error, γ∈(0;1) represents the discount factor, V ω (s t , a t ) is the evaluation function, V ω (s t , a t ) represents the value function of the next state st+1, that is, the expected cumulative return that can be obtained in the state st+1, and Vω(st) represents the value function of the current state st, that is, the expected cumulative return that can be obtained in the state st.

[0142] After fully processing an SFC placement, the critic network updates the parameter ω by minimizing the squared TD error loss, i.e.:

[0143]

[0144] Furthermore, the actor network updates the policy parameter θ to maximize the cumulative reward in the following way:

[0145]

[0146] Where θ represents the policy parameter; logπ θ (a t |s t )A(a t ,s t ) represents the cross entropy loss weighted by the TD error; ε θ represents the learning rate of the actor network; T represents the total number of time steps, which means the number of time steps involved in a complete sequence; Represents the gradient operator, which means the gradient calculation of the parameter θ.

[0147] After obtaining the updated policy parameters θ, the final network model is trained, that is, the adaptive SFC placement framework of deep reinforcement learning, so as to deploy virtual network functions based on the adaptive SFC placement framework of deep reinforcement learning.

[0148] In an optional embodiment, the method further comprises:

[0149] Determining whether to dynamically expand the capacity of a virtual network function instance in a service function chain based on communication delay. Specifically, determining whether to dynamically expand the capacity of a virtual network function instance in a service function chain based on communication delay may include the following steps 3-1 to 3-10:

[0150] Step 3-1, input a service function chaining orchestration request and a weighted topology graph corresponding to the service function chaining orchestration request;

[0151] Step 3-2, updating the communication delay of the service function chain orchestration request according to the maximum number of iterations, calculating the deployment cost of the CPU of the virtual node when deploying the service function chain, the memory of the virtual node when deploying the service function chain, and the bandwidth on the virtual link when deploying the service function chain;

[0152] Step 3-3, determine whether the communication delay does not exceed the constraint of the maximum tolerable delay;

[0153] Step 3-4, if yes, no adjustment is needed, end;

[0154] Step 3-5, if not, then for the service function chaining request, follow Arrange the nodes in ascending order; where, Indicates the CPU of the virtual node when deploying the service function chain; Indicates the memory of the virtual node when deploying the service function chain;

[0155] Step 3-6, judgement Whether it does not exceed the preset threshold;

[0156] Step 3-7, if yes, increase the CPU core allocation of the virtual network function of the virtual node and iteratively update the communication delay, if no, end;

[0157] Step 3-8, determining whether the updated communication delay does not exceed the constraint of the maximum tolerable delay;

[0158] Step 3-9: If the requirements are met, then the virtual network function instance corresponding to the communication delay requirement is expanded, and the new CPU core is used for allocation and iterative calculation of the new communication delay;

[0159] In step 3-10, if the requirements are not met, expand the capacity of the virtual network function instance with the largest number of CPU cores.

[0160] In one implementation, the SFC request i communication delay can be expressed as:

[0161]

[0162] is the communication delay of SFC request i on virtual link, which consists of propagation delay and transmission delay, that is, virtual link l i (u,v) The communication delay of SFC request i can be expressed as:

[0163]

[0164] in, is the propagation delay, which indicates the propagation delay on the physical link that the virtual link passes through, and is related to the distance between adjacent physical edges; is the transmission delay, which is calculated by dividing the size of the transmitted data packet by the bandwidth capacity allocated to the virtual link, that is: Transmit packet size for SFC request i.

[0165] In one embodiment, the communication delay of SFC request i may be constrained not to exceed the tolerable maximum delay constraint D i max , that is,

[0166] Furthermore, in practical applications, for the need to dynamically expand the capacity of VNF, the embodiment of the present application provides a virtual network function expansion strategy in the service function chain, so as to adapt to the rapid changes in network traffic demand and improve resource utilization. In one embodiment, when performing the expansion processing of the virtual network function in the service function chain, refer to Figure 5 As shown, the following steps may be included:

[0167] Step 4-1: Enter SFC request G i and weighted topology graph G', maximum number of iterations.

[0168] Step 4-2: Update the communication delay of the SFC request calculate and

[0169] Step 4-3: Determine communication delay Does it not exceed the maximum tolerable delay constraint D? i max If yes, no adjustment is needed and the process ends; if no, proceed to step 5-4.

[0170] Step 4-4: For SFC request i, follow Sort the nodes in ascending order.

[0171] Step 4-5: Judgment Does it not exceed the threshold ε? If yes, go to step 4-6; if no, end.

[0172] Step 4-6: Increase the CPU core allocation of the VNF of the node and iteratively calculate the new communication delay. The iterative calculation formula is as mentioned above

[0173] Step 4-7: Based on Determine whether the updated communication delay meets the requirements.

[0174] Step 4-8: If satisfied, scale up these VNF instances to use the new CPU core allocation scheme.

[0175] Step 4-9: If not satisfied, expand the capacity of the VNF instance with the largest number of CPU cores.

[0176] In this way, when it is detected that the service delay requested by the SFC exceeds the requirement, the historical network load of the physical node is first calculated, and the node with lower network load is selected to expand the VNF instance. When expanding, the node with more remaining resources is preferentially selected to ensure service continuity.

[0177] Based on the above method embodiment, the present application embodiment also provides a device for on-demand deployment of network functions for an unmanned cluster platform, see Figure 6 As shown, the device mainly includes the following parts:

[0178] A placement constraint module 610 is used to pre-configure a plurality of constraint conditions and to perform placement constraints on virtual network functions through the plurality of constraint conditions;

[0179] A placement request response module 620 is used to respond to a service function chain placement request initiated by the unmanned platform base station if one or more of the plurality of constraints are met, match the unmanned cluster network resources with the service function chain placement request, dynamically allocate the unmanned cluster network resources based on the service function chain placement request, and generate service function chain orchestration information;

[0180] The virtual network function deployment module 630 is used to deploy the virtual network function according to the service function chain arrangement information.

[0181] In a feasible implementation, the constraints include at least resource quantity constraints, bandwidth constraints, and physical node placement constraints, wherein:

[0182] The resource quantity constraint is that the total amount of resources required for service function chain placement requests mapped to the virtual node does not exceed the remaining resources;

[0183] The bandwidth constraint is that the total bandwidth of the service function chain placement requests mapped to the virtual link does not exceed the remaining bandwidth of the virtual link;

[0184] The physical node placement constraint is that each virtual node can only be placed on one physical node.

[0185] In a feasible implementation manner, the placement request response module 620 is used to:

[0186] Constructed a deep reinforcement learning-based adaptive service function chain placement framework;

[0187] A graph convolutional network that extracts network features of unmanned platforms and captures the ordered information of service function chain placement requests to generate a sequence-to-sequence model for placement strategies;

[0188] The service function chain orchestration information is generated by processing the sequence-to-sequence model of the placement strategy to maximize the long-term average benefit and minimize the communication delay.

[0189] In a feasible implementation manner, the above-mentioned device further includes a deployment path determination module, which is used to:

[0190] Determine the deployment cost based on the remaining CPU ratio, remaining memory ratio, and remaining bandwidth ratio on the virtual node when deploying the service function chain;

[0191] The candidate deployment paths are sorted according to the deployment costs to obtain the target path for deploying the service function chain in the network.

[0192] In a feasible implementation manner, the placement request response module 620 is used to:

[0193] The first current state of the unmanned cluster network is input into the graph neural network, and the nearest neighbor relationship between nodes is calculated by dynamically updating the graph structure to obtain the state information of the unmanned cluster network; wherein the first current state includes the resource usage, location information and network topology state of each drone node;

[0194] Input the second current state of the service function chaining request into the GRU encoder to obtain the hidden state of the current time step;

[0195] Generate a placement action decision set according to the last hidden state of the encoder, the current state of the GRU decoder and the deployment action; the placement action decision set is used to characterize the deployment order and deployment location of the service function chain in the physical network;

[0196] The correlation between the input sequence and the set of placement action decisions is calculated based on the context vector to generate service function chain orchestration information.

[0197] In a feasible implementation manner, the above device further includes: a capacity expansion module, which is used to:

[0198] Determine whether to dynamically expand the capacity of virtual network function instances in a service function chain based on communication delay.

[0199] In a feasible implementation manner, the above expansion module is used to:

[0200] Input a service function chaining orchestration request and a weighted topology graph corresponding to the service function chaining orchestration request;

[0201] Update the communication delay of the service function chain orchestration request according to the maximum number of iterations, calculate the deployment cost of the CPU of the virtual node when deploying the service function chain, the memory of the virtual node when deploying the service function chain, and the bandwidth on the virtual link when deploying the service function chain;

[0202] Determine whether the communication delay does not exceed the constraint of the maximum tolerable delay;

[0203] If yes, no adjustment is needed, end;

[0204] If not, then for the service function chaining request, follow Arrange the nodes in ascending order; where, Indicates the CPU of the virtual node when deploying the service function chain; Indicates the memory of the virtual node when deploying the service function chain;

[0205] judge Whether it does not exceed the preset threshold;

[0206] If yes, increase the CPU core allocation of the virtual node virtual network function and iteratively update the communication delay. If no, end.

[0207] Determine whether the updated communication delay does not exceed the constraint of the maximum tolerable delay;

[0208] If the requirement is met, the virtual network function instance corresponding to the communication delay requirement is expanded, and the new CPU core is used to allocate and iteratively calculate the new communication delay;

[0209] If not, expand the capacity of the VNF instance with the largest number of CPU cores.

[0210] The device for on-demand deployment of network functions for unmanned cluster platforms provided in the embodiments of the present application has the same implementation principle and technical effects as those of the aforementioned method embodiments. For the sake of brief description, for parts not mentioned in the embodiment of the device for on-demand deployment of network functions for unmanned cluster platforms, reference may be made to the corresponding contents in the aforementioned method embodiment for on-demand deployment of network functions for unmanned cluster platforms.

[0211] The present application also provides an electronic device, such as Figure 7 As shown, it is a structural diagram of the electronic device, wherein the electronic device 100 includes a processor 71 and a memory 70, the memory 70 stores computer executable instructions that can be executed by the processor 71, and the processor 71 executes the computer executable instructions to implement any of the above-mentioned methods for on-demand deployment of network functions applied to unmanned cluster platforms.

[0212] exist Figure 7 In the illustrated embodiment, the electronic device further includes a bus 72 and a communication interface 73 , wherein the processor 71 , the communication interface 73 and the memory 70 are connected via the bus 72 .

[0213] Among them, the memory 70 may include a high-speed random access memory (RAM), and may also include a non-volatile memory (non-volatile memory), such as at least one disk storage. The communication connection between the system network element and at least one other network element is realized through at least one communication interface 73 (which can be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. can be used. The bus 72 can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. The bus 72 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 7 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0214] The processor 71 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit or software instructions in the processor 71. The above processor 71 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor. The steps of the method disclosed in the embodiment of the present application can be directly embodied as a hardware decoding processor to execute, or the hardware and software modules in the decoding processor can be executed. The software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory, and the processor 71 reads the information in the memory and completes the steps of the method for on-demand deployment of network functions applied to an unmanned cluster platform in the aforementioned embodiment in combination with its hardware.

[0215] An embodiment of the present application also provides a computer-readable storage medium, which stores computer-executable instructions. When the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to implement the above-mentioned method for on-demand deployment of network functions applied to an unmanned cluster platform. The specific implementation can be found in the aforementioned method embodiment, which will not be repeated here.

[0216] The computer program product of the method and apparatus for on-demand deployment of network functions for unmanned cluster platforms provided in the embodiments of the present application includes a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the methods in the previous method embodiments. For specific implementations, please refer to the method embodiments, which will not be repeated here.

[0217] Unless otherwise specifically stated, the relative steps, numerical expressions and values ​​of the components and steps set forth in these embodiments do not limit the scope of the present application.

[0218] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a non-volatile computer-readable storage medium that is executable by a processor. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, and other media that can store program codes.

[0219] In the description of the present application, it should be noted that the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0220] In the description of this application, it should also be noted that, unless otherwise clearly specified and limited, the terms "set", "install", "connect", and "connect" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two elements. For ordinary technicians in this field, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0221] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for on-demand deployment of network functions for unmanned cluster platforms, It is characterized in that The method comprises: Pre-configure multiple constraints to constrain the placement of virtual network functions; If one or more of the multiple constraints are met, responding to a service function chain placement request initiated by the unmanned platform base station, matching unmanned cluster network resources with the service function chain placement request, dynamically allocating unmanned cluster network resources based on the service function chain placement request, and generating service function chain orchestration information; The virtual network function is deployed according to the service function chain orchestration information.

2. The method for on-demand deployment of network functions applied to an unmanned cluster platform according to claim 1, It is characterized in that The constraints include at least resource quantity constraints, bandwidth constraints, and physical node placement constraints, wherein: The constraint on the amount of resources is that the total amount of resources required for the service function chain placement request mapped to the virtual node does not exceed the remaining amount of resources; The bandwidth constraint is that the total bandwidth of the service function chain placement request mapped to the virtual link does not exceed the remaining bandwidth of the virtual link; The physical node placement constraint is that each virtual node can only be placed on one physical node.

3. The method for on-demand deployment of network functions applied to an unmanned cluster platform according to claim 1, It is characterized in that Responding to a service function chain placement request initiated by an unmanned platform base station, matching unmanned cluster network resources with the service function chain placement request, dynamically allocating unmanned cluster network resources based on the service function chain placement request, and generating service function chain orchestration information, including: Constructed a deep reinforcement learning-based adaptive service function chain placement framework; A graph convolutional network that extracts network features of unmanned platforms and captures the ordered information of service function chain placement requests to generate a sequence-to-sequence model for placement strategies; The service function chain orchestration information is generated by processing the sequence-to-sequence model of the placement strategy to maximize the long-term average benefit and minimize the communication delay.

4. The method for on-demand deployment of network functions applied to an unmanned cluster platform according to claim 3, It is characterized in that The method further comprises: Determine the deployment cost based on the remaining CPU ratio, remaining memory ratio, and remaining bandwidth ratio on the virtual node when deploying the service function chain; The candidate deployment paths are sorted according to the deployment costs to obtain the target path for deploying the service function chain in the network.

5. The method for on-demand deployment of network functions applied to an unmanned cluster platform according to claim 1, It is characterized in that Responding to a service function chain placement request initiated by an unmanned platform base station, matching unmanned cluster network resources with the service function chain placement request and dynamically allocating unmanned cluster network resources based on the service function chain placement request, and generating service function chain orchestration information, including: The first current state of the unmanned cluster network is input into the graph neural network, and the nearest neighbor relationship between nodes is calculated by dynamically updating the graph structure to obtain the state information of the unmanned cluster network; wherein the first current state includes the resource usage, location information and network topology state of each drone node; Input the second current state of the service function chaining request into the GRU encoder to obtain the hidden state of the current time step; Generate a placement action decision set according to the last hidden state of the encoder, the current state of the GRU decoder and the deployment action; the placement action decision set is used to characterize the deployment order and deployment position of the service function chain in the physical network; The correlation between the input sequence and the set of placement action decisions is calculated based on the context vector to generate service function chain orchestration information.

6. The method for on-demand deployment of network functions applied to an unmanned cluster platform according to claim 1, It is characterized in that The method further comprises: Determine whether to dynamically expand the capacity of virtual network function instances in a service function chain based on communication delay.

7. The method for on-demand deployment of network functions applied to an unmanned cluster platform according to claim 6, It is characterized in that Determining whether to dynamically expand the capacity of a virtual network function instance in a service function chain based on communication latency includes: Input a service function chaining arrangement request and a weighted topology graph corresponding to the service function chaining arrangement request; Update the communication delay of the service function chain orchestration request according to the maximum number of iterations, calculate the deployment cost of the CPU of the virtual node when deploying the service function chain, the memory of the virtual node when deploying the service function chain, and the bandwidth on the virtual link when deploying the service function chain; Determine whether the communication delay does not exceed the constraint of the maximum tolerable delay; If yes, no adjustment is needed, end; If not, then for the service function chaining request, follow Arrange the nodes in ascending order; where, Indicates the CPU of the virtual node when deploying the service function chain; Indicates the memory of the virtual node when deploying the service function chain; judge Whether it does not exceed the preset threshold; If yes, increase the CPU core allocation of the virtual node virtual network function and iteratively update the communication delay. If no, end. Determine whether the updated communication delay does not exceed the constraint of the maximum tolerable delay; If the requirement is met, the virtual network function instance corresponding to the communication delay requirement is expanded, and the new CPU core is used to allocate and iteratively calculate the new communication delay; If not, expand the capacity of the VNF instance with the largest number of CPU cores.

8. A device for on-demand deployment of network functions for unmanned cluster platforms, It is characterized in that The device comprises: A placement constraint module is used to pre-configure multiple constraints and to constrain the placement of virtual network functions through multiple constraints; A placement request response module, configured to respond to a service function chain placement request initiated by an unmanned platform base station if one or more of the plurality of constraints are met, match unmanned cluster network resources with the service function chain placement request, dynamically allocate unmanned cluster network resources based on the service function chain placement request, and generate service function chain orchestration information; A virtual network function deployment module is used to deploy virtual network functions according to the service function chain orchestration information.

9. An electronic device, It is characterized in that It includes a processor and a memory, the memory stores computer-executable instructions that can be executed by the processor, and the processor executes the computer-executable instructions to implement the method for on-demand deployment of network functions applied to an unmanned cluster platform as described in any one of claims 1 to 7.

10. A computer-readable storage medium, It is characterized in that The computer-readable storage medium stores computer-executable instructions. When the computer-executable instructions are called and executed by the processor, the computer-executable instructions prompt the processor to implement the method for on-demand deployment of network functions applied to an unmanned cluster platform as described in any one of claims 1 to 7.

Citation Information

Cited By

  • Novel power system satellite-ground converged communication network modeling and resource arrangement method and system

    CN119652774A