Optimized mapping method of service function chain in converged network architecture and related device

Abstract

The application provides a method for optimizing mapping of service function chains in a fusion network architecture and related equipment, the fusion network comprising a plurality of server nodes. The method comprises: obtaining a service function chain request and resource information of each server node; wherein the service function chain request comprises demand information of a plurality of virtual network functions in the service function chain; constructing a Markov decision process model according to the resource information and the demand information; wherein the Markov decision process model represents a mapping relationship between the server nodes and the virtual network functions; solving the Markov decision process model to obtain a mapping strategy, and executing the mapping strategy. The scheme of the application flexibly coordinates and schedules network resources, reduces resource congestion of the network, and meets the demand of low latency, high reliability, large bandwidth and other power business.

Description

Technical Field

[0001] This application relates to the field of communication technology, and in particular to an optimized mapping method and related equipment for service function chains in a converged network architecture. Background Technology

[0002] Network Function Virtualization (NFV) technology involves transferring hardware devices to virtual machines, thereby improving service flexibility and network openness. In NFV, Service Function Chains (SFCs) consist of ordered Virtual Network Functions (VNFs). VNFs provide flexibility by enabling the dynamic deployment and interconnection of network functions, thus realizing SFCs.

[0003] Effective Service Function Chain (SFC) mapping can flexibly handle massive data streams and perform filtering, learning, usage, compression, and processing, providing efficient, scalable, and economical network services for end-user IoT applications. However, under the requirements of low latency and high reliability in power network services, NFV-based networks have higher availability requirements than traditional networks, and the service function chain mapping schemes in related technologies cannot meet these needs. Summary of the Invention

[0004] In view of this, the purpose of this application is to propose an optimized mapping method and related equipment for service function chains in a converged network architecture, so as to solve or partially solve the above problems.

[0005] In a first aspect, this application provides an optimized mapping method for service function chains in a converged network architecture, wherein the converged network architecture includes a plurality of server nodes;

[0006] The method includes:

[0007] Obtain service function chain requests and resource information for each server node; wherein, the service function chain request includes the requirement information of several virtual network functions in the service function chain;

[0008] Based on the resource information and the demand information, a Markov decision process model is constructed; wherein, the Markov decision process model represents the mapping relationship between the server node and the virtual network function;

[0009] The mapping strategy is obtained by solving the Markov decision process model, and the mapping strategy is executed.

[0010] Optionally, solving the Markov decision process model to obtain the mapping strategy includes:

[0011] Based on the type of power service requested by the service function chain, determine the priority strategy and reward function;

[0012] Establish constraints corresponding to the resource information, and construct a joint optimization objective model based on the constraints;

[0013] Based on the priority strategy and the constraints, feasible nodes are selected in the converged network architecture.

[0014] In the feasible node, with the goal of maximizing the reward function, the joint optimization objective model is solved using the Markov decision process model to obtain the solution result, and the mapping strategy is determined based on the solution result and the reward function.

[0015] Optionally, the converged network architecture includes a 5G-based power converged access layer and a 5G-MEC multi-access edge computing layer; the power service types include latency-sensitive services and latency-tolerant services.

[0016] The step of determining the priority strategy based on the power service type requested by the service function chain includes:

[0017] In response to determining that the power service type of the service function chain request is a latency-sensitive service, the priority strategy is as follows:

[0018] From mapping the predecessor node to each neighbor node in the subnet of the 5G-based power converged access layer; from mapping the predecessor node to the cluster head node in the subnet of the 5G-based power converged access layer; from mapping the predecessor node to each node in the 5G-MEC multi-access edge computing layer;

[0019] In response to determining that the power service type of the service function chain request is a latency-tolerant service, the priority strategy is as follows:

[0020] From mapping the predecessor node to the cluster head node of the subnet based on the 5G-based power converged access layer; from mapping the predecessor node to each node in the 5G-MEC multi-access edge computing layer;

[0021] The predecessor node represents the previous node that was successfully mapped when mapping the current node.

[0022] Optionally, the resource information includes the total computing resources, total storage resources, and total bandwidth resources of the server node; the demand information includes the computing resource requirements, storage resource requirements, and bandwidth resource requirements of the virtual network function.

[0023] The step of establishing constraints corresponding to the resource information and constructing a joint optimization objective model based on the constraints includes:

[0024] Based on the total computing resources, total storage resources, and total bandwidth resources of the server nodes, establish latency constraints, remaining computing resource constraints, remaining storage resource constraints, and remaining bandwidth resource constraints.

[0025] Based on the aforementioned latency constraints, remaining computing resource constraints, remaining storage resource constraints, and remaining bandwidth resource constraints, a joint optimization objective model is constructed.

[0026] The time delay constraint is:

[0027]

[0028] Among them, T r Indicates the tolerable latency of the service function chain; t r The total response delay represents the sum of the communication delay of the physical link consisting of the server nodes mapped by the service function chain and the processing delay of the virtual network functions on all server nodes of the physical link; r∈R represents the set of service function chains.

[0029] The remaining computing resource constraints are as follows:

[0030]

[0031] in, This represents the remaining computing resources of server node v in time slot τ; This represents the total computing resources of server node v; f∈F represents the set of virtual network functions. This indicates the number of virtual network functions f mapped on server node v in time slot τ; This represents the computational resource requirements of virtual network functions; v∈V,V * This represents the set of server nodes based on the 5G-based power converged access layer and the 5G-MEC multi-access edge computing layer;

[0032] The constraints on the remaining storage resources are as follows:

[0033]

[0034] in, This represents the remaining storage resources of server node v in time slot τ; This represents the total storage resources of server node v; This indicates the storage resource requirements of the virtual network function;

[0035] The remaining bandwidth resource constraints are as follows:

[0036]

[0037] in, W represents the remaining bandwidth resources of the server node in time slot τ; v This represents the total bandwidth resources of the server node; y r ∈{0,1} indicates whether the service function chain r has been successfully mapped; This indicates the bandwidth requirements of the service function chain r; This indicates whether a virtual network function f is mapped on the server node v in time slot τ.

[0038] Optionally, the step of selecting feasible nodes in the converged network architecture based on the priority strategy and the constraints includes:

[0039] Based on the priority strategy, feasible nodes are searched in the converged network architecture to obtain feasible nodes; wherein, the feasible nodes satisfy the latency constraints, remaining computing resource constraints, remaining storage resource constraints, and remaining bandwidth resource constraints.

[0040] Optionally, the joint optimization objective function of the joint optimization objective model is:

[0041]

[0042] Where, ξ cpu This represents the computational cost per unit of server resources; ξ mem This represents the processing cost per unit of server resources; ξ W This represents the unit cost of bandwidth consumption.

[0043] Optionally, the reward function is:

[0044]

[0045] Wherein, R1 represents the reward function for a successful mapping of a power service type requested by the service function chain as a latency-sensitive service; R′1 represents the reward function for a failed mapping of a power service type requested by the service function chain as a latency-sensitive service; R2 represents the reward function for a successful mapping of a power service type requested by the service function chain as a latency-tolerant service; R′2 represents the reward function for a failed mapping of a power service type requested by the service function chain as a latency-tolerant service; κ1 represents the first weight coefficient; κ2 represents the second weight coefficient; κ3 represents the third weight coefficient; Ω r,s1 ,Ω r,s2 for Evaluation indicators; Ω r,s1 The first number represents the number of server nodes where the successfully mapped virtual network function resides in the 5G-based power converged access layer; Ω r,s2The second number represents the number of server nodes where the successfully mapped virtual network function resides in the 5G-MEC multi-access edge computing layer; sen r This is determined based on the type of power service requested according to the service function chain.

[0046] Optionally, constructing the Markov decision process model includes:

[0047] For each of the feasible nodes, determine the corresponding state and action;

[0048] The set of states of all feasible nodes is taken as the state space, and the set of actions is taken as the action space.

[0049] Based on the state space and the action space, a Markov decision process model is constructed.

[0050] The state space is as follows: for each state

[0051] in, This represents the remaining computing resources of all server nodes in time slot τ; This represents the remaining storage resources of all server nodes in time slot τ; This represents the remaining bandwidth resources of all server nodes in time slot τ; This represents the remaining latency space of the current service function chain, where t r,τ A represents the total response delay in the service function chain r within time slot τ; r The attribute information of the service function chain r includes the ordered set of server nodes, bandwidth requirements, tolerable latency, and time slots of the service function chain r; pre represents the mapping predecessor node;

[0052] The action space is:

[0053] in, This indicates whether the server node v is mapped to a virtual network function in the service function chain r.

[0054] Optionally, in the feasible node, with the goal of maximizing the reward function, the joint optimization objective model is solved using the Markov decision process model to obtain the solution result, and a mapping strategy is determined based on the solution result and the reward function, including:

[0055] In the feasible node, with the goal of maximizing the reward function, the joint optimization objective model is solved using the Markov decision process model based on the deep reinforcement learning method to obtain the solution result; wherein, the solution result represents the real-time reward;

[0056] Based on the solution results, the first number and the second number are calculated based on the reward function, and a mapping strategy is obtained based on the first number and the second number.

[0057] In a second aspect, this application provides an optimized mapping apparatus for service function chains in a converged network architecture, the converged network architecture including a plurality of server nodes;

[0058] The device includes:

[0059] The acquisition module is configured to: acquire service function chain requests and resource information of each server node; wherein, the service function chain request includes the requirement information of several virtual network functions in the service function chain;

[0060] The construction module is configured to: construct a Markov decision process model based on the resource information and the demand information; wherein the Markov decision process model represents the mapping relationship between the server node and the virtual network function;

[0061] The solution module is configured to: solve the Markov decision process model to obtain the mapping strategy, and execute the mapping strategy.

[0062] In a third aspect, this application provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable by the processor, characterized in that the processor executes the computer program to implement the method described in the first aspect.

[0063] In a fourth aspect, this application provides a non-transitory computer-readable storage medium storing computer instructions, characterized in that the computer instructions are used to cause a computer to perform the method described in the first aspect.

[0064] A fifth aspect of this application provides a computer program product comprising computer program instructions that, when executed on a computer, cause the computer to perform the method described in the first aspect.

[0065] As can be seen from the above, the optimized mapping method and related equipment for service function chains in the converged network architecture provided in this application model the mapping optimization problem between server nodes and virtual network functions of service function chains in the converged network architecture as a Markov decision process model. By solving the Markov decision process model, a mapping strategy is obtained and executed, thereby forming a complete service function chain. This enables more flexible coordinated scheduling of network resources, reduces network resource congestion, and meets the power service requirements such as low latency, high reliability, and high bandwidth. Attached Figure Description

[0066] To more clearly illustrate the technical solutions in this application or related technologies, the drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0067] Figure 1 This is a flowchart illustrating the optimized mapping method for the service function chain in the converged network architecture according to an embodiment of this application.

[0068] Figure 2 This is a schematic diagram illustrating an exemplary converged network architecture and its corresponding VNF mapping scheme for power service requests, as exemplified in this application.

[0069] Figure 3 This is a comparative diagram showing the average latency consumed by the exemplary PPO algorithm, random algorithm, and mapping strategy of this embodiment for mapping service function chains in this application.

[0070] Figure 4 This is a comparative diagram showing the average cost consumed by the exemplary PPO algorithm, random algorithm, and mapping strategy of this embodiment for mapping the service function chain, which are embodiments of this application.

[0071] Figure 5 This is a schematic diagram of the structure of the optimized mapping device for the service function chain in the converged network architecture of this application embodiment;

[0072] Figure 6 This is a schematic diagram of the structure of an electronic device according to an embodiment of this application. Detailed Implementation

[0073] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0074] It should be noted that, unless otherwise defined, the technical or scientific terms used in the embodiments of this application should have the ordinary meaning understood by one of ordinary skill in the art to which this application pertains. The terms "first," "second," and similar terms used in the embodiments of this application do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed after the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are only used to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0075] The accelerated construction of new power systems has led to a significant increase in the amount of business information to be processed. Furthermore, the increasing number of communication methods has resulted in the connection of diverse terminal devices to the power grid, further increasing the demand for communication channels and placing more stringent requirements on the coverage, reliability, access flexibility, and network performance of communication networks. The communication needs of power business terminals also vary depending on factors such as business type and environmental conditions.

[0076] While Network Function Virtualization (NFV) technology has been proposed in related technologies—that is, transferring hardware devices to virtual machines to improve service flexibility and enhance network openness—Service Function Chains (SFCs) in NFV consist of ordered Virtual Network Functions (VNFs). VNFs provide flexibility by enabling the dynamic deployment and interconnection of network functions to achieve SFCs. The delivery of power grid services often requires various network service functions that support NFV, such as virtual device drivers, data aggregators, data compressors, or feature extractors.

[0077] Effective SFC mapping can flexibly handle massive data streams and perform filtering, learning, utilization, compression, and processing, providing efficient, scalable, and economical network services for end-user IoT applications. However, under the requirements of low latency and high reliability in power network services, the failure of any VNF will lead to the interruption of the SFC link, thereby causing the collapse of a certain network service. Therefore, NFV-based networks have higher availability requirements than traditional networks.

[0078] In view of this, embodiments of this application provide an optimized mapping method and related equipment for service function chains in a converged network architecture. The method models the mapping optimization problem between server nodes and virtual network functions of service function chains in a converged network architecture as a Markov decision process model. By solving the Markov decision process model, a mapping strategy is obtained and executed, thereby enabling more flexible coordinated scheduling of network resources, reducing network resource congestion, and meeting the power service requirements such as low latency, high reliability, and high bandwidth.

[0079] It should be noted that, in this embodiment, the converged network is a fusion of a 5G-based power line converged access network and a 5G-MEC (Multi-access Edge Computing) network, which can bring computing power to the edge of the mobile network, thereby reducing latency and lowering the load on the 5G-MEC layer network. That is, the converged network in this embodiment includes a 5G-based power line converged access layer and a 5G-MEC layer.

[0080] The core of the converged access layer is a power converged access network composed of multi-link converged terminals. Each converged terminal can access the upper-layer transmission network via direct connection or through the last-hop node in a multi-hop ad hoc network (such as IAB, mesh, etc.). Power service terminals generate functional service chain requests within a specific occupied space, and these requests are transmitted to the converged terminals in the access network via radio interfaces. Adjacent converged terminal server nodes in the subnet then form a cluster to process the requests. In each subnet of this layer, only the subnet cluster head can connect to the 5G-MEC layer.

[0081] In the 5G-MEC layer, different MEC nodes form a communication, storage, and computing network. The converged access layer accesses this network through 5G base stations. This MEC network is a wide-area resource collaboration platform. It can not only call upon the nearest MEC node, but also other MEC nodes with idle resources. When each MEC node has sufficient resources, it can take over one or more subnets of the converged access layer's access network, thereby forming a connection point between the two networks. This architecture uses several connection points to connect the two networks and realize data transmission, and the converged terminal at the subnet cluster head can directly access the 5G-MEC layer.

[0082] It should be noted that, in this embodiment of the application, since the data flow of each SFC is only forwarded and processed by its VNF, the NFV technology can effectively ensure the security of power communication service data in the scenario of the convergence of the two networks.

[0083] Figure 1 A flowchart illustrating an optimized mapping method 100 for service function chains in a converged network architecture according to an embodiment of this application is shown. Figure 1 As shown, the method 100 may include the following steps.

[0084] Step S101: Obtain the service function chain request and the resource information of each server node; wherein, the service function chain request includes the requirement information of several virtual network functions in the service function chain.

[0085] First, the converged network architecture of this embodiment is specifically defined. The physical resources of the converged network are abstracted as an undirected weighted graph G = (V, E, A) V A E Where V represents the set of server nodes, and The number of subnet cluster heads in the converged access layer or server nodes in the 5G-MEC layer is m; E represents the set of physical links, and e(v,v) ′ Each physical link (E) consists of interconnected server nodes. Indicates server node attributes. This represents the total computing resources of server node v. A represents the total storage resources of server node v; E ={W v ,T v,u} represents the physical link attribute, W v T represents the total bandwidth resources of the server nodes in each physical link (the total bandwidth capacity of each physical link is the sum of the total bandwidth resources of all server nodes in that physical link). v,u The communication forwarding delay of the physical link (the sum of the communication delays between server nodes u and v in the physical link).

[0086] In other words, in this embodiment, for each service function chain request, its virtual network function request needs to be mapped to the underlying physical network. That is, several virtual nodes of the service function chain need to be mapped to several underlying physical nodes (server nodes of the converged terminal). The physical nodes provide the necessary computing resources, storage resources, etc., as well as the functions corresponding to these resources. Furthermore, the physical link allocates an appropriate amount of bandwidth to transmit service flows between virtual nodes. In this way, the virtual link is mapped to the physical link, enabling communication between the service function chain and the service terminal or between the virtual nodes.

[0087] In this embodiment, the resource information of the server node includes the total computing resources, total storage resources, and total bandwidth resources of the server node; the requirement information of the virtual network function includes the computing resource requirements, storage resource requirements, and bandwidth resource requirements of the virtual network function.

[0088] Secondly, define the virtual network function. The virtual network function service instance uses the variable f. i ∈F={f1,f2,...,f i} represents its attributes. in, This indicates the computing resource requirements of virtual network functions; Indicates the storage resource requirements of virtual network functionality; t f This indicates the processing latency of the virtual network function. It is understood that in the converged network architecture of this embodiment, multiple virtual network function instances can share the same server node for service processing until the remaining available resources of that server node are insufficient to deploy more virtual network function instances.

[0089] Furthermore, we define a service function chain. A service function chain is represented as r∈R={r1,r2,...,r...} |R|}, whose attributes are 4-tuples. in, Represents an ordered set of server nodes in the service function chain r; This represents the bandwidth requirement of service function chain r, that is, the sum of the bandwidth requirements of all virtual network functions in service function chain r; T r This indicates the tolerable latency for the service function chain. It can be understood that for a service function chain request, if the cumulative latency of its business flow exceeds this tolerable latency, it is considered an SLA violation, and the service function chain request will be rejected.

[0090] Step S102: Construct a Markov decision process model based on the resource information and the demand information; wherein the Markov decision process model represents the mapping relationship between the server node and the virtual network function.

[0091] Figure 2 An exemplary converged network architecture and its corresponding VNF mapping scheme for power service requests are illustrated. For example... Figure 2 As shown, there are 3 server nodes. The remaining resources of server node 1 and server node 2 can meet the deployment resource requirements of VNF1 and VNF2 respectively, while server node 3 is idle. NFV-MANO is the VNF node deployment orchestrator of the SFC chain, which is responsible for collecting requests and processing them in order.

[0092] When deploying SFC requests, there are two options: Option 1, VNF1 is deployed on server node 1 and VNF2 is deployed on server node 2; Option 2, VNF1 and VNF2 are integrated and deployed on server node 3. It can be understood that Option 1 keeps server node 3 idle while deploying VNFs, thus reducing system operating costs, but increasing system bandwidth usage and request communication latency; Option 2 improves the QoS (Quality of Service) of requests by integrating VNFs, maximizing the total throughput of accepted requests, but increasing the deployment cost of network services.

[0093] As the examples above illustrate, optimizing the mapping between server nodes and virtual network functions in a converged network architecture requires considering the trade-off between network service quality and deployment cost. However, current mapping schemes in related technologies do not balance both factors.

[0094] Related technologies have proposed a service function chain deployment method, but for the network service deployment cost, only the computing resources consumed by the virtual network functions are considered. Neither the storage resources consumed by the virtual network functions are taken into account in the deployment cost, nor are the network service deployment cost and latency simultaneously optimized as joint objectives.

[0095] In this embodiment, maximizing network throughput and minimizing link mapping cost are taken as joint optimization objectives. A joint optimization objective model is constructed, and a reward function is further constructed based on the joint optimization objective model, thereby constructing a Markov decision process model.

[0096] In practical implementation, based on the orchestration architecture of 5G-based power converged access network and 5G-MEC, constraints corresponding to resource information are established, and a joint optimization objective model is constructed based on these constraints. Specifically, based on the total computing resources, storage resources, and bandwidth resources of the server nodes, constraints and joint optimization objectives for the function request chain mapping are established, thereby constructing the joint optimization objective model.

[0097] Specifically, the constraints of this joint optimization objective model include latency constraints, remaining computing resource constraints, remaining storage resource constraints, and remaining bandwidth resource constraints, as detailed below.

[0098] For a service function chain request, if the cumulative latency of its service flow exceeds its tolerable latency, it is considered an SLA violation, and the service function chain request will be rejected. That is, for a service function chain request to function normally, all virtual network functions of the service function chain request r∈R must be successfully mapped, and the actual end-to-end total response latency cannot exceed its maximum response latency limit (tolerable latency). Therefore, latency constraints are established.

[0099] The time delay constraint is:

[0100]

[0101] Among them, T r Indicates the tolerable latency of the service function chain; t r This indicates the total response delay.

[0102] The total response delay represents the sum of the communication delay of the physical link consisting of server nodes mapped by the service function chain and the processing delay of the virtual network functions on all server nodes of the physical link.

[0103] The total response delay can be expressed as:

[0104]

[0105] Where u and v represent server nodes, respectively; It is a binary variable, that is f represents a virtual network function instance in a service function chain r∈R. i Whether it is mapped to server node v. Correspondingly, f represents a virtual network function instance in a service function chain r∈R. i Whether it is mapped to server node u.

[0106] In addition, with This represents the number of virtual network function instances belonging to the service function chain r∈R mapped on server node v in time slot τ. Then:

[0107] The remaining computing resource constraints are as follows:

[0108]

[0109] in, This represents the remaining computing resources of server node v in time slot τ; F represents the set of virtual network functions. This indicates the number of virtual network functions f mapped on server node v in time slot τ; This indicates the computing resource requirements of virtual network functions.

[0110] The constraints on the remaining storage resources are as follows:

[0111]

[0112] in, This represents the remaining storage resources of server node v in time slot τ; This indicates the storage resource requirements of the virtual network function.

[0113] The remaining bandwidth resource constraints are as follows:

[0114]

[0115] in, W represents the remaining bandwidth resources of the server node in time slot τ; v This represents the total bandwidth resources of the server node; This indicates the bandwidth requirements of the service function chain r; Indicates whether a virtual network function f is mapped on server node v in time slot τ; y r It is a binary variable, i.e., y r ∈{0,1} indicates whether the service function chain r has been successfully mapped.

[0116] It can be understood that the condition for successful mapping of service function chain r is that all virtual network functions of service function chain r are successfully deployed, i.e. And it satisfies the time delay constraint. At this time, y r =1, otherwise 0.

[0117] Thus, this embodiment considers constraints from four aspects: latency limitations, computing resources, storage resources, and bandwidth resources. Based on these constraints, the joint optimization objective function of this joint optimization objective model is:

[0118]

[0119] Where τ represents a time slot; v∈V,V * This represents the set of server nodes based on the 5G-based power converged access layer and the 5G-MEC multi-access edge computing layer; This represents the total computing resources of server node v; W represents the total storage resources of server node v; v ξ represents the total bandwidth resources of server node v; cpu This represents the computational cost per unit of server resources; ξ mem This represents the processing cost per unit of server resources; ξ W Represents the unit cost of bandwidth consumption; r∈R represents the set of service function chains; y r ∈{0,1} indicates whether the service function chain r was successfully mapped; W r This represents the bandwidth requirement of service function chain r; t r This represents the total response delay. It can be understood that for ξ... cpu ξ mem and ξ W The sum of these three factors is determined by the network functions virtualization market and network functions virtualization service providers, and the total sum of these three factors is 1.

[0120] in addition, It is a binary variable, that is This indicates whether a virtual network function f is mapped on the server node v in time slot τ.

[0121] Furthermore, with This represents the number of virtual network function service instances mapped on server node v in time slot τ. Then: when hour and,

[0122] This minimizes the deployment costs of network services, including the total cost of deploying virtual network functions and the cost of mapping virtual links to service function chains. Simultaneously, it maximizes the total throughput of accepted service function chain requests, thereby maximizing link throughput within the network within a limited latency range.

[0123] In some embodiments, a reward function is further constructed based on the information of resources of each node in the fusion network architecture obtained through statistics, using the joint optimization target model.

[0124] Specifically, the first number of variables Ω r,s1 Second number Ω r,s2 Represented as a joint optimization objective model The evaluation index, that is, the evaluation index of the joint optimization objective model. The variable is transformed into Ω in the reward function formula. r,s1 ,Ω r,s2 These two parameters are used for trade-offs and optimization.

[0125] The specific explanation is as follows: For latency-tolerant tasks, fewer Ω values ​​are needed when selecting VNF deployment nodes. r,s1 And more Ω r,s2 This can indirectly alleviate network congestion by allocating idle resources in the converged access layer (where propagation latency is lower) to latency-sensitive tasks. Furthermore, for latency-sensitive tasks, more Ω values ​​are considered when selecting VNF deployment nodes. r,s1 and fewer Ω r,s2 While ensuring latency, the server nodes of the 5G-MEC layer can be allocated to latency-tolerant tasks with high computing and storage resource requirements.

[0126] Wherein, the first number represents the number of server nodes where the successfully mapped virtual network function resides in the 5G-based power converged access layer; the second number represents the number of server nodes where the successfully mapped virtual network function resides in the 5G-MEC multi-access edge computing layer. Specifically, when y r When = 1, the first number Second number

[0127] It should be noted that the power service types requested by the service function chain include latency-sensitive services and latency-tolerant services. However, the mapping schemes in related technologies do not differentiate between the type characteristics of power service scenarios, and cannot take into account both latency-sensitive and latency-tolerant services. Therefore, this can easily lead to situations where the latency cannot be exceeded within the tolerance range of the task, or even network server node congestion.

[0128] Therefore, this embodiment designs priority strategies for latency-sensitive and latency-tolerant power services respectively, so as to achieve efficient mapping of service function request chains for different power services.

[0129] In response to determining that the power service type of the service function chain request is a latency-sensitive service, the priority strategy is as follows: from mapping the predecessor node to each neighbor node in the subnet of the 5G-based power converged access layer; from mapping the predecessor node to the cluster head node of the subnet of the 5G-based power converged access layer; from mapping the predecessor node to each node in the 5G-MEC multi-access edge computing layer.

[0130] In response to determining that the power service type of the service function chain request is a latency-tolerant service, the priority strategy is as follows: from mapping the predecessor node to the cluster head node of the subnet based on the 5G power converged access layer; from mapping the predecessor node to each node in the 5G-MEC multi-access edge computing layer.

[0131] The predecessor node represents the previous node that was successfully mapped when mapping the current node.

[0132] Furthermore, in some embodiments, the reward function is determined based on the type of power service requested by the service function chain.

[0133] The reward function is:

[0134]

[0135] Wherein, R1 represents the reward function for a successful mapping of a power service type requested by the service function chain as a latency-sensitive service; R′1 represents the reward function for a failed mapping of a power service type requested by the service function chain as a latency-sensitive service; R2 represents the reward function for a successful mapping of a power service type requested by the service function chain as a latency-tolerant service; R′2 represents the reward function for a failed mapping of a power service type requested by the service function chain as a latency-tolerant service; κ1 represents the first weight coefficient; κ2 represents the second weight coefficient; κ3 represents the third weight coefficient; Ω r,s1 Indicates the first number; indicates Ω r,s2 Second number; sen r∈{0,1}, when the power service type requested by the service function chain is a time-sensitive service, sen r =1, when the power service type requested by the service function chain is a latency-tolerant service. r =0.

[0136] In other words, different services correspond to different reward functions. Since deploying virtual network functions at network edge nodes incurs lower transmission latency and deployment costs, nodes in the converged access layer are more suitable for latency-sensitive tasks. Therefore, for latency-sensitive tasks, Ω... r,s1 The larger the value, the more nodes that prioritize deploying virtual network functions at the converged access layer, the greater the reward; if it is a latency-tolerant task, then Ω... r,s2 The larger the number of nodes that prioritize deploying virtual network functions at the 5G-MEC layer, the greater the reward. This reduces latency and deployment costs, effectively alleviates node congestion and latency degradation at the 5G-MEC layer, and thus makes efficient use of network resources.

[0137] In some embodiments, based on the constructed converged network architecture, constraints, and joint optimization objective, the mapping optimization problem of virtual network functions between server nodes and service function chains in the converged network architecture is transformed into an NP-hard Markov decision process (MDP).

[0138] In practice, feasible nodes are selected in the converged network architecture by prioritizing the power service type requested by the service function chain and establishing constraints. Feasible nodes can be searched within the converged network architecture based on the priority strategy to obtain feasible nodes; wherein, the feasible nodes satisfy the latency constraints, remaining computing resource constraints, remaining storage resource constraints, and remaining bandwidth resource constraints.

[0139] Then, for each feasible node, the corresponding state and action are determined; the set of states of all feasible nodes is taken as the state space, and the set of actions is taken as the action space; thus, a Markov decision process model is constructed based on the state space and the action space.

[0140] For the state space: For each state

[0141] in, This represents the remaining computing resources of all server nodes in time slot τ; This represents the remaining storage resources of all server nodes in time slot τ; This represents the remaining bandwidth resources of all server nodes in time slot τ; This represents the remaining latency space of the current service function chain, where tr,τ A represents the total response delay in the service function chain r within time slot τ; r The attribute information of the service function chain r includes the ordered set of server nodes, bandwidth requirements, tolerable latency, and time slots of the service function chain r; pre represents the mapping predecessor node.

[0142] For the action space, the fusion network architecture contains |V| = K + N nodes, and is based on a t,v A tuple representing the action space. If This indicates that the deployment of the virtual network function has failed, then a t,v =0; if Then a t,v Let v be the index number of the server nodes. Therefore, the action space is represented as a. t ={a t,1 ,a t,2 ,…,a t,|V|}

[0143] and,

[0144] Step S103: Solve the Markov decision process model to obtain the mapping strategy, and execute the mapping strategy.

[0145] This embodiment considers deployment costs and latency. During the adaptive learning process of how the agent selects the node mapping for each virtual network function in the service function chain request, it can automatically take appropriate actions in each state by continuously learning policy π, thereby maximizing the reward function. To enable the agent to search for feasible nodes within the current range according to a priority policy, and to sort the feasible nodes in the set according to the latency required to deploy and run the virtual network function from smallest to largest, thereby mapping them sequentially at the corresponding nodes, a Markov decision process model is solved. It should be understood that the feasible nodes satisfy the latency constraints, remaining computing resource constraints, remaining storage resource constraints, and remaining bandwidth resource constraints.

[0146] In some alternative embodiments, a neural network model can be constructed using an A3C-based deep reinforcement learning approach to solve the MDP problem, thereby maximizing the reward function and achieving the joint optimization objective of minimizing task latency and deployment cost.

[0147] In some optional embodiments, at the feasible node, with the goal of maximizing the reward function, the joint optimization objective model is solved using the Markov decision process model to obtain the solution result, and a mapping strategy is determined based on the solution result and the reward function. It is understood that the solution result represents the real-time reward.

[0148] In this way, by using deep reinforcement learning methods to solve and obtain rewards in the feasible nodes, and then calculating the first number and the second number based on the rewards and the reward function, a mapping strategy is obtained.

[0149] The A3C algorithm consists of a global actor-critic network and multiple threaded sub-actor-critic networks. π(s|θ) and V(s|φ) represent the policy and value function of the global actor-critic, respectively. ′ ) and V(s|φ ′ ) represent the policy and value functions of the sub-actor-critic network, respectively, where θ and θ' are the policy and value functions of the sub-actor-critic network. ′ φ and φ ′ These are the parameters for actor-critic. The network can be updated every n steps or it can reach a certain termination condition.

[0150] Specifically, the value function gradient and policy gradient of each thread can be calculated using the following formulas, where H represents the entropy that can avoid convergence to a suboptimal deterministic policy, and δ is the entropy hyperparameter table that can control the strength of the entropy regularization term.

[0151]

[0152] Optionally, the designed adaptive online method for service function chain mapping based on A3C is as follows.

[0153]

[0154]

[0155] As can be seen from the algorithm above, it mainly includes two processes: algorithm initialization and embedding SFC's A3C. Furthermore, three variables are designed into the algorithm: the deployment of the predecessor node, the cumulative latency, and the cumulative deployment cost. Thus, during the deep reinforcement learning process, the agent iteratively optimizes policy π according to the reward function, ultimately enabling the agent to output actions with higher reward functions after learning, thereby achieving the joint optimization objective of minimizing task latency and deployment cost.

[0156] The specific explanation of the above algorithm is as follows: When a service function chain request generated by a business terminal is forwarded to the corresponding converged terminal in the converged access network, the source node where the terminal is located directly deploys the virtual network function of the service function chain request and points the previous node to the source node; then, the corresponding reward function is determined according to the business type, and the corresponding parameters of the actor-critic are initialized, and the learning agent learns online through the actor-critic and the asynchronous learner; next, the gradient update output is added separately to update the parameters of the global network, and these parameters can be copied to each thread; finally, the action is obtained when the stopping condition is reached.

[0157] This algorithm searches for feasible nodes within the current scope according to priority. Within the set of feasible nodes, it sorts them in ascending order of the latency required to deploy and run the virtual network function on each node. The algorithm then deploys the virtual network function sequentially on each node, pointing the previous node to the corresponding node. It updates the cumulative latency and deployment cost of the service function chain, as well as the available resources of the corresponding node. Finally, it outputs an optimized mapping strategy for the service function chain in the converged network architecture, specifically the number of server nodes where the successfully mapped virtual network functions reside in the 5G-based power converged access layer and the 5G-MEC multi-access edge computing layer, along with the specific mapping relationships.

[0158] Therefore, the mapping strategy based on this embodiment maps the service function chain. Specifically, according to the correspondence between the server nodes and the virtual network functions of the service function chain in the mapping strategy, the virtual network functions are deployed to the server nodes in sequence according to the priority strategy.

[0159] Thus, the proposed solution, based on the converged network architecture of 5G-based power access network and 5G-MEC, establishes a joint optimization objective of minimizing the total cost of functional service chain mapping and maximizing network throughput, considering constraints such as latency, computing resources, storage resources, and bandwidth resources. This balances the network service deployment cost and the latency consumed by each service function chain. Furthermore, it designs a priority strategy that differentiates the latency sensitivity of services, defines the state space, action space, and reward function, and designs an adaptive online algorithm for service function chain mapping based on A3C. Therefore, it meets the low-latency power service requirements, minimizes the cost of functional service chain mapping, enables more flexible collaborative scheduling of network resources, and reduces network resource congestion.

[0160] Because converged networks extend and extend edge computing capabilities, they place stricter demands on the coverage, reliability, access flexibility, and network performance of communication networks. The solution proposed in this application addresses the shortcomings of related technologies in meeting the diverse communication needs, service types, and resource requirements of multi-link power service terminals, as well as the inability to satisfy the network quality sensitivity of refined services. It can support low-latency, high-bandwidth, and highly reliable power communication services.

[0161] Finally, each of the above embodiments is further explained in detail with reference to exemplary simulation experiments.

[0162] Optionally, a Visual Studio Code simulation platform can be used, with Python 3.10 as the environment. The maximum number of service function chains is set to 250, with the ratio of latency-sensitive services to latency-tolerant services being 1:1.

[0163] refer to Figure 3 This is a comparative diagram showing the average latency consumed by the exemplary PPO algorithm, the random algorithm, and the mapping strategy of this embodiment for mapping service function chains. Figure 3 As shown, with the increase in the number of service function chain requests, the average latency consumed by the three algorithms in mapping the service function chain changes as follows: When the number of requests is 0-50, the available network resources are sufficient, and the differences between the three algorithms are small. However, relatively speaking, the mapping strategy in this embodiment consumes lower and more stable latency. As the number of requests increases, the PPO algorithm performs better than the random algorithm due to the expansion of training data and the iterative learning of the agent. In the range of 180-250, the mapping strategy in this embodiment performs best because it distinguishes between the priority decisions of latency-sensitive services and latency-tolerant services, ensuring the low latency requirements of power services while balancing the number of nodes deployed in the virtual network function of the service function chain at the 5G-MEC layer and the converged access layer, thus alleviating network congestion.

[0164] Figure 4 A comparative diagram illustrates the average cost consumed in mapping service function chains using the exemplary PPO algorithm, the random algorithm, and the mapping strategy of this embodiment. Figure 4 As shown, the increased number of requests expands the training data of the deep reinforcement learning algorithm, enhancing the learning effect of the agent. This results in the deployment cost of the agent's decision-making being lower than that of the random algorithm. Furthermore, the mapping strategy in this embodiment enables the agent to learn more efficiently, ultimately making its performance slightly better than the PPO algorithm.

[0165] Therefore, the solution proposed in this application ensures low latency while avoiding network server node congestion, thus meeting the high reliability and high bandwidth requirements of power services.

[0166] It should be noted that the above description describes some embodiments of this application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in a different order than that shown in the above embodiments and still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0167] Based on the same technical concept, corresponding to any of the above embodiments, this application also provides an optimized mapping device 500 for service function chains in a converged network architecture, wherein the converged network includes a plurality of server nodes.

[0168] refer to Figure 5 The optimized mapping device 500 for the service function chain in the converged network architecture includes:

[0169] The acquisition module 501 is configured to: acquire service function chain requests and resource information of each server node; wherein, the service function chain request includes the requirement information of several virtual network functions in the service function chain;

[0170] The construction module 502 is configured to: construct a Markov decision process model based on the resource information and the demand information; wherein the Markov decision process model represents the mapping relationship between the server node and the virtual network function;

[0171] The solver module 503 is configured to: solve the Markov decision process model to obtain a mapping strategy, and execute the mapping strategy.

[0172] In some alternative embodiments, the converged network includes a 5G-based power converged access layer and a 5G-MEC multi-access edge computing layer.

[0173] Optionally, the solution module 503 is specifically configured to: determine a priority strategy and reward function based on the power service type requested by the service function chain; establish constraints corresponding to the resource information and construct a joint optimization objective model based on the constraints; select feasible nodes in the fused network architecture based on the priority strategy and the constraints; among the feasible nodes, solve the joint optimization objective model using the Markov decision process model with the goal of maximizing the reward function, obtain the solution result, and determine a mapping strategy based on the solution result and the reward function.

[0174] For ease of description, the above devices are described in terms of function, divided into various modules. Of course, in implementing this application, the functions of each module can be implemented in one or more software and / or hardware.

[0175] The apparatus described above is used to implement the optimized mapping method of the service function chain in the corresponding converged network architecture in any of the foregoing embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0176] Based on the same technical concept, corresponding to any of the above embodiments, this application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable by the processor. When the processor executes the computer program, it implements the optimized mapping method of the service function chain in the converged network architecture as described in any of the above embodiments.

[0177] Figure 6 This embodiment illustrates a more specific hardware structure of an electronic device, which may include a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040, and a bus 1050. The processor 1010, memory 1020, input / output interface 1030, and communication interface 1040 are interconnected internally via the bus 1050.

[0178] The processor 1010 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.

[0179] The memory 1020 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 1020 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented by software or firmware, the relevant program code is stored in the memory 1020 and is called and executed by the processor 1010.

[0180] The input / output interface 1030 is used to connect input / output modules to realize information input and output. Input / output modules can be configured as components within the device (not shown in the figure) or externally connected to the device to provide corresponding functions. Input devices may include keyboards, mice, touchscreens, microphones, various sensors, etc., while output devices may include displays, speakers, vibrators, indicator lights, etc.

[0181] The communication interface 1040 is used to connect a communication module (not shown in the figure) to enable communication between this device and other devices. The communication module can communicate via wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0182] Bus 1050 includes a pathway for transmitting information between various components of the device, such as processor 1010, memory 1020, input / output interface 1030, and communication interface 1040.

[0183] It should be noted that although the above-described device only shows the processor 1010, memory 1020, input / output interface 1030, communication interface 1040, and bus 1050, in specific implementations, the device may also include other components necessary for normal operation. Furthermore, those skilled in the art will understand that the above-described device may only include the components necessary for implementing the embodiments of this specification, and not necessarily all the components shown in the figures.

[0184] The electronic devices described above are used to implement the optimized mapping method of the service function chain in the corresponding converged network architecture in any of the foregoing embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0185] Based on the same technical concept, corresponding to any of the above embodiments, this application also provides a non-transitory computer-readable storage medium storing computer instructions for causing a computer to execute the optimized mapping method of service function chains in the converged network architecture as described in any of the above embodiments.

[0186] The computer-readable medium of this embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device.

[0187] The computer instructions stored in the storage medium of the above embodiments are used to cause the computer to execute the optimized mapping method of the service function chain in the converged network architecture as described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0188] Based on the same technical concept, corresponding to any of the above embodiments, this application also provides a computer program product, which includes computer program instructions. In some embodiments, the computer program instructions can be executed by one or more processors of a computer to cause the computer and / or the processors to perform the optimized mapping method for service function chains in the converged network architecture. Corresponding to the execution entity for each step in each embodiment of the optimized mapping method for service function chains in the converged network architecture, the processor executing the corresponding step may belong to the corresponding execution entity.

[0189] The computer program product of the above embodiments is used to enable the computer and / or the processor to execute the optimized mapping method of the service function chain in the converged network architecture as described in any of the above embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0190] Those skilled in the art should understand that the discussion of any of the above embodiments is merely exemplary and is not intended to imply that the scope of this application (including the claims) is limited to these examples; within the framework of this application, the technical features of the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations of different aspects of the embodiments of this application as described above, which are not provided in the details for the sake of brevity.

[0191] Additionally, to simplify the description and discussion, and to avoid obscuring the embodiments of this application, the well-known power / ground connections to integrated circuit (IC) chips and other components may or may not be shown in the provided drawings. Furthermore, the apparatus may be shown in block diagram form to avoid obscuring the embodiments of this application, and this also takes into account the fact that the details of the implementation of these block diagram apparatuses are highly dependent on the platform on which the embodiments of this application will be implemented (i.e., these details should be fully understood by those skilled in the art). While specific details (e.g., circuits) have been set forth to describe exemplary embodiments of this application, it will be apparent to those skilled in the art that the embodiments of this application can be implemented without these specific details or with variations thereof. Therefore, these descriptions should be considered illustrative rather than restrictive.

[0192] Although this application has been described in conjunction with specific embodiments thereof, many substitutions, modifications, and variations of these embodiments will be apparent to those skilled in the art from the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may be used with the embodiments discussed.

[0193] The embodiments of this application are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the embodiments of this application should be included within the protection scope of this application.

Claims

1. An optimized mapping method for service function chains in a converged network architecture, characterized in that, The converged network architecture includes several server nodes, a 5G-based power converged access layer, and a 5G-MEC multi-access edge computing layer; The method includes: Obtain service function chain requests and resource information for each server node; wherein, the service function chain request includes the requirement information of several virtual network functions in the service function chain; Based on the resource information and the demand information, a Markov decision process model is constructed; wherein, the Markov decision process model represents the mapping relationship between the server node and the virtual network function; Based on the power service type requested by the service function chain, a priority strategy and reward function are determined; the power service type includes latency-sensitive services and latency-tolerant services; wherein, determining the priority strategy based on the power service type requested by the service function chain includes: in response to determining that the power service type requested by the service function chain is a latency-sensitive service, the priority strategy is: from mapping the predecessor node to each neighbor node in the subnet of the 5G-based power converged access layer; from mapping the predecessor node to the cluster head node of the subnet of the 5G-based power converged access layer; from mapping the predecessor node to each node in the 5G-MEC multi-access edge computing layer; in response to determining that the power service type requested by the service function chain is a latency-tolerant service, the priority strategy is: from mapping the predecessor node to the cluster head node of the subnet of the 5G-based power converged access layer; from mapping the predecessor node to each node in the 5G-MEC multi-access edge computing layer; the predecessor node represents the previous node that was successfully mapped when mapping the current node; Establish constraints corresponding to the resource information, and construct a joint optimization objective model based on the constraints; select feasible nodes in the fusion network architecture based on the priority strategy and the constraints; among the feasible nodes, solve the joint optimization objective model using the Markov decision process model with the goal of maximizing the reward function, obtain the solution result, and determine the mapping strategy based on the solution result and the reward function; Execute the mapping strategy.

2. The method according to claim 1, characterized in that, The resource information includes the total computing resources, total storage resources, and total bandwidth resources of the server node; the demand information includes the computing resource requirements, storage resource requirements, and bandwidth resource requirements of the virtual network function. The step of establishing constraints corresponding to the resource information and constructing a joint optimization objective model based on the constraints includes: Based on the total computing resources, total storage resources, and total bandwidth resources of the server nodes, establish latency constraints, remaining computing resource constraints, remaining storage resource constraints, and remaining bandwidth resource constraints. Based on the aforementioned latency constraints, remaining computing resource constraints, remaining storage resource constraints, and remaining bandwidth resource constraints, a joint optimization objective model is constructed. The time delay constraint is: ； in, Indicates the tolerable latency of the service function chain; It represents the total response latency, which is the sum of the communication latency of the physical link consisting of server nodes mapped by the service function chain and the processing latency of the virtual network functions on all server nodes of the physical link; Represents a set of service function chains; The remaining computing resource constraints are as follows: ； in, Indicates in time slot server node The remaining computing resources; Indicates server node Total computing resources; Represents a set of virtual network functions; Indicates in time slot server node The top mapping has virtual network functionality. Quantity; This indicates the computing resource requirements of virtual network functions; This represents the set of server nodes based on the 5G-based power converged access layer and the 5G-MEC multi-access edge computing layer; The constraints on the remaining storage resources are as follows: ； in, Indicates in time slot server node The remaining storage resources; Indicates server node The total amount of storage resources; This indicates the storage resource requirements of the virtual network function; The remaining bandwidth resource constraints are as follows: ； in, Indicates in time slot The remaining bandwidth resources of the server node; This represents the total bandwidth resources of the server node; Represents the service function chain Was the mapping successful? Represents the service function chain Bandwidth requirements; Indicates in time slot server node Does the above map to a virtual network? .

3. The method according to claim 2, characterized in that, The step of selecting feasible nodes in the converged network architecture based on the priority strategy and the constraints includes: Based on the priority strategy, feasible nodes are searched in the converged network architecture to obtain feasible nodes; wherein, the feasible nodes satisfy the latency constraints, remaining computing resource constraints, remaining storage resource constraints, and remaining bandwidth resource constraints.

4. The method according to claim 2, characterized in that, The joint optimization objective function of the joint optimization objective model is: in, This represents the computational cost per unit of server resources. This indicates the processing cost per unit of server resources. This represents the unit cost of bandwidth consumption.

5. The method according to claim 2, characterized in that, The reward function is: in, The reward function represents the type of power service requested by the service function chain as a time-sensitive service and the mapping is successful; The reward function represents the reward function for the power service type requested by the service function chain being a time-sensitive service and for which mapping has failed. The reward function represents the type of power service requested by the service function chain as a latency-tolerant service and indicates that the mapping is successful; The reward function represents the reward function for the power service type requested by the service function chain as a latency-tolerant service and for which the mapping fails. Indicates the first weighting coefficient; This represents the second weighting coefficient; This represents the third weighting coefficient; for Evaluation indicators The first number represents the number of server nodes where the successfully mapped virtual network function is located in the 5G-based power converged access layer. The second number represents the number of server nodes where the successfully mapped virtual network function is located in the 5G-MEC multi-access edge computing layer; This is determined based on the type of power service requested according to the service function chain.

6. The method according to claim 2, characterized in that, The construction of the Markov decision process model includes: For each of the feasible nodes, determine the corresponding state and action; The set of states of all feasible nodes is taken as the state space, and the set of actions is taken as the action space. Based on the state space and the action space, a Markov decision process model is constructed. The state space is as follows: for each state ; in, Indicates in time slot The remaining computing resources of all server nodes; Indicates in time slot Remaining storage resources on all server nodes; Indicates in time slot Remaining bandwidth resources of all server nodes; This represents the remaining latency space of the current service function chain, where Indicates in time slot Service Function Chain Total response latency; Represents the service function chain The attribute information, including the service function chain. The ordered set of server nodes, bandwidth requirements, tolerable latency, and time slots; This indicates that the previous node is being mapped. The action space is in, Indicates server node Does the above map to a service function chain? The virtual network function in A tuple representing the action space.

7. The method according to claim 5, characterized in that, In the feasible node, with the goal of maximizing the reward function, the joint optimization objective model is solved using the Markov decision process model to obtain the solution result. Based on the solution result and the reward function, a mapping strategy is determined, including: In the feasible node, with the goal of maximizing the reward function, the joint optimization objective model is solved using the Markov decision process model based on the deep reinforcement learning method to obtain the solution result; wherein, the solution result represents the real-time reward; Based on the solution results, the first number and the second number are calculated based on the reward function, and a mapping strategy is obtained based on the first number and the second number.

8. An optimized mapping device for service function chains in a converged network architecture, characterized in that, The converged network architecture includes several server nodes, a 5G-based power converged access layer, and a 5G-MEC multi-access edge computing layer; The device includes: The acquisition module is configured to: acquire service function chain requests and resource information of each server node; wherein, the service function chain request includes the requirement information of several virtual network functions in the service function chain; The construction module is configured to: construct a Markov decision process model based on the resource information and the demand information; wherein the Markov decision process model represents the mapping relationship between the server node and the virtual network function; The solution module is configured to: determine a priority strategy and reward function based on the power service type requested by the service function chain; the power service type includes latency-sensitive services and latency-tolerant services; wherein, determining the priority strategy based on the power service type requested by the service function chain includes: in response to determining that the power service type requested by the service function chain is a latency-sensitive service, the priority strategy is: from mapping the predecessor node to each neighbor node in the subnet of the 5G-based power converged access layer; from mapping the predecessor node to the cluster head node of the subnet of the 5G-based power converged access layer; from mapping the predecessor node to each node in the 5G-MEC multi-access edge computing layer; in response to determining that the power service type requested by the service function chain is a latency-tolerant service. The priority strategy is as follows: mapping the predecessor node to the cluster head node of the subnet of the 5G-based power converged access layer; mapping the predecessor node to each node in the 5G-MEC multi-access edge computing layer; the predecessor node represents the previous node that was successfully mapped when mapping the current node; establishing constraints corresponding to the resource information, and constructing a joint optimization objective model based on the constraints; selecting feasible nodes in the converged network architecture based on the priority strategy and the constraints; among the feasible nodes, solving the joint optimization objective model using the Markov decision process model with the goal of maximizing the reward function, obtaining the solution result, and determining the mapping strategy based on the solution result and the reward function; and executing the mapping strategy.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable by the processor, characterized in that, When the processor executes the computer program, it implements the method as described in any one of claims 1-7.

10. A non-transitory computer-readable storage medium storing computer instructions, characterized in that, The computer instructions are used to cause the computer to perform the method as described in any one of claims 1-7.

11. A computer program product comprising computer program instructions that, when executed on a computer, cause the computer to perform the method as described in any one of claims 1-7.

Images