CN115314355A - Electric power communication network architecture system and method based on deterministic network - Google Patents

Abstract

The invention proposes a power communication network architecture system based on a deterministic network. The system includes a control layer, an infrastructure layer and a knowledge layer; the control layer, for service and policy fitting, adopts a knowledge-based scheduling mechanism according to user business requirements, Form a definite routing and resource strategy that meets the requirements of specific business service quality and network behavior characteristics, and realize the decision-making adaptation of routing and scheduling; the infrastructure layer, strategy and resource matching, is composed of a series of TSN network devices, which are used for routing and scheduling. The scheduling function is mapped to a refined network resource combination to realize data forwarding; the knowledge layer, fitting resources and knowledge, communicates with the infrastructure layer and the control layer, and is responsible for collecting and storing the network resources of the infrastructure layer in real time. The resource allocation strategy of the control layer realizes the dynamic allocation strategy of network resources, and feeds back the learned scheduling strategy to the control layer in time.

Description

Power communication network architecture system and method based on deterministic network

technical field

The present invention relates to the technical field of network communication, in particular to a deterministic network-based power communication network architecture system and method.

Background technique

As an important support and guarantee of the power grid, the power communication network is an important basis for realizing the intelligence and interaction of the power grid and the operation control of the large power grid. With the continuous in-depth development and promotion of smart grid and ubiquitous electric power Internet of things, the emergence of new technologies such as 5G communication, ICT infrastructure and holographic communication, more and more new services are connected to the network, and the power communication business is moving towards video The development of large-bandwidth and low-latency services such as , multimedia, and precision loads poses new demands and challenges for power communication networks. Because the traditional power communication network architecture adopts the best-effort design mode, the exchange mode is single, the transmission delay of data packets in the network is difficult to determine, and the network controllability is weak. Therefore, the current power system urgently needs to introduce new technologies and architectures to provide deterministic and low-latency differentiated services.

Deterministic Network (Deterministic Network, DetNet) is a recently proposed network that guarantees deterministic bandwidth, delay, jitter and packet loss rate indicators. Its underlying layer uses Ethernet TSN technology to ensure real-time data transmission and delay determinism through methods such as resource reservation, display routing, and redundant transmission under the premise of ensuring accurate time synchronization. The deterministic network system model has the characteristics of "punctuality, reliability, and large scale". It is deeply integrated with the existing power communication network architecture and is considered to be a key technology platform for power grid perception, calculation, and analysis capabilities. It can match the power grid industry's core requirements for business controllability and differentiated processing, and provides a new technology choice for ubiquitous, flexible, efficient and stable end-to-end data transmission services for the power access network and core network, and improves the network architecture. support.

Therefore, starting from the basic concept of deterministic network, the present invention proposes a three-layer model of electric power communication network that integrates deterministic network technology. From the aspect of architectural function design, the role of each layer is elaborated. In order to obtain the real-time network status and make the optimal decision independently, the collaborative work of the management control module of the control layer and the management module of the knowledge layer is analyzed, and finally a real-time information feedback system of the deterministic network is realized.

Contents of the invention

Purpose of the invention: The technical problem to be solved by the present invention is how to effectively reduce the end-to-end delay of data transmission in the power communication network. The present invention proposes a deterministic network-based power communication network architecture and method. Due to the particularity of the deterministic network, it is very important to analyze its network performance in different application scenarios. The core idea of software-defined network (SDN) is to separate control and forwarding, to realize the programmability and ease of operation of management and control. It has been widely used in power communication network and smart grid. Therefore, based on the idea of separation of control and forwarding in SDN, a three-layer deterministic network architecture similar to SDN is constructed. Among them, the control layer is completely separated from the technical facility layer, while the knowledge layer can communicate with the two layers and provide timely feedback information.

Technical solution: In order to solve the above technical problems, the present invention proposes a power communication network architecture system based on a deterministic network, which includes a control layer, an infrastructure layer and a knowledge layer;

The control layer, service and policy fitting, adopts a knowledge-based scheduling mechanism based on user business needs to form a definite routing and resource strategy that meets the requirements of specific business service quality and network behavior characteristics, and realizes the decision-making adaptation of routing and scheduling;

The infrastructure layer, strategy and resource fitting, is composed of a series of TSN network devices, which are used to map routing and scheduling functions into refined network resource combinations to realize data forwarding;

The knowledge layer, resources and knowledge fitting, communicates with the infrastructure layer and the control layer, and is responsible for collecting and storing network resources of the infrastructure layer in real time. By learning the resource allocation strategy of the control layer, the dynamic allocation strategy of network resources is realized, and Feedback the learned scheduling strategy to the control layer in time.

Further, the control layer is responsible for implementing routing and scheduling functions, carrying services upwards, and controlling the data layer downwards, including a data analysis module, a resource management module, and a path calculation module, and adopts a knowledge-based scheduling mechanism. The method is as follows: data analysis The module conducts abstract modeling of user business requests to realize the mapping from business indicators to service models; the path calculation module calculates the deterministic path of the service based on the service model and routing algorithm; the resource management module calculates the service path and the feedback from the knowledge layer Information, using different scheduling algorithms to achieve resource allocation between nodes;

其中，v为业务等级，c^p为运营商收益，c^s为标准等级收益，δ为时延，λ为抖动，ε为分组丢失率；若用户请求的业务总数为m,则分析模块需要处理的基本参数及网络效益分别为(B₁,B₂,...,B_m)和(R₁,R₂,...,R_m)，综上所述，业务需求模型I表示为：Among them, the business requirements in the data analysis module include business basic parameters B and network benefits R, the basic parameters are defined as five-tuple B=(s,d,l,t _max ,r), s is the source node, d is the destination node , l is the data flow length, t _max is the maximum end-to-end delay, r is the allocation rate; the network benefit is defined as

Among them, v is the service level, c ^p is the operator's income, c ^s is the standard level income, δ is the delay, λ is the jitter, and ε is the packet loss rate; if the total number of services requested by the user is m, the analysis module needs to process The basic parameters and network benefits of are respectively (B ₁ ,B ₂ ,...,B _m ) and (R ₁ ,R ₂ ,...,R _m ). To sum up, the business demand model I is expressed as:

The service model S refers to the service performance index and the necessary service function requirements. The performance index is defined as Q=(t, δ, λ, ε), where t is the throughput rate, δ is the delay, λ is the jitter rate, and ε is the packet Loss rate; service function requirements F include: network function F ^N , function dependency relationship F ^Q and network function dependent resource type ^FR , that is, F=(F ^N , F ^Q , ^FR ); when the number of business requests is m, the service The model is expressed as:

Among them, Q _i and F _i respectively represent the performance index of the i-th service and the corresponding service function requirement, i∈(1,m).

映射得到服务路径，路由算法Rt指最短路径算法或K最短路径算法，

映射函数的功能是依据服务性能需求，为业务选取所有可能的传输路径，控制层将服务需求转化为服务路径的过程如下表示，即：The route addressing function in the path calculation module combines the service demand matrix S defined above with the routing algorithm through the mapping function

The service path is mapped, and the routing algorithm Rt refers to the shortest path algorithm or K shortest path algorithm,

The function of the mapping function is to select all possible transmission paths for the business according to the service performance requirements. The process of the control layer transforming the service requirements into service paths is as follows, namely:

After the service path is established, data transmission is carried out. If the application requirements cannot be met or the constraints of routing adjustment cannot be met, the path calculation will be re-executed;

When the service path meets the conditions, the resource management module uses the circular queue forwarding (CQF) mechanism or the scheduling algorithm based on the asynchronous shaping ATS to allocate resources reasonably according to the feedback information of the path P and the knowledge layer, so that the service end-to-end delay Minimize and maximize the utilization of network resources.

Further, the infrastructure layer is located at the bottom of the entire architecture, including deterministic forwarding devices and deterministic processing devices. The deterministic forwarding devices do not have routing functions and only forward data; the deterministic processing devices not only have data forwarding capabilities , also has the function of data processing through programming, specifically:

分别代表节点在计算、缓存及传输功能的第i类能力上的服务实例，i∈(1,j)。基于以上定义，对于服务路径P上得第k个服务节点所依赖的物理节点的资源实例化结果为：The functions of the infrastructure layer are realized by node state information, service capabilities and function instances, where N is defined as the set of all physical nodes n _i in the network, and i is the number of physical nodes; the state information of nodes is defined as F=(N _l , N _s , N _d , N _im ), the element components in the state information respectively represent the node position, node type, node connection degree and node importance degree; the service capability is defined as C=(C _c ,C _h ,C _t ), including Computing capability C _c , caching capability C _h , and transmission capability C _t , where each type of service capability is specifically divided into the following subscript j; function instance E=(E _c , E _h , E _t ) corresponds to the resource service capability of the node, where E _c is an instance of computing function, E _h is an instance of cache function, E _t is an instance of transmission function, that is, the service instance of node n on the jth capability of transmission resource instance E is defined as e, where

respectively represent the service instance of the node in the i-th capability of computing, caching and transmission functions, i∈(1,j). Based on the above definition, the resource instantiation result of the physical node on which the kth service node depends on the service path P is:

Therefore, the mapping from service path to data layer service instance resource combination is expressed as:

表示映射函数，其将路径P经过的节点，依据服务能力C和服务实例F转化为细化的资源实例。d_i表示第i个服务节点的资源实例化结果，i∈(1,k)。in,

Indicates the mapping function, which converts the nodes passed by the path P into refined resource instances according to the service capability C and service instance F. d _i represents the resource instantiation result of the i-th service node, i∈(1,k).

Further, the knowledge layer communicates with the infrastructure layer and the control layer through a programmable interface, including an enhanced knowledge management module and a state management information base; wherein, the state management information base stores service resource combination instances for real-time monitoring of equipment status. Update and enhance the knowledge management module to record the scheduling strategy of the control layer, combine the machine learning algorithm to learn the scheduling algorithm of different business requirements, and predict the optimal scheduling strategy of the data flow according to the learned knowledge. The knowledge is defined as K=(lh,br,D _max , D _T , C _l ), lh is the length of the data flow, the data burst rate br, the maximum end-to-end delay D _max , the flow cut-off time D _T and the link capability C _l .

Further, the control layer and the knowledge layer work together to implement a timely and accurate feedback mechanism, including: when the control layer schedules and allocates data streams, it first parses the data streams to obtain the performance requirements of the data streams, including The maximum delay that can be tolerated, the length of the data packet, and the data burst rate; then, according to the performance model of the analyzed data flow, the routing and service mapping are performed to obtain the deterministic service path; secondly, the control layer requests the knowledge layer to The existing scheduling knowledge includes synchronous scheduling or asynchronous scheduling, and selects the corresponding scheduling strategy according to the underlying network resource status recorded in the information status database; at the same time, the knowledge layer makes intelligent learning decisions based on the existing scheduling algorithms and traffic characteristics. The knowledge is used to predict the scheduling strategy of subsequent traffic. Finally, the routing and scheduling strategy is sent to the infrastructure layer for corresponding forwarding and processing.

Further, the knowledge management module and the device information status database of the knowledge layer collect and update the operating status information of the underlying forwarding equipment, and whenever the operating status of the underlying forwarding equipment changes, autonomously send status update information to the management module to which it belongs ;Enhance the knowledge management module to periodically send status maintenance information to the underlying forwarding device to prevent the failure to communicate with the knowledge layer autonomously when a fault occurs, and the control layer directly accesses the information status database when performing data flow scheduling and allocation to obtain real-time network device operation State information, so as to select the appropriate strategy for optimal allocation according to business needs; the enhanced management module mines and learns knowledge algorithms according to the needs of the control layer, and sends the corresponding knowledge to the control layer to realize intelligent management and decision-making.

The present invention also proposes a power communication network resource scheduling method based on the power communication network architecture system of the above-mentioned deterministic network, and the steps of the method are as follows:

Step 1, for the deterministic business of the power grid, the control layer obtains the performance requirements of the business through the data analysis module, and maps it to the service function model S;

Step 2, transmit the service function model to the routing control module, and combine the routing algorithm to realize the deterministic transmission path;

Step 3, the resource management module obtains knowledge related to resource policies from the knowledge layer, combines the transmission path P, calls the scheduling algorithm, and performs resource allocation;

Step 4: According to the device state information recorded in the state information database of the knowledge layer, if the combination of service resources meets the routing constraints and business performance requirements, scheduling has already occurred at this time, and the control layer allocates resources to the bottom layer to determine the path to complete the buffering of deterministic requirements mechanism, queue scheduling mechanism, and path selection mechanism, and distribute commands to the infrastructure layer; if the scheduling conditions are not met, re-schedule calculation and strategy learning;

Step 5: The infrastructure layer receives the scheduling command from the control layer and transmits the service data packets; when the data flow transmission is completed, the network status is updated and the status information is stored in the status information database of the knowledge layer.

Beneficial effects: Compared with the prior art, the present invention adopts the above technical solutions to have the following beneficial technical effects:

The new deterministic network-based power communication network architecture and method of the present invention adopts the SDN-based deterministic network architecture, integrates with the existing power network architecture, and realizes substation communication by setting a professional level and applying core technologies such as clock synchronization and display routing. Synchronous transmission between networks, low latency and high reliability of scheduling automation.

The new deterministic network-based power communication network architecture and method of the present invention, with deterministic transmission as the core, supports reasonable coexistence of multiple deterministic services and common services, provides efficient and convenient management through an efficient queue management mechanism and information feedback mechanism, and realizes Deterministic service guarantee, while adhering to the traditional advantages of optical transmission.

The novel deterministic network-based power communication network architecture and method of the present invention has the characteristics of low cost and multiplexing of Ethernet, and solves the defects of poor scalability of traditional power architecture and inability to effectively support time-sensitive business and sudden business. The deterministic network architecture is used in power integrated networking scenarios to provide resources and deterministic service guarantees for business decisions.

Description of drawings

Fig. 1 is a schematic diagram of a power communication network architecture based on a deterministic network according to an embodiment of the present invention;

Fig. 2 is a flow chart of a communication method for a power communication network based on a deterministic network according to an embodiment of the present invention;

3 is a schematic diagram of an information feedback mechanism based on a deterministic network according to an embodiment of the present invention;

Fig. 4 is a schematic diagram of management functions based on a deterministic network according to an embodiment of the present invention.

Detailed ways

Below in conjunction with accompanying drawing, technical scheme of the present invention is described in further detail:

In today's era of surge in data flow, the existing Internet often has problems such as data congestion, high packet loss rate, and unguaranteed delay. The deterministic service function provided by the deterministic network can guarantee the data transmission with special requirements to meet the requirements of extremely low delay, zero packet loss and high reliable service quality in any network environment. Based on the deterministic network architecture and integrated with the existing communication network architecture, it provides a solution for the data transmission of the power communication network.

The deterministic network-based power communication network architecture according to the embodiment of the present invention, as shown in FIG. 1 , includes: an infrastructure layer, a control layer and a knowledge layer.

Specifically, the infrastructure layer is composed of a series of TSN network devices, which communicate upward with the control layer for the definite forwarding of data. The control layer, composed of multiple controllers, logically forms a central controller, communicates with the infrastructure layer, and is responsible for controlling and managing the underlying devices. The knowledge layer communicates with the infrastructure layer and the control layer, and is responsible for collecting and storing network resources of the infrastructure layer in real time. By learning the resource allocation strategy of the control layer, the dynamic allocation strategy of network resources is realized, and timely feedback has been made to the control layer. Scheduling policy for learning.

Among them, the infrastructure layer and the control layer realize the separation of control and forwarding, and the knowledge layer communicates with the infrastructure layer and the control layer to form a dynamic and real-time feedback mechanism to determine the transmission path of the data flow and the reasonable allocation of resources.

将物理节点的资源进行实例化，得到服务实例资源组合，以供知识层访问和存储。According to some embodiments of the present invention, as shown in FIG. 1 , the infrastructure layer includes two types of deterministic network nodes, deterministic edge nodes and deterministic forwarding nodes. The deterministic edge node can be used as the starting point or the terminating point of the deterministic flow, and its main functions include adding or deleting data packet sorting, data packet duplication and combination. The deterministic forwarding node only implements the DetNet forwarding sublayer, which is responsible for routing the message from the source to the destination, and focuses on determining the data forwarding of the network. After receiving the control information of the control layer, it realizes the data forwarding of the business, and at the same time provides its own status information to the control layer to ensure the real-time update of the network topology by the control layer. It also provides node status and resource usage information to the control and knowledge planes, and updates its own running status information in real time, so that the control plane can better allocate resources and select routes. Specifically, the function of the infrastructure layer is to realize the refined resource service combination of routing and resource allocation strategies, and mainly use three attributes of node status information, service capabilities and function instances to instantiate nodes. Node state information is defined as a collection of node location, node type, node connectivity and node importance; service capabilities include computing capability C _c , cache capability C _h and transmission capability C _t ; resource service capabilities of nodes corresponding to functional instances; the above attributes by mapping function

The resources of the physical nodes are instantiated to obtain the combination of service instance resources for access and storage by the knowledge layer.

According to some embodiments of the present invention, as shown in FIG. 1 , the control layer is composed of multiple controllers, actually forming a centralized logic controller. Each controller is composed of multiple control modules, including resource management module, path calculation module, traffic scheduling module and data analysis module. The modules work together to control and manage the underlying equipment. By interacting with the knowledge plane, they can obtain timely Network status and call the corresponding knowledge algorithm.

Specifically, the data analysis module abstracts and models user business requests to realize the mapping from business indicators to service models; the path calculation module calculates the deterministic path of the service based on the service model and routing algorithms, such as the shortest path algorithm; at the same time, The resource management module implements resource allocation between nodes by using scheduling algorithms based on feedback information from service paths and knowledge layers.

The scheduling algorithm used in this paper can be divided into two types: synchronous scheduling and asynchronous scheduling. The cyclic forwarding queue based on the cyclic forwarding queue or the cycle-specified cyclic forwarding queue is a synchronous scheduling algorithm. The network adopts the clock synchronization protocol and frame preemption mechanism of the TSN time-sensitive network to realize the fast transmission of delay-sensitive flows and ensure an extremely low packet loss rate. The asynchronous traffic shaper ATS used in the asynchronous scheduling algorithm is a scheduler based on the urgency level (UBS) to achieve deterministic transmission of each flow. It is characterized by no need for strict clock synchronization, strong scalability, and full use of network bandwidth.

According to some embodiments of the present invention, as shown in FIG. 1 , the knowledge layer consists of an enhanced knowledge management module and a device information status database, responsible for collecting and storing network status information, such as the status of flows passing through each network node. At the same time, according to the requirements of the control plane, the existing scheduling algorithms are mined and learned, and the network status information and intelligent learning algorithms can be sent to the control plane according to the requirements of the control plane.

The enhanced knowledge management module is composed of multiple devices, and each device manages an area. It is mainly responsible for communicating with the forwarding nodes of the infrastructure layer, collecting and updating the operating status information of the underlying forwarding devices, such as link transmission delay and remaining buffer space , available bandwidth, forwarding delay of end devices, etc. And summarize the collected information into a device information state database. Whenever the operating status of the underlying forwarding device changes, it will autonomously send status update information to the management module it belongs to. In addition, the enhanced knowledge management module will periodically send status maintenance information to the underlying forwarding device to prevent failure to autonomously communicate with the knowledge layer when a failure occurs.

According to the deterministic network-based power communication network networking method according to the embodiment of the present invention, the method adopts the above-mentioned deterministic network architecture-based power communication network architecture, and the flow chart of the business scheduling method is shown in Figure 2, including:

(1) When the user request arrives, the control layer parses the data packet through the data analysis model, thereby obtaining the service function requirements of the business, and transforming the service into a service function model;

(2) The service function model is transmitted to the routing control module, combined with the routing algorithm, to realize the deterministic transmission path P;

(3) The resource management module obtains resource policy-related knowledge from the knowledge layer, combines the transmission path P, calls the scheduling algorithm, and judges the allocation of resources;

(4) According to the device state information recorded in the state information database of the knowledge layer, if the combination of service resources meets the routing constraints and business performance requirements, the scheduling has already occurred at this time, and the control layer allocates resources to the bottom layer to determine the path to complete the buffering of deterministic requirements mechanism, queue scheduling mechanism, and path selection mechanism, and distribute commands to the infrastructure layer; if the scheduling conditions are not met, the scheduling calculation and policy learning will be re-performed.

(5) The infrastructure layer receives the scheduling command from the control layer and transmits the business data packets; when the data flow transmission is completed, the network status is updated and the status information is stored in the status information database of the knowledge layer.

A schematic diagram of layers among modules of the present invention is shown in FIG. 1 . System function modules include: SDN-based deterministic network architecture, deterministic network information feedback mechanism. Deterministic networks require limited delay upper limit, jitter, and extremely low packet loss rate, which not only puts higher requirements on the forwarding capability of the forwarding layer, but also puts forward stricter control capabilities for upper-layer management. In order to achieve timely and accurate information feedback, the knowledge management module and the equipment information status database are enhanced in the knowledge layer, which are coordinated with the resource management module of the control layer to form a closed-loop information feedback mechanism.

Realization of power communication architecture function of deterministic network:

The network nodes adopt deterministic delay guarantee. The power core network architecture mainly includes the knowledge layer, control layer and infrastructure layer. Deterministic network architecture and protocols are deployed at each layer, and the deterministic power architecture can be completed through the collaborative processing of each layer. Scalability and collaborative business capabilities, a schematic diagram of the deterministic power communication network architecture is shown in Figure 1.

The realization of the feedback function of the power communication architecture of the deterministic network is shown in Figure 3:

The information feedback function is based on the enhanced management module and the deterministic routing protocol to realize the information interaction between the knowledge layer and the control layer; based on the information status device library, it realizes communication with the infrastructure layer and obtains the changing network status in time. Whenever the operating status of the underlying forwarding device changes, it will autonomously send status update information to the management module it belongs to. In addition, the enhanced knowledge management module will periodically send status maintenance information to the underlying forwarding device to prevent failure to autonomously communicate with the knowledge layer when a failure occurs. When the control layer performs data flow scheduling and allocation, it first analyzes the data flow to obtain the performance requirements of the data flow, such as the maximum tolerable delay, the length of the data packet, and the data burst rate; then, according to the analyzed data Flow performance model, mapping routes and services, to obtain deterministic service paths; secondly, the control layer requests existing scheduling knowledge from the knowledge layer, such as synchronous scheduling or asynchronous scheduling, and according to the status of the underlying network resources recorded in the information status database , to select the corresponding scheduling strategy; at the same time, the knowledge layer will make intelligent learning decisions based on the existing scheduling algorithms and traffic characteristics, and the learned knowledge can be used to predict the scheduling strategy of subsequent traffic. Finally, the routing and scheduling policies are delivered to the infrastructure layer for corresponding forwarding and processing.

The realization of the power communication architecture management function of the deterministic network is shown in Figure 4:

The control layer is mainly responsible for the resource and computing management of the entire architecture, mainly including the resource management module, path calculation module, traffic scheduling module and data analysis module. The modules work together to control and manage the underlying devices. By interacting with the knowledge plane, they can Obtain the network status in time and call the corresponding scheduling algorithm.

To sum up, on the basis of the SDN-based deterministic network architecture, the present invention provides a deterministic network-based fusion architecture and method for the power communication network system, supports large-scale deterministic service services, and provides a flexible and scalable architecture , to achieve low latency, high-speed mobility, high reliability, low packet loss and other functions of power communication.

Claims (7)

1. A power communication network architecture system based on a deterministic network is characterized by comprising a control layer, an infrastructure layer and a knowledge layer;

a control layer, which is used for fitting services and strategies, and forming a determined route and resource strategy meeting the requirements of specific service quality and network behavior characteristics by adopting a knowledge-based scheduling mechanism aiming at the service requirements of users, so as to realize the decision adaptation of the route and the scheduling;

the infrastructure layer is used for matching strategies and resources, consists of a series of TSN network devices, and is used for mapping the routing and scheduling functions into refined network resource combinations to realize data forwarding;

and the knowledge layer is used for fitting resources with knowledge, is in communication connection with the infrastructure layer and the control layer, is responsible for collecting and storing the network resources of the infrastructure layer in real time, realizes a dynamic allocation strategy of the network resources by learning a resource allocation strategy of the control layer, and timely feeds back the learned scheduling strategy to the control layer.

2. The deterministic network-based power communication network architecture system of claim 1, wherein the control layer is responsible for implementing routing and scheduling functions, upward carries traffic, and downward controls the data layer, and comprises a data analysis module, a resource management module and a path computation module, and adopts a knowledge-based scheduling mechanism, and the method comprises the following steps: the data analysis module carries out abstract modeling on the user service request to realize the mapping from the service index to the service model; the path calculation module calculates a deterministic path of the service according to the service model and the routing algorithm; the resource management module adopts different scheduling algorithms according to the feedback information of the service type path and the knowledge layer to realize the resource allocation among the nodes;

the service requirement in the data analysis module includes a service basic parameter B and a network benefit R, and the basic parameter is defined as a quintuple B = (s, d, l, t) _max R), s is the source node, d is the destination node, l is the data stream length, t _max R is the distribution rate for the maximum end-to-end delay; the network benefit is defined as

Wherein v is the service class, c ^p For operator revenue, c ^s The standard grade gain is obtained, delta is time delay, lambda is jitter rate, and epsilon is packet loss rate; if the total number of the services requested by the user is m, the basic parameters and the network benefits which need to be processed by the analysis module are respectively (B) ₁ ,B ₂ ,...,B _m) and (R₁ ,R ₂ ,...,R _m ) In summary, the business requirement model I is expressed as:

the service model S refers to a service performance index and a necessary service function requirement, wherein the performance index is defined as Q = (t, delta, lambda and epsilon), wherein t is a throughput rate, delta is a time delay, lambda is a jitter rate, and epsilon is a packet loss rate; the service function requirements F include: network function F ^N Functional dependency relationship F ^Q And network function dependent resource type F ^R I.e. F = (F) ^N ,F ^Q ,F ^R ) (ii) a When the number of service requests is m, the service model is expressed as:

wherein ,Q_i And F _i Respectively representing the performance index of the ith service and the corresponding service function requirement, i belongs to (1, m);

the route addressing function in the path calculation module combines the service requirement matrix S defined above with the routing algorithm through a mapping function

Mapping to obtain a service path, wherein a routing algorithm Rt refers to a shortest path algorithm or a K shortest path algorithm,

the mapping function is used for selecting all possible transmission paths for the service according to the service performance requirement, and the process of converting the service requirement into the service path by the control layer is expressed as follows:

after the service path is established, data transmission is carried out, and if the application requirement cannot be met or the constraint condition of routing adjustment cannot be met, path calculation is executed again;

when the service path meets the condition, the resource management module reasonably distributes resources by adopting a scheduling algorithm based on a Circular Queue Forwarding (CQF) mechanism or asynchronous shaping ATS (automatic traffic Forwarding) according to the path P and the feedback information of the knowledge layer, so that the service end-to-end delay is minimized and the network resource utilization is maximized.

3. The deterministic network-based power communication network architecture system of claim 1,

the infrastructure layer is positioned at the bottommost layer of the whole framework and comprises deterministic forwarding equipment and deterministic processing equipment, and the deterministic forwarding equipment does not have a routing function and only forwards data; the deterministic processing device not only has data forwarding capability, but also has processing functions for implementing data by programming, specifically:

the infrastructure layer functions are implemented by node state information, service capabilities and function instances, where N is defined as all physical nodes N in the network _i I is the number of physical nodes; the state information of the node is defined as F = (N) _l ,N _s ,N _d ,N _im ) Element components in the state information respectively represent node positions, node types, node connectivity and node importance; service capability is defined as C = (C) _c ,C _h ,C _t ) Including computing power C _c Buffer capacity C _h And transmission capacity C _t Wherein each class of service capability is specifically divided by a subscript j; function example E = (E) _c ,E _h ,E _t ) Resource service capability of the corresponding node, wherein E _c To calculate function instances, E _h For example of a caching function, E _t For the transport of a function instance, i.e. a service instance of node n on class j capability of transport resource instance E is defined as E, where

Respectively representing service instances of the nodes on the ith type capability of the computing, caching and transmitting functions, i e (1, j), and based on the definition, the resource instantiation result of the physical node on which the kth service node depends on the service path P is as follows:

the mapping from the service path to the data layer service instance resource combination is therefore represented as:

wherein ,

representing a mapping function that transforms the nodes traversed by the path P into refined resource instances according to the service capabilities C and the service instances F, d _i The resource instantiation result, i ∈ (1, k), of the ith service node is represented.

4. The deterministic network-based power communication network architecture system of claim 1,

the knowledge layer is communicated with the infrastructure layer and the control layer through a programmable interface and comprises an enhanced knowledge management module and a state management information base; the state management information base stores a service resource combination example for real-time updating of equipment states, the enhanced knowledge management module records a scheduling strategy of a control layer, learns scheduling algorithms with different service requirements by combining a machine learning algorithm, and predicts an optimal scheduling strategy of a data stream according to the learned knowledge, wherein the knowledge is defined as K = (lh, br, D) _max ,D _T ,C _l ) Lh is the data flow length, data burst rate br, maximum end-to-end delay D _max Flow cutoff time D _T And link capability C _l 。

5. A deterministic network based power communication network architecture system according to claim 4,

the control layer with knowledge layer collaborative work realizes timely accurate feedback mechanism, includes: when the control layer performs data stream scheduling, the control layer firstly analyzes the data stream to obtain the performance requirements of the data stream, including tolerable maximum time delay, data packet length and data burst rate; then, mapping the route and the service according to the analyzed performance model of the data stream to obtain a deterministic service path; secondly, the control layer requests the existing scheduling knowledge including synchronous scheduling or asynchronous scheduling from the knowledge layer, and selects a corresponding scheduling strategy according to the underlying network resource state recorded by the information state database; meanwhile, the knowledge layer carries out intelligent learning decision according to the existing scheduling algorithm and traffic characteristics, the learned knowledge is used for predicting the scheduling strategy of the subsequent traffic, and finally, the routing and scheduling strategy is issued to the infrastructure layer for corresponding forwarding and processing.

6. A deterministic network based power communication network architecture system according to claim 4,

the knowledge management module and the equipment information state database of the knowledge layer collect and update the running state information of the bottom layer forwarding equipment, and autonomously send state update information to the management module when the running state of the bottom layer forwarding equipment changes; the enhanced knowledge management module periodically sends state maintenance information to the bottom forwarding equipment to prevent the communication with the knowledge layer independently when a fault occurs, and the control layer directly accesses the information state database to obtain real-time network equipment running state information when carrying out data flow scheduling distribution, so that an appropriate strategy is selected to carry out optimal distribution according to service requirements; the enhancement management module excavates and learns the algorithm of knowledge according to the demand of the control layer, and sends corresponding knowledge to the control layer, thereby realizing intelligent management and decision.

7. The method for scheduling power communication network resources, implemented by the power communication network architecture system based on deterministic networks according to any of claims 1 to 6, characterized in that the method comprises the following steps:

step 1, aiming at a power grid deterministic service, a control layer obtains a performance requirement of the service through a data analysis module and maps the performance requirement into a service function model S;

step 2, transmitting the service function model to a routing control module, and combining a routing algorithm to realize a deterministic transmission path;

step 3, the resource management module obtains resource strategy related knowledge from the knowledge layer, and invokes a scheduling algorithm to allocate resources by combining the transmission path P;

step 4, according to the equipment state information recorded by the state information base of the knowledge layer, if the service resource combination meets the routing constraint and the service performance requirement, the scheduling is already carried out at this moment, the control layer determines a path allocation resource to the bottom layer, completes the execution of a buffer mechanism, a queue scheduling mechanism and a path selection mechanism of the deterministic requirement, and transfers an allocation command to the infrastructure layer; if the scheduling conditions are not met, performing scheduling calculation and strategy learning again;

step 5, the infrastructure layer receives the dispatching command of the control layer and transmits the service data packet; and after the data stream transmission is finished, updating the network state and storing the state information in a state information base of the knowledge layer.

Images