Service orchestration method for industry dedicated communication network

Abstract

The invention relates to the technical field of cloud servitization network orchestration, in particular to a service orchestration method of an industry dedicated communication network, which comprises the following steps of: firstly, modeling a network service process by adopting an object-oriented thought to realize unified modeling of a service function and network resources; secondly, parameterizing network node resources through a mathematical model, describing the operation state of a bearer network, providing node service capability definiteness and flow control parameters, and supporting the certainty of bearer resources; and finally, realizing deterministic arrangement of reachable paths and execution time of each functional layer when multiple service flows are parallel in a dynamic environment. Therefore, the problem that the existing industrial bearer network cannot meet the service scene with the deterministic requirement in the industrial internet is solved.

Description

Technical Field

[0001] The present invention relates to the field of cloud-based service-oriented network orchestration technology, and in particular to a service orchestration method for an industry-specific communication network. Background Art

[0002] Next-generation information technology, represented by 5G, is driving the industrial internet toward cloudification, service-oriented development, and virtualization. Through Network Function Virtualization (NFV) and Software-Defined Network (SDN) technologies, NFV provides independent network resource control, open network programmability, and customized network resources for different business scenarios. This enables rapid, customized service rollout and flexible, on-demand resource allocation. This has led to the concept of Network as a Service (NaaS), which transforms underlying heterogeneous network resources into services and orchestrates them on-demand as bearer services.

[0003] In the NaaS architecture, the network services required by a business consist of a series of connected virtualized network functions (VNFs). VNFs can be dynamically created and deleted, offering high flexibility and scalability. The construction and deployment of network services primarily consider the number of VNFs, the order of network functions within the service function chain, and the resource allocation of the entire service function chain within the virtualized infrastructure (NFVI). The logical topology encompassing multiple service chains can be represented by a virtual network function forwarding graph (VNF-FG), allowing service requests to be fulfilled through the orchestration of the VNF-FG.

[0004] VNF-FG orchestration can design network service templates based on specific application scenarios, and then drive the relevant function opening and resource management of the underlying infrastructure domain based on the network service templates, ultimately meeting specific application requirements. Specifically, it can be divided into high-level business orchestration and low-level resource orchestration. Business orchestration is mainly responsible for ensuring end-to-end business availability, while resource orchestration provides corresponding low-level basic resource scheduling. Its architecture is as follows: Figure 1As shown. First, the service is defined according to the application scenario. The Communication Service Management Function (CSMF) converts the service's communication service requirements into network slicing requirements for the Network Slice Management Function (NSMF), namely the network service requirements, which include the service's regional capacity, coverage area, isolation level, accessibility, reliability, bandwidth, latency and other requirements. Then, the NSMF decomposes the network slicing requirements into the Service Level Agreement (SLA) requirements of the network subslice and sends the network subslice deployment request to the Network Slice Subnet Management Function (NSSMF). Finally, the NSSMFs in various fields convert the SLA requirements of the network slice subnet into the service parameters of the network element and send them down.

[0005] The existing bearer infrastructure used by the Industrial Internet, such as 5G / 6G networks, non-terrestrial networks (NTN), and fiber optic networks, still uses the ideas and structures of the traditional Internet field. It consists of a bearer network + industrial applications, which is a simple layered structure. The bearer network is only regarded as a transmission channel for industrial application data. It lacks integration with business control processes and cannot meet the business scenarios with deterministic requirements in the Industrial Internet.

[0006] Currently, there are many solutions to the problem of network service orchestration. CN 110858807 B discloses a layered collaborative architecture that dynamically distributes service policy objects through a dual mapping table mechanism to coordinate the dynamic orchestration of service policies in multi-operator networks. CN 115633083 A discloses a priority-aware reinforcement learning algorithm that dynamically jointly optimizes routing and spectrum selection under non-global information to minimize network overhead. CN 106533966B formally models network topology and service requirements, and adopts a greedy algorithm to dynamically jointly plan traffic paths and function deployment locations to achieve collaborative optimization of resource utilization and load balancing. The above-mentioned network orchestration scheme has the following functional defects: it does not uniformly model network resources and business flows, and cannot formally describe the behavior between business flow layers, and the policy orchestration is based on static resource orchestration and lacks adaptability to dynamic network environments. Therefore, it is necessary to refer to the NaaS architecture, construct a unified model description by abstracting heterogeneous bearer network resources, enhance the coupling relationship between functional layers, and design an orchestration algorithm in a dynamic network environment to ensure that the spatiotemporal dimensions of industrial services are reachable and must be reached. Summary of the Invention

[0007] The purpose of the present invention is to provide a service orchestration method for an industry-specific communication network, aiming to solve the problem that the existing industrial bearer network cannot meet the business scenarios with deterministic requirements in the industrial Internet.

[0008] To achieve the above object, the present invention provides a service orchestration method for an industry-specific communication network, characterized in that it includes the following steps:

[0009] Use time-delay PetriNet to model network service processes, achieve a unified description of industrial services and telecom cloud networks, and perform deadlock avoidance and reachability analysis.

[0010] Deterministic network calculus is used to parameterize delay PetriNet Token and arc, providing node service capability and delay bound for the delay PetriNet service orchestration model;

[0011] Use process algebra to complete the abstract refinement and model conversion of the bearer layer PetriNet, providing resource constraints and operational determinism for service orchestration;

[0012] A deterministic service orchestration solution based on reachability graphs and heuristic optimization algorithms, combined with genetic algorithms to dynamically combine and optimize the end-to-end reachable paths and execution times of different services, unifies the temporal determinism and spatial reachability of the functional layer.

[0013] Among them, in "Using Delay PetriNet to Model Network Service Processes, Achieve a Unified Description of Industrial Services and Telecom Cloud Networks, and Perform Deadlock Avoidance and Reachability Analysis", the delay PetriNet is redefined as a sextuple, including instantiated network functions or devices, changes, and data flow relationships, to formally describe the behavior between business flow layers and the service interaction process.

[0014] The following steps are included in the "Using Delay PetriNet to Model Network Service Processes, Implement a Unified Description of Industrial Services and Telecom Cloud Networks, and Perform Deadlock Avoidance and Reachability Analysis":

[0015] Define business timing logic relationships and perform deadlock avoidance and reachability analysis;

[0016] Translate business intent into automated operational intent that can be understood by the network;

[0017] Use time-delay PetriNet to model the static spatiotemporal logical relationships of network functions;

[0018] After completing the service process modeling, dynamic property testing and feedback are carried out.

[0019] The following steps are included in "Using deterministic network calculus to parameterize delay PetriNet tokens and arcs to provide node service capabilities and delay bounds for the delay PetriNet service orchestration model":

[0020] Provide node service capabilities and delay bounds for the latency PetriNet service orchestration model to derive the latency of the entire service process;

[0021] Use network calculus arrival curves and service curves to describe the delay PetriNet transition triggering conditions and execution capabilities.

[0022] The "Deterministic Service Orchestration Solution Based on Reachability Graph and Heuristic Optimization Algorithm" includes the following steps:

[0023] Define chromosome as a matrix;

[0024] The initial population uses a greedy strategy to initialize candidate solutions after selecting the business request;

[0025] Calculate the fitness of each individual in the population, and randomly select the best individual to enter the offspring population according to the binary tournament selection strategy;

[0026] A new fitness function is used to calculate the fitness value of individuals in the population. The processing rate and execution time parameters of the nodes are adjusted based on network calculations until the maximum number of iterations is met. The optimal individual is obtained and matched with the solution space to obtain the optimal service path under the current environment of different businesses.

[0027] The present invention provides a service orchestration method for an industry-specific communication network. First, the method uses object-oriented thinking to model the network service process, achieving unified modeling of service functions and network resources. Second, it uses mathematical models to parameterize network node resources and describe the operating status of the bearer network, providing node service capability bounds and flow control parameters to support the determinism of bearer resources. Finally, it achieves deterministic orchestration of reachable paths and execution times for each functional layer when multiple business flows run in parallel in a dynamic environment. This solves the problem that existing industrial bearer networks cannot meet the business scenarios with deterministic requirements in the Industrial Internet. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0029] Figure 1 It is a layered network service and orchestration management architecture.

[0030] Figure 2 This is a diagram of the industrial business service orchestration process.

[0031] Figure 3 This is a diagram of the Petri Net service orchestration object model.

[0032] Figure 4 It is a class diagram of the Petri Net service orchestration model.

[0033] Figure 5 This is a diagram of the Petri Net service orchestration construction process.

[0034] Figure 6 It is a diagram of deadlock analysis and deadlock avoidance strategy.

[0035] Figure 7 It is a schematic diagram of the mapping relationship between functional models.

[0036] Figure 8 This is a schematic diagram of the delay composition of the business service process.

[0037] Figure 9 It is a diagram of the Petri Net model of the internal process of the device.

[0038] Figure 10 This is a diagram of the steps of the MOGA-DO algorithm.

[0039] Figure 11 This is a flow chart of a service orchestration method for an industry-specific communication network provided by the present invention.

[0040] Figure 12 It is a flowchart that uses time-delay PetriNet to model the network service process, achieve a unified description of industrial services and telecom cloud networks, and perform deadlock avoidance and reachability analysis.

[0041] Figure 13 It is a flowchart that uses deterministic network calculus to parameterize delay PetriNetToken and arc, and provides node service capabilities and delay bounds for the delay PetriNet service orchestration model.

[0042] Figure 14 It is a deterministic service orchestration solution based on reachability graphs and heuristic optimization algorithms. It combines genetic algorithms to dynamically combine and optimize the end-to-end reachable paths and execution times of different businesses, and provides a unified flow chart for the time determinism and spatial reachability of the functional layer. DETAILED DESCRIPTION

[0043] The following describes embodiments of the present invention in detail, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and are not to be construed as limiting the present invention.

[0044] See also Figures 1 to 14 The present invention provides a service arrangement method for an industry-specific communication network, comprising the following steps:

[0045] S1 uses time-delayed PetriNet to model network service processes, achieving a unified description of industrial services and telecom cloud networks, and performing deadlock avoidance and reachability analysis.

[0046] The delay PetriNet is redefined as a six-tuple, including instantiated network functions or devices, transitions, and data flow relationships, to formally describe the behavior between business flow layers and the service interaction process.

[0047] Industrial Internet delay Petri Net is redefined as a six-tuple Φ=(P,T L ,T G ,F,M,I), Figure 3 The service orchestration object model is shown, where:

[0048] 1)p i ∈P is an instantiated network function or device containing service attributes and parameters;

[0049] 2)t i ∈T is a transition abstracted from data operations. T L and T G are the node-local and network-global transition sets respectively;

[0050] 3) F represents the business data flow relationship;

[0051] 4) Token is defined as a business flow containing parameters such as data object, priority, port ID, logical address, etc.

[0052] S11 defines the business temporal logic relationship and performs deadlock avoidance and reachability analysis;

[0053] Specifically, the deadlock analysis and deadlock avoidance strategy, the specific process is as follows:

[0054] 1) Define deadlock state s d Contains deadlock place set P d =(p1,p2,…,p k ), select the starting node p i ∈P dGenerate a reachable tree, and if the deadlock-free constraint is not satisfied, output (s d ,P d ), go to step 2), otherwise go to step 3);

[0055] 2) In P d Reselect p r , if p r If it has been selected, go to step 3), otherwise repeat this operation;

[0056] 3) If there is an arc (p i ,t,p r ), and satisfy the triggering condition of transition t, let p i =p r , return to step 1).

[0057] 4) Repeat the above steps 1)-3) to obtain the deadlock identification set {s1,s2,…,s n} and the corresponding deadlock place set {P1,P2,…,P n}, avoid deadlock by increasing system resources in the deadlock library or adding control libraries to ensure system activity.

[0058] S12 translates business intent into automated operational intent that can be understood by the network;

[0059] Specifically, the business intent is translated into automated operation intent that can be understood by the network. The attributes, methods and other factors contained in the service determine the data negotiation and exchange between network functional entities.

[0060] S13 uses time-delay PetriNet to model the static spatiotemporal logical relationships of network functions;

[0061] Specifically, the time-delay Petri Net is used to model the static spatiotemporal logical relationship of network functions, and the available nodes and reachable VNF-FG paths of the entire service process are presented in the form of a Petri Net reachability graph.

[0062] After completing the service process modeling, S14 conducts dynamic property inspection and feedback.

[0063] Specifically, after the service model is built, the system activity test is first performed. Distributed industrial equipment triggers local state transfer through local changes, and is interconnected with the 5G network through global changes to achieve wide-area information transmission. Industrial equipment needs to coordinate the running time of functions with each other to meet the deadlock-free condition constraints. However, in the actual operation process, the constraints are not always met. In order to reduce the cost of operation and maintenance, deadlock avoidance must be performed during the model building phase. Activity is a necessary condition for the continuous operation of the system, and deadlock-free is a typical manifestation. When multiple business flows are running in parallel, improper resource allocation or process conflicts will cause unreasonable design and deployment of VNF-FG, thereby causing deadlock or crosstalk in the service process. The activity and deadlock-free nature of the system are defined according to the following formula:

[0064]

[0065] Among them, formula (1) is the definition of system activity, and formula (2) is the definition of system deadlock-free.

[0066] Figure 6 The deadlock analysis and deadlock avoidance strategy are shown. The specific process is as follows:

[0067] 1) Define deadlock state s d Contains deadlock place set P d =(p1,p2,…,p k ), select the starting node p i ∈P d Generate a reachable tree, and if the deadlock-free constraint is not satisfied, output (s d ,P d ), go to step 2), otherwise go to step 3);

[0068] 2) In P d Reselect p r , if p r If it has been selected, go to step 3), otherwise repeat this operation;

[0069] 3) If there is an arc (p i ,t,p r ), and satisfy the triggering condition of transition t, let p i =p r , return to step 1).

[0070] 4) Repeat the above steps 1)-3) to obtain the deadlock identification set {s1,s2,…,s n} and the corresponding deadlock place set {P1,P2,…,P n}, avoid deadlock by increasing system resources in the deadlock library or adding control libraries to ensure system activity.

[0071] The core concept of the deadlock avoidance strategy is to add places to constrain network system behavior, ensuring deadlock-free networks and thus ensuring service availability. In PetriNet, adding places or tokens within places corresponds to replanning network paths and increasing bandwidth, computing, and other network resources in real-world scenarios.

[0072] After completing the system activity test, the service path reachability analysis needs to be performed. The Industrial Internet delay Petri Net Φ=(P,T L ,T G ,F,M,I)’s time dimension is defined as:

[0073]

[0074] The notation of s=(m,u) is extended to m∈M(|P|,d), satisfying The state update formula is:

[0075]

[0076] Among them, the time expansion initial identifier is The expanded state equation provides a reachable identification time-space dimension relationship. If in the global transition sequence σ=t1,t2,…,t n ∈T G Trigger the next m (n) Since m (0) is reachable, then the state equation is expressed as:

[0077] m (n) =m (0)· R n-1 +C·ψ σ (5)

[0078] Where C is the time-extended incidence matrix, satisfying:

[0079] C∶=(C (1) ,…,C (|T|) ),C (k) ∈M(|P|,d),k∈{1,…,|T|} (6)

[0080]

[0081] Among them, R∈M(d,d) is the process matrix, r i,j ∈R satisfies:

[0082]

[0083] ψ∈M(d·|T|,d) is the Parikh matrix. i ∈σ, satisfying the following formula:

[0084]

[0085] Industrial business designated industrial Internet delay Petri Net initial identifier m (0) and the termination mark m (n) , according to the state equation, the spatiotemporal correlation of different states s = (m,u) is established.

[0086] Abstract the virtual network into a directed graph G v =(N v ,L v ), virtual node set Virtual Link Set Compute resource requests Bandwidth resource request amount The physical network is modeled as an undirected graph G p (N p ,L p ), servers, switches and other devices form a physical node collection Representation node To Node Physical links, nodes can use CPU computing resources The bandwidth capacity is Business service request set Γ={Γ k |k∈[1,|Γ|]}, defined as the quintuple Γ k ={s k ,d k ,A k ,F k ,T k}, where s k ,d k Representative business request Γ k The source node and destination node, A k is the business attribute, T k is the upper bound of the maximum allowed delay, and the service function set h represents Γ k Required order of function execution.

[0087] Petri Net service orchestration model is based on service request Γ k Match network functions and nodes and guide traffic. The mapping relationship between them is further abstracted using time-delay Petri Net. is a Petri Net place, and the node state composition is identified as s:R→V, V={v i |i∈[1,|V|]} ordered connection representation The corresponding service operation is transition, and the optimization goals are as follows:

[0088]

[0089] Among them, u is the description of the virtual node and physical nodes The binary variable of the mapping relationship. Formula (12) is the time constraint. The actual execution time cannot exceed the time upper bound. Formula (13) is the network function timing logic constraint. For nodes with strong dependency, the processing time of the previous function must be strictly earlier than the latter. Formula (14) is the computing resource constraint. The time allocated to the physical node The amount of computing resources requested cannot exceed the available amount; Formula (15) is the bandwidth constraint, and the physical link The bandwidth request of the bearer cannot exceed the available bandwidth resources.

[0090] S2 uses deterministic network calculus to parameterize delay PetriNet Tokens and arcs, providing node service capabilities and delay bounds for the delay PetriNet service orchestration model;

[0091] S21 provides node service capabilities and delay bounds for the latency PetriNet service orchestration model to derive the latency of the entire service process;

[0092] Specifically, deterministic network calculus is used to parameterize the delay Petri Net Token and arc, providing node service capabilities and delay bounds for the delay Petri Net service orchestration model, so as to derive the delay of the entire service process and realize dynamic time planning of the service process.

[0093] S22 uses network calculus arrival curves and service curves to describe the delay PetriNet transition triggering conditions and execution capabilities.

[0094] Specifically, we use network calculus arrival curves and service curves to describe the delay PetriNet transition trigger condition β(p,t) and execution capability β(t,p). During the service process, the classifier identifies the traffic type, and then passes through each service function in order. The node data exchange is completed by the forwarder. Therefore, the entire service process time is the sum of the fixed delay and variable delay of the forwarding node, functional node, and link, as shown in Figure 2. Figure 13 As shown. The business flow flowing through the library and transition is subject to a fixed delay λ i Limitations, providing service curves On the other hand, affected by variable delay, the service curve ξ i Network capabilities The service curves of each node are still connected in series to form a rate-delay curve. k The reachable path needs to pass through n nodes, so the upper and lower bounds of the entire process delay of a single service path satisfy:

[0095]

[0096] S3 uses process algebra to complete the abstract refinement and model conversion of the bearer layer PetriNet, providing resource constraints and operational determinism for service orchestration;

[0097] Specifically, the process algebra is used to simplify the Petri Net service process model of the carrier layer. Figure 9 Taking the Petri Net model of the concurrent process inside the device shown as an example, the initial state m0 = (1, 0, 0, 0, 0, 0, 0), the process algebra is expressed as:

[0098]

[0099] S4 is a deterministic service orchestration solution based on reachability graphs and heuristic optimization algorithms. It combines genetic algorithms to dynamically combine and optimize the end-to-end reachable paths and execution times of different services, unifying the temporal determinism and spatial reachability of the functional layer.

[0100] S41 defines chromosomes as matrices;

[0101] Specifically, the chromosome is defined as n×|F k | matrix, n is the number of places, |F k |For BusinessΓ k The number of network functions required, the chromosome matrix uses binary encoding, if the library can be the current Provides service, then the value is 1, that is, the column vector is Available Petri Net reachable identifiers, the row vector is the library p i In different The chromosome fragment reflects the service operation corresponding to the PetriNet state change, according to different business requests Γ k ={s k ,d k ,A k ,F k ,T k}, network performance parameters such as latency, bandwidth resources, and computing power are used to select chromosomes to form individual X.

[0102] The initial population S42 uses a greedy strategy to initialize candidate solutions after selecting the business request;

[0103] Specifically, the initial population Po0 is selected k ,d k ,F k Then a greedy strategy is used to initialize the candidate solutions.

[0104] S43 calculates the fitness of each individual in the population and randomly selects the best individual to enter the offspring population according to the binary tournament selection strategy;

[0105] Specifically, during the evolution from generation i to generation i+1, calculate the number of N individuals in the population Po i The fitness fit(X i ), according to the binary tournament selection strategy, the best individuals are randomly selected to enter the offspring population until the population size is equal. The probability of each individual being selected is p k =c(1-c) k-1 , k is the optimal individual X best1 The probability of being selected. Then select two parents R1 and R2 for partial matching crossover. According to the Bernoulli distribution B(n,p), the offspring has a probability of p to copy the parent chromosome with a higher fitness value. The mutation operator uses replacement mutation to obtain M new individuals by randomly replacing genes, forming the population Po i+1 Since crossover and mutation operations may result in individuals not meeting the constraints of the time-delay Petri Net network service model, a feasibility test of the new individuals is required after this step, that is, to verify the dynamic properties of PetriNet and whether the system constraints (12)-(15) are met, and to correct or delete individuals that do not meet the constraints. The individual selection and update process of the multi-objective genetic algorithm is the fitness sharing of the value range from a global perspective. In order to narrow the individual fitness, the Euclidean distance shown in formula (18) is used to evaluate the similarity d between individuals. x,y , and update the fitness function F'(x) to limit the large increase of individuals, maintain population diversity, and prevent network congestion.

[0106]

[0107] in: Po for the population i+1 The number of permutations and combinations of any two individuals, x i ,y i ; x,y∈Po i+1 ,i={1,2} represents the delay and bandwidth optimization objectives of individuals x and y respectively.

[0108] S44 uses a new fitness function to calculate the fitness value of individuals in the population, and adjusts the processing rate and execution time parameters of the nodes based on network calculations until the maximum number of iterations is met. The optimal individual is obtained and matched with the solution space to obtain the optimal service path under the current environment of different businesses.

[0109] Specifically, the new fitness function is used to calculate the population Po i+1The fitness value of the individuals in the network is selected, and the first N individuals are selected to form the offspring for the i+2th generation evolution. At the same time, the processing rate, execution time and other parameters of the nodes are adjusted based on the network calculation until the maximum number of iterations G is met. The optimal individual is obtained and matched with the solution space to obtain the optimal service path under the current environment of different businesses.

[0110] Example:

[0111] During the service modeling phase, an object-oriented model is used to build an industrial Internet service orchestration system. This system implements unified model planning for industrial Internet services and communication network message carrying, performs deadlock avoidance and reachability analysis, and searches for multi-flow combination paths that meet the timing logic of industrial computing and control in complex two-dimensional network topologies.

[0112] During the parameter configuration phase, mathematical models are used to derive the latency bounds and node resource parameters of the service path. Computing, storage, and bandwidth capacity resource parameters are configured for each network node object. These network node resource parameters are then converted into multi-dimensional constraints, providing resource constraints and operational determinism for service orchestration.

[0113] During the dynamic optimization phase, in multi-business scenarios, the end-to-end reachable paths and execution times of different businesses are dynamically combined and optimized, and computing, storage, and bandwidth resources are allocated to industrial multi-stream businesses in real time to achieve the unification of functional layer time determinism and spatial accessibility.

[0114] Among them, in the "service modeling phase, an object-oriented model is used to build an industrial Internet service orchestration system, realize unified model planning between industrial Internet services and communication network message carrying, and perform deadlock avoidance and reachability analysis, and find multi-flow combination paths that meet the timing logic of industrial computing and control in complex two-dimensional network topology", the following steps are included:

[0115] The "Network Node" class is used to implement unified modeling of physical devices and virtual functions. The attribute types describing network nodes include resource parameters, connection topology, timestamp, and current load status.

[0116] By formalizing the characteristics of network nodes in different industrial services, and defining the start and end tags of each characteristic based on the attributes, methods, and functional information of the network nodes, this approach encompasses all information about the network nodes and clearly describes the hierarchical structure of reference relationships between nodes.

[0117] Deploy deadlock prevention mechanisms to preemptively avoid deadlocks that may occur during system operation by increasing network resources or adding controllers to coordinate process sequences and run times.

[0118] Perform service path reachability judgment and generate a multi-flow combination path template diagram that meets the timing logic and strict time constraints of industrial control.

[0119] The parameter configuration phase involves deriving the service path's latency bounds and path node resource parameters through mathematical models, configuring computing, storage, and bandwidth capacity resource parameters for each network node object, and converting the network node resource parameters into multi-dimensional constraints to provide resource constraints and operational determinism for service orchestration. The following steps are included:

[0120] Mathematical models are used to provide network node service capabilities and latency bounds for the service orchestration model, thereby deriving the latency of the entire service process.

[0121] Parameterize the resource variables of network nodes, refine the operating status of local devices, and provide deterministic support for bearer resources.

[0122] In the dynamic optimization phase, in multi-service scenarios, the end-to-end reachable paths and execution times of different services are dynamically combined and optimized, and computing, storage, and bandwidth resources are allocated to services in real time to achieve the unification of functional layer temporal determinism and spatial reachability. The following steps are included:

[0123] According to the real-time resource occupancy status of the business flow, the optimal service path in the current environment is obtained by adjusting parameters such as the processing rate and execution time of the network nodes.

[0124] Beneficial effects:

[0125] 1. In response to the problems of consistency in the expression of new industrial Internet models and the effective transmission capability of parameters, this application uses time-delayed Petri Net to construct a unified information communication model to formally describe the behavior between business flow layers and the service interaction process, including: 1) defining the Petri Net network service object model, converting the business timing logic into a network service process Petri Net model, and achieving consistent modeling of business function requirements, VNF dependencies, and the behavior between functional layers of physical equipment operation; 2) using synchronization distance analysis, deadlock detection avoidance, and reachability judgment to ensure the coordination of the interaction process of the Petri Net service orchestration model, and pre-planning the deadlock-free available service paths and network service templates for the business in a static environment.

[0126] 2. To address the problem that static service path generation algorithms cannot achieve deterministic matching between services and bearer resources in time-varying networks, this application 1) uses network calculus to parameterize Petri Net Tokens and arcs to dynamically provide communication performance parameters such as latency upper and lower bounds, backlog time, etc. for network orchestration; 2) uses communication process algebra to optimize the local operating status of the Petri Net network service model, verify device behavior, and provide deterministic service capability support for the bearer layer; 3) proposes a dynamic service orchestration algorithm based on a multi-objective genetic algorithm to optimize the solution process of the Petri Net network service model, combines network calculus and communication process algebra to plan the reachable paths and execution times of multiple services in real time, and achieve network service quality assurance in a dynamic environment.

[0127] The above disclosure is merely a preferred embodiment of a method for orchestrating services in an industry-specific communication network according to the present invention. It is understood that this is not intended to limit the scope of the present invention. Persons skilled in the art will appreciate that any equivalent modifications made by implementing all or part of the above embodiment in accordance with the claims of the present invention are still within the scope of the invention.

Claims

1. A service orchestration method for an industry-specific communication network, characterized in that: The following steps are involved: Use time-delay PetriNet to model network service processes, achieve a unified description of industrial services and telecom cloud networks, and perform deadlock avoidance and reachability analysis. Deterministic network calculus is used to parameterize delay PetriNet Token and arc, providing node service capability and delay bound for the delay PetriNet service orchestration model; Use process algebra to complete the abstract refinement and model conversion of the bearer layer PetriNet, providing resource constraints and operational determinism for service orchestration; A deterministic service orchestration solution based on reachability graphs and heuristic optimization algorithms, combined with genetic algorithms to dynamically combine and optimize the end-to-end reachable paths and execution times of different services, unifies the temporal determinism and spatial reachability of the functional layer.

2. The service arrangement method of the industry-specific communication network according to claim 1, characterized in that: In the paper "Using Delay PetriNet to Model Network Service Processes, Achieve a Unified Description of Industrial Services and Telecom Cloud Networks, and Perform Deadlock Avoidance and Reachability Analysis," the delay PetriNet is redefined as a sextuple, including instantiated network functions or devices, transitions, and data flow relationships, to formally describe inter-service flow layer behavior and service interaction processes.

3. The service arrangement method of the industry-specific communication network according to claim 1, characterized in that: The following steps are included in the "Using Delay PetriNet to Model Network Service Processes, Implement a Unified Description of Industrial Services and Telecom Cloud Networks, and Perform Deadlock Avoidance and Reachability Analysis": Define business timing logic relationships and perform deadlock avoidance and reachability analysis; Translate business intent into automated operational intent that can be understood by the network; Use time-delay PetriNet to model the static spatiotemporal logical relationships of network functions; After completing the service process modeling, dynamic property testing and feedback are carried out.

4. The service arrangement method of the industry-specific communication network according to claim 1, characterized in that: In "Using deterministic network calculus to parameterize time-delayed PetriNet tokens and arcs to provide node service capabilities and delay bounds for the time-delayed PetriNet service orchestration model," the following steps are included: Provide node service capabilities and delay bounds for the latency PetriNet service orchestration model to derive the latency of the entire service process; Use network calculus arrival curves and service curves to describe the delay PetriNet transition triggering conditions and execution capabilities.

5. The service arrangement method of the industry-specific communication network according to claim 1, characterized in that: The following steps are included in the "Deterministic Service Orchestration Solution Based on Reachability Graph and Heuristic Optimization Algorithm": Define chromosome as a matrix; The initial population uses a greedy strategy to initialize candidate solutions after selecting the business request; Calculate the fitness of each individual in the population, and randomly select the best individual to enter the offspring population according to the binary tournament selection strategy; A new fitness function is used to calculate the fitness value of individuals in the population. The processing rate and execution time parameters of the nodes are adjusted based on network calculations until the maximum number of iterations is met. The optimal individual is obtained and matched with the solution space to obtain the optimal service path under the current environment of different businesses.

Images