Digital twin modeling method and device for multi-modal network element
By building a digital twin model in a multimodal network and verifying the modal control strategy, the risk problem when directly acting on the physical network is solved, and higher policy accuracy and network stability are achieved.
Patent Information
- Application Number
- CN202510395301.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-31
- Publication Date
- 2025-07-01
AI Technical Summary
In a multimodal network environment, when the modal control strategy of the control system directly acts on the physical network, it faces high risks, which may lead to configuration errors and network performance degradation.
By introducing network modal twins, a digital twin model of the physical network is built, so that the modal control strategy can be experimental and verified on the digital twin model and then act on the physical network modal, thereby reducing risks. The specific methods include constructing and fusion of corresponding models based on the appearance model, southbound protocol, association information, modal identification and constraint information of network element devices in a multimodal network, and finally building a digital twin model of multimodal network element.
The accurate experiment and verification of modal control strategies on the digital twin model of multimodal network elements is achieved, which improves the accuracy and reliability of the strategy and reduces the risk of direct effect on the physical network.
Smart Images
Figure CN120234976A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of communication technologies, and particularly to a digital twin modeling method and device for a multi-modal network element. Background Art
[0002] With the development of information technologies, the types of network modalities are also continuously increasing. As the basic units constituting a complex network, multi-modal network elements have strong discreteness in functional characteristics, and there are significant differences in protocol support, data processing capabilities, and topological structures among network elements of different modalities. In addition, the physical network composed of multi-modal network elements may be extremely large in scale and complex in topological relationships, which further increases the difficulty of network management.
[0003] In a multi-modal network environment, when the modal control strategy of a control system directly acts on a physical network, it faces relatively high risks. Due to the complexity and diversity of network element modalities, directly applying the modal control strategy to the physical network may lead to configuration errors and thus affect the normal operation of the network. The performance differences of devices of different modalities may cause the control modal strategy to be unable to be effectively executed on some physical devices, affecting the overall performance of the network.
[0004] To solve the above problems, by introducing network modal twins and constructing a digital twin model of the physical network, enabling the modal control strategy to be experimented and verified on the digital twin model before acting on the physical network modality, the risks brought by network programming can be effectively reduced. However, accurately performing digital twin modeling of multi-modal network elements is the key to realizing accurate experimental verification of the modal control strategy on the digital twin model. Therefore, how to accurately perform digital twin modeling of multi-modal network elements is a technical problem that urgently needs to be solved. Summary of the Invention
[0005] The present application provides a digital twin modeling method and device for a multi-modal network element, which realizes accurate digital twin modeling of the multi-modal network element, thereby improving the accuracy of the experimental verification results of the control modal strategy.
[0006] In a first aspect, an embodiment of the present application provides a digital twin modeling method for a multi-modal network element. The digital twin modeling method for the multi-modal network element includes:
[0007] According to the appearance model information, southbound protocol information, association information, modal recognition information, and corresponding constraint information of various types of network element devices in a multi-modal network, respectively construct an appearance model, a southbound protocol model, an association information model, a modal recognition model, and a corresponding constraint model of the corresponding type;
[0008] Fuse all types of appearance models, southbound protocol models, association information models, modality recognition models and their corresponding constraint models respectively to construct a fused appearance model, a fused southbound protocol model, a fused association information model and a fused modality recognition model;
[0009] Based on the modality feature model output after the application of the fused modality recognition model, combine the path selection mechanism of the multi-modal network to construct a fused modality control model;
[0010] Fuse the fused appearance model, the fused southbound protocol model, the fused association information model, the fused modality recognition model and the fused modality control model to construct a multi-modal network element digital twin model.
[0011] Combined with the first aspect, in one implementation, various types of network element devices in the multi-modal network include P4 devices and DCI devices.
[0012] Combined with the first aspect, in one implementation, construct the appearance model and the corresponding constraint model of the corresponding type according to the appearance model information and the corresponding constraint information of various types of network element devices in the multi-modal network, and fuse all types of appearance models and the corresponding constraint models to construct a fused appearance model, including:
[0013] Construct the appearance model of the corresponding type based on the appearance pictures and identification information of each component in various types of network element devices, where the identification information includes the model of the network element device and the type of the component in the network element device, and the components include the chassis, board cards and single disks;
[0014] If the network element device is a frame device, use the board card information, port information and optical module information of the network element device as the constraint information to construct the corresponding appearance constraint model;
[0015] If the network element device is a non-frame device, use the slot information occupied by the network element device, the information of the single disk inserted in each slot, the port information and the optical module information as the constraint information to construct the corresponding appearance constraint model;
[0016] Fuse the appearance models of various types of network element devices and the corresponding appearance constraint models to generate a fused appearance model:
[0017] M_1 = ∪_i = 1^i = j(A_i ∩ Ca_i)
[0018] Where, M_1 is the fused appearance model, A_i is the appearance model of the i-th type of network element device, Ca_i is the appearance constraint model of the i-th type of network element device, and j is the type of network element device in the multi-modal network.
[0019] In combination with the first aspect, in one implementation, a southbound protocol model of the corresponding type and a corresponding constraint model are constructed according to the southbound protocol information and the corresponding constraint information of various types of network element devices in the multi-modal network, and a fused southbound model is constructed by fusing all types of southbound protocol models and the corresponding constraint models, including:
[0020] If the southbound protocol of the network element device supports the P4Runtime protocol, a corresponding southbound protocol model is constructed according to the data model, message passing model, and operation model of the P4Runtime protocol;
[0021] If the southbound protocol of the network element device supports the Netconf protocol, a corresponding southbound protocol model is constructed according to the content layer model, operation layer model, RPC layer model, and transport layer model of the Netconf protocol;
[0022] If the type of the network element device is a P4 device, the southbound protocol of the P4 device is supported to be the P4Runtime protocol and does not support the Netconf protocol as the constraint information, and a corresponding southbound protocol constraint model of the P4 device is constructed;
[0023] If the type of the network element device is a DCI device, the southbound protocol of the DCI device is supported to be the Netconf protocol and does not support the P4Runtime protocol as the constraint information, and a corresponding southbound protocol constraint model of the DCI device is constructed;
[0024] The southbound protocol models of various types of network element devices and the corresponding southbound protocol constraint models are fused to construct a fused southbound protocol model:
[0025] M_2 = ∪_i = 1^i = j (P_i ∩ Cp_i)
[0026] Among them, M_2 is the fused southbound protocol model, P_i is the southbound protocol model of the i-th type of network element device, Cp_i is the southbound protocol constraint model of the i-th type of network element device, and j is the type of network element device in the multi-modal network.
[0027] In combination with the first aspect, in one implementation, an association information model of the corresponding type and a corresponding constraint model are constructed according to the association information and the corresponding constraint information of various types of network element devices in the multi-modal network, and a fused association information model is constructed by fusing all types of association models and the corresponding constraint models, including:
[0028] A P4 device association information model is established according to the link information and neighbor information of the multi-modal network;
[0029] A DCI device association information model is derived and constructed according to the P4 device association information model;
[0030] If the network element device is a P4 device, use the neighbors of the P4 device, including the P4 device, the DCI device, and / or the Host device, as constraint information to construct a corresponding P4 device association information constraint model;
[0031] If the network element device is a DCI device, use the neighbors of the DCI device, including the P4 device and / or the DCI device, excluding the Host device, as constraint information to construct a corresponding DCI device association information constraint model;
[0032] Fuse the association information models of various types of network element devices and the corresponding association information constraint models to construct a fused association information model:
[0033] M_3 = ∪_i = 1^i = j (R_i ∩ Cr_i)
[0034] Among them, M_3 is the fused association information model, R_i is the association information model of the i-th type of network element device, Cr_i is the association information constraint model of the i-th type of network element device, and j is the type of network element device in the multimodal network.
[0035] Combined with the first aspect, in an implementation, construct corresponding type appearance models and corresponding modal recognition models according to the modal recognition information and corresponding constraint information of various types of network element devices in the multimodal network, and fuse all types of modal recognition models and corresponding constraint models to construct a fused modal recognition model, including:
[0036] Parse the network packets received by various types of network element devices respectively to obtain corresponding modal identifiers;
[0037] Construct corresponding type modal recognition models according to the modal identifiers of various types of network element devices;
[0038] If the type of the network element device is a p4 device, use the network element device supporting the recognition of IP, identity, geography, and content modalities as constraint information to construct a corresponding P4 device modal recognition constraint model;
[0039] If the type of the network element device is a DCI device, use the network element device supporting the recognition of optical modality as constraint information to construct a corresponding DCI device modal recognition constraint model;
[0040] Fuse the modal recognition models of various types of network element devices and the corresponding modal recognition constraint models to generate a fused modal recognition model:
[0041] M_4 = ∪_i = 1^i = j (D_i ∩ Cd_i)
[0042] Among them, M_4 is the fusion modality recognition model, D_i is the modality recognition model of the i-th type of network element device, Cd_i is the modality recognition constraint model of the i-th type of network element device, and j is the type of network element device in the multimodal network.
[0043] Combined with the first aspect, in an implementation, for the modality feature model output after applying the fusion modality recognition model, combined with the path selection mechanism of the multimodal network, a fusion modality control model is constructed, including:
[0044] Add the addressing model and routing algorithm of the multimodal network to the modality feature model to construct the fusion modality control model.
[0045] Combined with the first aspect, in an implementation, fuse the fusion appearance model, the fusion southbound protocol model, the fusion association information model, the fusion modality recognition model, and the fusion modality control model to construct a multimodal network element digital twin model, including:
[0046] M_f = ∪_n = 1^n = 5(M_n)
[0047] Among them, M_f is the multimodal network element model, the range of n in M_n is 1 to 5, M_1 is the fusion appearance model, M_2 is the fusion southbound protocol model, M_3 is the fusion association information model, M_4 is the fusion modality recognition model, and M_5 is the fusion modality control model.
[0048] Combined with the first aspect, in an implementation, the method further includes:
[0049] Map the modality of the multimodal network element digital twin model to the modality of the physical network element device to obtain the corresponding mapping relationship:
[0050] C_f = f_1(t, p) ∩ f_2(C_o, C_a, C_s)
[0051] Among them, C_f is the set of modalities supported by the physical device, f_1(t, p) is the maximum set of modalities supported by the multimodal network element in the multimodal network, t is the network element device type, p is the southbound protocol, f_2(C_o, C_a, C_s) is the set of modalities supported by the multimodal network element in the multimodal network after implementing the modality control strategy in the multimodal network element digital twin model, C_o is the original set of modalities before implementing the control strategy, C_a is the set of modalities added during the implementation of the control strategy, and C_s is the set of modalities reduced during the implementation of the control strategy.
[0052] In the second aspect, an embodiment of the present application provides a digital twin modeling device for a multimodal network element. The digital twin modeling device for the multimodal network element includes:
[0053] The first construction module is used to respectively construct the appearance model, southbound protocol model, association information model, modality recognition model and corresponding constraint model of the corresponding type according to the appearance model information, southbound protocol information, association information, modality recognition information and corresponding constraint information of various types of network element devices in the multimodal network;
[0054] The first fusion module is used to respectively fuse all types of appearance models, southbound protocol models, association information models, modality recognition models and corresponding constraint models to construct a fused appearance model, a fused southbound protocol model, a fused association information model and a fused modality recognition model;
[0055] The second construction module is used to construct a fused modality control model based on the modality feature model output after applying the fused modality recognition model and in combination with the path selection mechanism of the multimodal network;
[0056] The second fusion module is used to fuse the fused appearance model, the fused southbound protocol model, the fused association information model, the fused modality recognition model and the fused modality control model to construct a multimodal network element digital twin model.
[0057] The beneficial effects brought by the technical solution provided by the embodiments of the present application include:
[0058] By respectively constructing the appearance model, southbound protocol model, association information model, modality recognition model and corresponding constraint model of the corresponding type according to the appearance model information, southbound protocol information, association information, modality recognition information and corresponding constraint information of various types of network element devices in the multimodal network; respectively fusing all types of appearance models, southbound protocol models, association information models, modality recognition models and corresponding constraint models to construct a fused appearance model, a fused southbound protocol model, a fused association information model and a fused modality recognition model; constructing a fused modality control model based on the modality feature model output after applying the fused modality recognition model and in combination with the path selection mechanism of the multimodal network; fusing the fused appearance model, the fused southbound protocol model, the fused association information model, the fused modality recognition model and the fused modality control model to construct a multimodal network element digital twin model, it realizes the fusion modeling of multiple types of multimodal network element devices from multiple dimensions, constructs an accurate multimodal network element digital twin model, thereby improving the accuracy of the experimental verification results of the modality control strategy on the multimodal network element digital twin model, and further greatly reducing the risk of the modality control strategy directly acting on the physical network. Description of the Drawings
[0059] Figure 1 It is a schematic flowchart of an embodiment of the digital twin modeling method for the multimodal network element of the present application;
[0060] Figure 2 Flow diagram for constructing a fusion appearance model for this application;
[0061] Figure 3 Flow diagram for constructing a fusion southbound protocol model for this application;
[0062] Figure 4 Flow diagram for constructing a fusion associated information model for this application;
[0063] Figure 5 Flow diagram for constructing a fusion modality recognition model for this application;
[0064] Figure 6 Schematic diagram of functional modules of an embodiment of a digital twin modeling device for a multi-modal network element of this application. Detailed implementation manners
[0065] In order to enable those skilled in the art of this technology to better understand the solution of this application, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of this application.
[0066] First, some technical terms in this application are explained to facilitate the understanding of this application by those skilled in the art.
[0067] P4 device: Refers to a network device that supports P4 language programming, such as a switch, etc. P4 is an open-source domain-specific language used to program data plane devices.
[0068] DCI device: Data Center Interconnect device, used to connect computer networks of different data centers to achieve data exchange and sharing.
[0069] Southbound protocol: Refers to the protocol for communication between the controller and underlying network devices (such as switches, routers, etc.) in the software-defined network (SDN) architecture.
[0070] P4Runtime protocol: A protocol used to control the data plane of network devices defined by P4 programs;
[0071] Netconf protocol: (Network Configuration Protocol) is a network configuration protocol used to manage and configure network devices.
[0072] Host device: refers to a terminal device connected to a network.
[0073] To make the objectives, technical solutions, and advantages of this application clearer, the following will further describe the embodiments of this application in detail with reference to the accompanying drawings.
[0074] In a first aspect, an embodiment of this application provides a digital twin modeling method for a multi-modal network element.
[0075] In one embodiment, referring to Figure 1 , Figure 1 is a schematic flowchart of the first embodiment of the digital twin modeling method for the multi-modal network element of this application. As Figure 1 shown, the digital twin modeling method for the multi-modal network element includes:
[0076] Step S101: According to the appearance model information, southbound protocol information, association information, modality recognition information, and corresponding constraint information of various types of network element devices in the multi-modal network, respectively construct the corresponding appearance model, southbound protocol model, association information model, modality recognition model, and corresponding constraint model.
[0077] Step S102: Respectively fuse all types of appearance models, southbound protocol models, association information models, modality recognition models, and corresponding constraint models to construct a fused appearance model, a fused southbound protocol model, a fused association information model, and a fused modality recognition model.
[0078] Step S103: Based on the modality feature model output after the application of the fused modality recognition model, combine with the path selection mechanism of the multi-modal network to construct a fused modality control model.
[0079] Step S104: Fuse the fused appearance model, the fused southbound protocol model, the fused association information model, the fused modality recognition model, and the fused modality control model to construct a digital twin model of the multi-modal network element.
[0080] It should be noted that this application conducts digital twin modeling based on dimensions such as the appearance model, southbound protocol, association information, modality recognition, modality control, and corresponding constraint information of various types of network element devices, and performs model fusion, which can construct a high-precision digital twin model of the multi-modal network element, enabling the modality control strategy to be accurately experimentally verified in the digital twin model of the multi-modal network element and obtaining accurate verification results, thereby improving the modality control ability of the modality control strategy for the multi-modal network and effectively reducing the risk of directly applying the modality control strategy to physical devices.
[0081] Among them, the types of network element devices in the multi-modal network include P4 devices, DCI devices, and other types of network element devices. To help better understand the solution of this application, the following takes the types of network element devices in the multi-modal network as P4 devices and DCI devices as an example to illustrate the digital twin modeling method of the multi-modal network elements of this application.
[0082] In one embodiment, constructing a fused appearance model includes: constructing appearance models and corresponding constraint models of corresponding types according to the appearance model information and corresponding constraint information of various types of network element devices in the multi-modal network, and fusing all types of appearance models and corresponding constraint models to construct a fused appearance model.
[0083] Specifically, as Figure 2 shown, constructing a fused appearance model includes:
[0084] Step S201, constructing an appearance model of a corresponding type based on the appearance pictures and identification information of each component in various types of network element devices, where the identification information includes the model of the network element device and the type of the component in the network element device, and the components include chassis, boards, and single disks.
[0085] Exemplarily, before constructing the fused appearance model, appearance pictures of each component of each type of network element device in different directions can be collected. In this embodiment, appearance pictures of each component of P4 devices and DCI devices in different directions are collected. Among them, the components include chassis, boards, single disks, etc., and the directions include the front, back, and side, etc. The collected appearance pictures are processed by PS to improve the clarity of the appearance pictures, and the processed appearance pictures are received and stored by the multi-modal modeling system. At the same time, the multi-modal modeling system also receives and stores the identification information such as the device model and component type corresponding to the input appearance pictures in the database for convenient calling.
[0086] The multi-modal modeling system can retrieve the appearance pictures and identification information of each component in each type of network element device stored in the database to construct appearance models of different types respectively. In this embodiment, a P4 appearance model can be constructed according to the appearance pictures and identification information of each component in the P4 device, and a DCI appearance model can be constructed according to the appearance pictures and identification information of each component in the DCI device.
[0087] Step S202, if the network element device is a frame device, constructing a corresponding appearance constraint model with the board information, port information, and optical module information of the network element device as constraint information.
[0088] Among them, the board information of the network element device includes the board model information supported by the network element device, the port information includes the port rate information supported by the network element device, and the optical module information includes the optical module model information of the network element device.
[0089] Step S203: If the network element device is a non-frame device, use the slot information occupied by the network element device, the information of the single board inserted in each slot, the port information, and the optical module information as constraint information to construct a corresponding appearance constraint model.
[0090] Among them, the slot information includes the slot numbers occupied by the network element device. For example, if the network element device occupies slots 1 to 8, the information of the single board inserted in each slot includes the single board signals of each slot.
[0091] For example, if the P4 device is a frame device, the multi-modal modeling system constructs a corresponding P4 appearance constraint model according to the board card information, port information, and optical module information of the received network element device.
[0092] If the DCI device is not a frame device, the multi-modal modeling system constructs a corresponding DCI appearance constraint model according to the received slot information, the information of the single board inserted in each slot, the port information, and the optical module information.
[0093] Step S204: Integrate the appearance models of various types of network element devices and the corresponding appearance constraint models to generate an integrated appearance model:
[0094] M_1 = ∪_i = 1^i = j(A_i ∩ Ca_i)
[0095] Among them, M_1 is the integrated appearance model, A_i is the appearance model of the i-th type of network element device, Ca_i is the appearance constraint model of the i-th type of network element device, j is the type of network element device in the multi-modal network, ∩ is the intersection symbol, and ∪ is the union symbol. In this embodiment, the network element devices include P4 devices and DCI devices, so the value of j is 2.
[0096] Explanatorily, the above formula represents the construction process of the integrated appearance model M_1. The meaning of the formula is that for each type of network element device i (from 1 to j), take the intersection A_i ∩ Ca_i of its appearance model A_i and appearance constraint model Ca_i. Perform the union operation on the intersections of all types of network element devices (from i = 1 to i = j) to obtain the final integrated appearance model M_1. Therefore, it can be understood that the appearance model of each type of network element device in this embodiment is also the set of appearance model information of the corresponding type of network element device, the appearance constraint model is also the set of corresponding constraint information, and the integrated appearance model is the union of the appearance models and appearance constraint models of all types of network element devices.
[0097] In one embodiment, constructing the integrated southbound protocol model includes: constructing the southbound protocol models of corresponding types and the corresponding constraint models according to the southbound protocol information and the corresponding constraint information of various types of network element devices in the multi-modal network, and integrating all types of southbound protocol models and the corresponding constraint models to construct the integrated southbound model.
[0098] Specifically, as Figure 3 shown, constructing a fused southbound protocol model includes:
[0099] Step S301: If the southbound protocol of the network element device supports the P4Runtime protocol, then construct a corresponding southbound protocol model according to the data model, message passing model, and operation model of the P4Runtime protocol.
[0100] Among them, the data model, message passing model, and operation model are three key components that make up the P4Runtime protocol. Among them, the data model is used to describe the runtime state of the P4 program; the message passing model is used to define the message format; the operation model defines the operation method of the controller on the network element device.
[0101] Step S302: If the southbound protocol of the network element device supports the Netconf protocol, then construct a corresponding southbound protocol model according to the content layer model, operation layer model, RPC layer model, and transport layer model of the Netconf protocol.
[0102] Among them, the Netconf protocol adopts a hierarchical structure, divided into four layers: the content layer, operation layer, RPC layer, and transport layer. Among them, the content layer defines the structure and content of the configuration data and status data of the network device, the operation layer defines the basic operations of the interaction between the client and the server, the RPC layer provides a mechanism for remote procedure call (RPC), and the transport layer is responsible for providing secure and reliable message transmission between the client and the server.
[0103] It should be noted that based on whether the southbound protocol of the network element device supports the P4Runtime protocol or the Netconf protocol, the southbound protocol models of each type of device can be constructed separately. For example, in this embodiment, the P4 device supports the P4Runtime protocol, then construct the corresponding southbound protocol model of the P4 device according to the data model, message passing model, and operation model of the P4Runtime protocol. The DCI device supports the Netconf protocol, then construct the southbound protocol model of the DCI device according to the content layer model, operation layer model, RPC layer model, and transport layer model.
[0104] Step S303: If the type of the network element device is a P4 device, then construct a corresponding southbound protocol constraint model of the P4 device with the southbound protocol supporting the P4Runtime protocol and not supporting the Netconf protocol as the constraint information.
[0105] Step S304: If the type of the network element device is a DCI device, then construct a corresponding southbound protocol constraint model of the DCI device with the southbound protocol supporting the Netconf protocol and not supporting the P4Runtime protocol as the constraint information.
[0106] Step S305: Integrate the southbound protocol models and corresponding southbound protocol constraint models of various types of network element devices to form an integrated southbound protocol model:
[0107] M_2 = ∪_i = 1^i = j (P_i ∩ Cp_i)
[0108] Among them, M_2 is the integrated southbound protocol model, P_i is the southbound protocol model of the i-th type of network element device, Cp_i is the southbound protocol constraint model of the i-th type of network element device, and j is the type of network element devices in the multimodal network. The principle of generating the integrated southbound protocol model is the same as that of generating the integrated appearance model, which will not be elaborated here.
[0109] In one embodiment, constructing the integrated association information model includes: constructing the association information model and corresponding constraint model of the corresponding type according to the association information and corresponding constraint information of various types of network element devices in the multimodal network, and integrating all types of association models and corresponding constraint models to construct the integrated association information model.
[0110] Specifically, as Figure 4 shown, constructing the integrated association information model includes:
[0111] Step S401: Establish a P4 device association information model according to the link information and neighbor information of the multimodal network.
[0112] It should be noted that the link information of the multimodal network includes the connection status and characteristics between network element devices, which includes: link status, link bandwidth, delay, packet loss rate, link type, modal characteristics of the link, etc. Neighbor information is used to describe the status and relationship between a network element device and other surrounding network element devices, which includes: the identity of the neighbor network element device, the status of the neighbor network element device, the function of the neighbor network element device, and the relationship with the neighbor network element device.
[0113] Step S402: Derive and construct a DCI device association information model according to the P4 device association information model.
[0114] Exemplarily, the multimodal modeling system uses the P4 device association information model as the reference model for deriving and converting the association information model, and derives the association data of the DCI device by analyzing the link and neighbor data in the reference model, so as to construct the DCI association information model.
[0115] Step S403: If the network element device is a P4 device, use the neighbors of the P4 device including P4 devices, DCI devices, and / or Host devices as constraint information to construct the corresponding P4 device association information constraint model.
[0116] Step S404: If the network element device is a DCI device, construct a corresponding DCI device association information constraint model with the neighbors of the DCI device including P4 devices and / or DCI devices and excluding Host devices as constraint information.
[0117] Step S405: Integrate the association information models of various types of network element devices and the corresponding association information constraint models to construct an integrated association information model:
[0118] M_3 = ∪_i = 1^i = j (R_i ∩ Cr_i)
[0119] where M_3 is the integrated association information model, R_i is the association information model of the i-th type of network element device, Cr_i is the association information constraint model of the i-th type of network element device, and j is the type of network element devices in the multi-modal network. The principle of generating the integrated association information model is the same as that of generating the integrated appearance model and will not be elaborated here.
[0120] In one embodiment, constructing the integrated modality recognition model includes: constructing the corresponding appearance model and the corresponding modality recognition model according to the modality recognition information and the corresponding constraint information of various types of network element devices in the multi-modal network, and integrating all types of modality recognition models and the corresponding constraint models to construct the integrated modality recognition model.
[0121] Specifically, as Figure 5 shown, the process of constructing the integrated modality recognition model includes:
[0122] Step S501: Parse the network packets received by various types of network element devices respectively to obtain the corresponding modality identifiers.
[0123] Exemplarily, perform network packet splitting, verification, and parsing on the received packets of various types of network element devices in sequence, then parse the data link layer packets in sequence, and then parse the Ethernet data packets to parse the fields related to the modality identifier. For example, parse the Ethernet type field, where 0x0800 represents the IP modality, 0x8947 represents the geographical modality, etc., so as to obtain the corresponding modality identifier.
[0124] Step S502: Construct the corresponding modality recognition models for various types of network element devices according to their modality identifiers respectively.
[0125] Specifically, construct a P4 device modality recognition model according to the modality identifier of the P4 device, and construct a DCI device modality recognition model according to the modality identifier of the DCI device.
[0126] Step S503: If the type of the network element device is a p4 device, construct a corresponding P4 device modality recognition constraint model with the network element device supporting the recognition of IP, identity, geographical, and content modalities as constraint information.
[0127] Step S504: If the type of the network element device is a DCI device, construct a corresponding DCI device modality recognition constraint model with the network element device's support for identifying optical modalities as the constraint information.
[0128] Step S505: Integrate the modality recognition models and the corresponding modality recognition constraint models of various types of network element devices to generate an integrated modality recognition model:
[0129] M_4 = ∪_i = 1^i = j (D_i ∩ Cd_i)
[0130] where M_4 is the integrated modality recognition model, D_i is the modality recognition model of the i-th type of network element device, Cd_i is the modality recognition constraint model of the i-th type of network element device, and j is the number of types of network element devices in the multimodal network. The principle of generating the integrated modality recognition model is the same as that of generating the integrated appearance model, which will not be elaborated here.
[0131] In one embodiment, constructing the integrated modality control model includes: combining the modality feature model output after applying the integrated modality recognition model with the path selection mechanism of the multimodal network to construct the integrated modality control model.
[0132] Specifically, the path selection mechanism of the multimodal network includes an addressing model and a routing algorithm. Constructing the integrated modality control model includes: adding the addressing model and routing algorithm of the multimodal network to the modality feature model through the multimodal control system to construct the integrated modality control model. Among them, the modality feature model is the output of the multimodal control system based on the specific application of the integrated modality recognition model and is the key information extracted from different modality data.
[0133] Further, integrating the integrated appearance model, the integrated southbound protocol model, the integrated association information model, the integrated modality recognition model, and the integrated modality control model to construct a multimodal network element digital twin model, including:
[0134] M_f = ∪_n = 1^n = 5 (M_n)
[0135] where M_f is the multimodal network element model, the range of n in M_n is 1 to 5, M_1 is the integrated appearance model, M_2 is the integrated southbound protocol model, M_3 is the integrated association information model, M_4 is the integrated modality recognition model, and M_5 is the integrated modality control model.
[0136] Explanatorily, the above formula represents the construction process of M_f of the multimodal network element digital twin model. Its actual meaning is that for each model M_n (from n = 1 to n = 5), perform a union operation on them to obtain the final multimodal network element model M_f.
[0137] As a preferred embodiment, the method further includes mapping the modalities of the multimodal network element digital twin model to the modalities of the physical network element device to obtain a corresponding mapping relationship:
[0138] C_f = f_1(t, p) ∩ f_2(C_o, C_a, C_s)
[0139] Wherein, C_f is the set of modalities supported by the physical device, f_1(t, p) is the maximum set of modalities supported by the multimodal network element in the multimodal network, t is the type of network element device, p is the southbound protocol, f_2(C_o, C_a, C_s) is the set of modalities supported by the multimodal network element in the multimodal network after implementing the modality control strategy in the multimodal network element digital twin model, C_o is the original set of modalities before implementing the control strategy, C_a is the added set of modalities in implementing the control strategy, and C_s is the reduced set of modalities in implementing the control strategy.
[0140] The digital twin modeling method of the multimodal network element provided by the embodiments of the present application constructs digital twin models of multiple dimensions based on information such as the appearance model, southbound protocol, association information, modality recognition, modality control, and constraints of P4 devices and DCI devices, then fuses the twin models of the two types of devices from each dimension respectively, and finally combines each fusion model into a multimodal network element digital twin model, and describes the core content of the modeling through multiple modeling formulas during the modeling process, constructing a high-precision multimodal network element digital twin model, enabling the modality control strategy to be accurately experimentally verified in the multimodal network element digital twin model, obtaining accurate verification results, thereby improving the modality control ability of the modality control strategy for the multimodal network and effectively reducing the risk of directly applying the modality control strategy to physical devices.
[0141] In a second aspect, the embodiments of the present application further provide a digital twin modeling device for a multimodal network element.
[0142] In one embodiment, referring to Figure 6 , Figure 6 is a schematic diagram of the functional modules of an embodiment of the digital twin modeling device for the multimodal network element of the present application. As shown in Figure 6 , the digital twin modeling device for the multimodal network element includes:
[0143] A first construction module, which is used to respectively construct corresponding appearance models, southbound protocol models, association information models, modality recognition models, and corresponding constraint models according to the appearance model information, southbound protocol information, association information, modality recognition information, and corresponding constraint information of various types of network element devices in the multimodal network;
[0144] The first fusion module is used to fuse all types of appearance models, southbound protocol models, association information models, modality recognition models and corresponding constraint models respectively to construct a fused appearance model, a fused southbound protocol model, a fused association information model and a fused modality recognition model;
[0145] The second construction module is used to construct a fused modality control model based on the modality feature model output after the application of the fused modality recognition model and in combination with the path selection mechanism of the multimodal network;
[0146] The second fusion module is used to fuse the fused appearance model, the fused southbound protocol model, the fused association information model, the fused modality recognition model and the fused modality control model to construct a multimodal network element digital twin model.
[0147] Further, in one embodiment, various types of network element devices in the multimodal network include P4 devices and DCI devices.
[0148] Further, in one embodiment, the device is further configured to:
[0149] Construct an appearance model of the corresponding type and a corresponding constraint model according to the appearance model information and the corresponding constraint information of various types of network element devices in the multimodal network, and fuse all types of appearance models and the corresponding constraint models to construct a fused appearance model, including:
[0150] Construct an appearance model of the corresponding type based on the appearance pictures and identification information of each component in various types of network element devices, where the identification information includes the model of the network element device and the type of the component in the network element device, and the components include the chassis, board cards and single disks;
[0151] If the network element device is a frame-type device, use the board card information, port information and optical module information of the network element device as constraint information to construct a corresponding appearance constraint model;
[0152] If the network element device is a non-frame-type device, use the slot information occupied by the network element device, the information of the single disk inserted in each slot, port information and optical module information as constraint information to construct a corresponding appearance constraint model;
[0153] Fuse the appearance models of various types of network element devices and the corresponding appearance constraint models to generate a fused appearance model:
[0154] M_1 = ∪_i = 1^i = j(A_i ∩ Ca_i)
[0155] Where M_1 is the fused appearance model, A_i is the appearance model of the i-th type of network element device, Ca_i is the appearance constraint model of the i-th type of network element device, and j is the type of network element device in the multimodal network.
[0156] Further, in one embodiment, the apparatus is further configured to: construct a southbound protocol model of a corresponding type and a corresponding constraint model according to the southbound protocol information and the corresponding constraint information of various types of network element devices in the multi-modal network, and fuse all types of southbound protocol models and the corresponding constraint models to construct a fused southbound model, including:
[0157] If the southbound protocol of the network element device supports the P4Runtime protocol, construct a corresponding southbound protocol model according to the data model, message passing model, and operation model of the P4Runtime protocol;
[0158] If the southbound protocol of the network element device supports the Netconf protocol, construct a corresponding southbound protocol model according to the content layer model, operation layer model, RPC layer model, and transport layer model of the Netconf protocol;
[0159] If the type of the network element device is a P4 device, construct a corresponding southbound protocol constraint model for the P4 device with the southbound protocol supporting the P4Runtime protocol and not supporting the Netconf protocol as the constraint information;
[0160] If the type of the network element device is a DCI device, construct a corresponding southbound protocol constraint model for the DCI device with the southbound protocol supporting the Netconf protocol and not supporting the P4Runtime protocol as the constraint information;
[0161] Fuse the southbound protocol models of various types of network element devices and the corresponding southbound protocol constraint models to construct a fused southbound protocol model:
[0162] M_2 = ∪_i = 1^i = j (P_i ∩ Cp_i)
[0163] where M_2 is the fused southbound protocol model, P_i is the southbound protocol model of the i-th type of network element device, Cp_i is the southbound protocol constraint model of the i-th type of network element device, and j is the type of network element device in the multi-modal network.
[0164] Further, in one embodiment, the apparatus is further configured to: construct a corresponding association information model and a corresponding constraint model according to the association information and the corresponding constraint information of various types of network element devices in the multi-modal network, and fuse all types of association models and the corresponding constraint models to construct a fused association information model, including:
[0165] Establish a P4 device association information model according to the link information and neighbor information of the multi-modal network;
[0166] Derive and construct a DCI device association information model according to the P4 device association information model;
[0167] If the network element device is a P4 device, use the neighbors of the P4 device, including the P4 device, the DCI device, and / or the Host device, as constraint information to construct a corresponding P4 device association information constraint model;
[0168] If the network element device is a DCI device, use the neighbors of the DCI device, including the P4 device and / or the DCI device, excluding the Host device, as constraint information to construct a corresponding DCI device association information constraint model;
[0169] Fuse the association information models of various types of network element devices and the corresponding association information constraint models to construct a fused association information model:
[0170] M_3 = ∪_i = 1^i = j (R_i ∩ Cr_i)
[0171] Where M_3 is the fused association information model, R_i is the association information model of the i-th type of network element device, Cr_i is the association information constraint model of the i-th type of network element device, and j is the type of network element device in the multimodal network.
[0172] Further, in one embodiment, the apparatus is further configured to: construct a corresponding appearance model and a corresponding modality recognition model according to the modality recognition information and the corresponding constraint information of various types of network element devices in the multimodal network, and fuse all types of modality recognition models and the corresponding constraint models to construct a fused modality recognition model, including:
[0173] Parse the network packets received by various types of network element devices respectively to obtain corresponding modality identifiers;
[0174] Construct corresponding modality recognition models for various types of network element devices according to the modality identifiers of various types of network element devices;
[0175] If the type of the network element device is a p4 device, use the network element device supporting the recognition of IP, identity, geography, and content modalities as constraint information to construct a corresponding P4 device modality recognition constraint model;
[0176] If the type of the network element device is a DCI device, use the network element device supporting the recognition of optical modality as constraint information to construct a corresponding DCI device modality recognition constraint model;
[0177] Fuse the modality recognition models of various types of network element devices and the corresponding modality recognition constraint models to generate a fused modality recognition model:
[0178] M_4 = ∪_i = 1^i = j (D_i ∩ Cd_i)
[0179] Among them, M_4 is the fusion modality recognition model, D_i is the modality recognition model of the i-th type of network element device, Cd_i is the modality recognition constraint model of the i-th type of network element device, and j is the type of network element device in the multimodal network.
[0180] Further, in one embodiment, the device is further configured to: based on the modality feature model output after applying the fusion modality recognition model, combine with the path selection mechanism of the multimodal network to construct a fusion modality control model, including:
[0181] Add the addressing model and routing algorithm of the multimodal network to the modality feature model to construct the fusion modality control model.
[0182] Further, in one embodiment, the device is further configured to: fuse the fusion appearance model, the fusion southbound protocol model, the fusion association information model, the fusion modality recognition model, and the fusion modality control model to construct a multimodal network element digital twin model, including:
[0183] M_f = ∪_n = 1^n = 5(M_n)
[0184] Among them, M_f is the multimodal network element model, the range of n in M_n is 1 to 5, M_1 is the fusion appearance model, M_2 is the fusion southbound protocol model, M_3 is the fusion association information model, M_4 is the fusion modality recognition model, and M_5 is the fusion modality control model.
[0185] Further, in one embodiment, the device is further configured to: the method further includes:
[0186] Map the modality of the multimodal network element digital twin model to the modality of the physical network element device to obtain the corresponding mapping relationship:
[0187] C_f = f_1(t,p) ∩ f_2(C_o,C_a,C_s)
[0188] Among them, C_f is the set of modalities supported by the physical device, f_1(t,p) is the maximum set of modalities supported by the multimodal network element in the multimodal network, t is the network element device type, p is the southbound protocol, f_2(C_o,C_a,C_s) is the set of modalities supported by the multimodal network element in the multimodal network after implementing the modality control strategy in the multimodal network element digital twin model, C_o is the original set of modalities before implementing the control strategy, C_a is the set of modalities added during the implementation of the control strategy, and C_s is the set of modalities reduced during the implementation of the control strategy.
[0189] Among them, the function implementation of each module in the above multimodal network element digital twin modeling device corresponds to each step in the above multimodal network element digital twin modeling method embodiment, and its function and implementation process will not be elaborated here one by one.
[0190] It should be noted that the serial numbers of the embodiments of the present application above are only for description and do not represent the superiority or inferiority of the embodiments.
[0191] In the description of the embodiments of the present application, the terms "including" and "having" and any variations thereof in the specification, claims and drawings of the present application are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally further include steps or units not listed, or may optionally further include other steps or units inherent to these processes, methods, products or devices. The descriptions of the terms "first", "second", "third", etc. are used to distinguish different objects, etc., and do not represent a sequence, nor do they limit that "first", "second" and "third" are of different types.
[0192] In the description of the embodiments of the present application, "exemplary", "for example" or "for instance" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "exemplary", "for example" or "for instance" in the embodiments of the present application should not be construed as being more preferred or having more advantages than other embodiments or designs. Rather, the use of the words "exemplary", "for example" or "for instance" is intended to present the relevant concepts in a specific manner.
[0193] In the description of the embodiments of the present application, unless otherwise specified, " / " means "or". For example, A / B may mean A or B; "and / or" in the text is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B may mean: A exists alone, A and B exist simultaneously, and B exists alone. In addition, in the description of the embodiments of the present application, "a plurality of" means two or more than two.
[0194] In some processes described in the embodiments of the present application, there are multiple operations or steps that appear in a specific order. However, it should be understood that these operations or steps may not be executed in the order in which they appear in the embodiments of the present application or may be executed in parallel. The serial numbers of the operations are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations or steps may be executed in order or in parallel, and these operations or steps may be combined.
[0195] Through the description of the above embodiments, those skilled in the art can clearly understand that the above-described embodiment methods can be implemented by means of software plus a necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above and includes several instructions for causing a terminal device to execute the methods described in the various embodiments of the present application.
[0196] The above are only the preferred embodiments of the present application, and do not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made by using the content of the specification and drawings of the present application, or directly or indirectly applied in other related technical fields, shall be equally included in the patent protection scope of the present application.
Claims
1. A digital twin modeling method for a multimodal network element, characterized in that: The digital twin modeling method of the multimodal network element includes: According to the appearance model information, southbound protocol information, association information, modal identification information and corresponding constraint information of various types of network element devices in the multimodal network, the appearance model, southbound protocol model, association information model, modal identification model and corresponding constraint model of the corresponding type are respectively constructed; All types of appearance models, southbound protocol models, associated information models, modal recognition models and corresponding constraint models are fused to construct fused appearance models, fused southbound protocol models, fused associated information models and fused modal recognition models. Based on the modal feature model output after the fusion modal recognition model is applied, combined with the path selection mechanism of the multimodal network, a fusion modal control model is constructed; The fused appearance model, the fused southbound protocol model, the fused association information model, the fused modal identification model and the fused modal control model are integrated to construct a multimodal network element digital twin model.
2. The digital twin modeling method of a multimodal network element according to claim 1, characterized in that: Various types of network element devices in the multimodal network include P4 devices and DCI devices.
3. The digital twin modeling method of a multimodal network element according to claim 2, characterized in that: According to the appearance model information and corresponding constraint information of various types of network element devices in the multimodal network, corresponding types of appearance models and corresponding constraint models are constructed, and all types of appearance models and corresponding constraint models are fused to construct a fused appearance model, including: Building appearance models of corresponding types based on appearance pictures and identification information of various components in various types of network element devices, wherein the identification information includes the model of the network element device and the type of components in the network element device, and the components include a chassis, a board, and a single disk; If the network element device is a frame-type device, the board information, port information and optical module information of the network element device are used as constraint information to build a corresponding appearance constraint model; If the network element device is a non-frame type device, the slot information occupied by the network element device, the information of the single disk inserted in each slot, the port information and the optical module information are used as constraint information to build the corresponding appearance constraint model; The appearance models of various types of network element devices and the corresponding appearance constraint models are merged to generate a fused appearance model: M_1=∪_i=1^i=j(A_i∩Ca_i) Among them, M_1 is the fused appearance model, A_i is the appearance model of the i-th type of network element device, Ca_i is the appearance constraint model of the i-th type of network element device, and j is the type of network element device in the multimodal network.
4. The digital twin modeling method of a multimodal network element according to claim 2, characterized in that: According to the southbound protocol information and corresponding constraint information of various types of network element devices in the multimodal network, corresponding types of southbound protocol models and corresponding constraint models are constructed, and all types of southbound protocol models and corresponding constraint models are integrated to construct an integrated southbound model, including: If the southbound protocol of the network element device supports the P4Runtime protocol, a corresponding southbound protocol model is constructed according to the data model, message transmission model, and operation model of the P4Runtime protocol; If the southbound protocol of the network element device supports the Netconf protocol, the corresponding southbound protocol model is constructed according to the content layer model, operation layer model, RPC layer model and transport layer model of the Netconf protocol; If the type of the network element device is a P4 device, the southbound protocol supports the P4Runtime protocol and does not support the Netconf protocol as constraint information, and the corresponding P4 device southbound protocol constraint model is constructed; If the type of the network element device is a DCI device, the southbound protocol supports the Netconf protocol and does not support the P4Runtime protocol as constraint information, and a corresponding DCI device southbound protocol constraint model is constructed; The southbound protocol models of various types of network element devices and the corresponding southbound protocol constraint models are integrated to build a converged southbound protocol model: M_2=∪_i=1^i=j(P_i∩Cp_i) Among them, M_2 is the converged southbound protocol model, P_i is the southbound protocol model of the i-th type of network element device, Cp_i is the southbound protocol constraint model of the i-th type of network element device, and j is the type of network element device in the multimodal network.
5. The digital twin modeling method of a multimodal network element according to claim 2, characterized in that: According to the association information and corresponding constraint information of various types of network element devices in the multimodal network, corresponding types of association information models and corresponding constraint models are constructed, and all types of association models and corresponding constraint models are integrated to construct a fused association information model, including: Establish a P4 device association information model based on the link information and neighbor information of the multimodal network; Derivation and construction of a DCI device association information model based on the P4 device association information model; If the network element device is a P4 device, the neighbors of the P4 device include the P4 device, the DCI device and / or the Host device as constraint information to construct a corresponding P4 device association information constraint model; If the network element device is a DCI device, the neighbors of the DCI device include P4 devices and / or DCI devices, but do not include Host devices, as constraint information, and a corresponding DCI device association information constraint model is constructed; The associated information models of various types of network element devices and the corresponding associated information constraint models are integrated to construct an integrated associated information model: M_3=∪_i=1^i=j(R_i∩Cr_i) Among them, M_3 is the fusion association information model, R_i is the association information model of the i-th type of network element equipment, Cr_i is the association information constraint model of the i-th type of network element equipment, and j is the type of network element equipment in the multimodal network.
6. The digital twin modeling method of a multimodal network element according to claim 2, characterized in that: According to the modal recognition information and corresponding constraint information of various types of network element devices in the multimodal network, the corresponding type of appearance model and the corresponding modal recognition model are constructed, and all types of modal recognition models and corresponding constraint models are fused to construct a fused modal recognition model, including: Analyze the network messages received by various types of network element devices respectively to obtain corresponding mode identifiers; According to the modal identification of various types of network element equipment, corresponding types of modal recognition models are constructed respectively; If the type of the network element device is a p4 device, the network element device supports identification of IP, identity, geography, and content modality as constraint information to construct a corresponding P4 device modality identification constraint model; If the type of the network element device is a DCI device, the optical mode recognition supported by the network element device is used as constraint information to construct a corresponding DCI device mode recognition constraint model; The modal recognition models of various types of network element devices and the corresponding modal recognition constraint models are integrated to generate a fused modal recognition model: M_4=∪_i=1^i=j(D_i∩Cd_i) Among them, M_4 is the fusion modal recognition model, D_i is the modal recognition model of the i-th type of network element equipment, Cd_i is the modal recognition constraint model of the i-th type of network element equipment, and j is the type of network element equipment in the multimodal network.
7. The digital twin modeling method of a multimodal network element according to claim 2, characterized in that: The modal feature model output after the fusion modal recognition model is applied is combined with the path selection mechanism of the multimodal network to construct a fusion modal control model, including: The addressing model and routing algorithm of the multimodal network are added to the modal feature model to construct the fusion modal control model.
8. The digital twin modeling method of a multimodal network element according to claim 1, characterized in that: The fused appearance model, the fused southbound protocol model, the fused association information model, the fused modal identification model and the fused modal control model are integrated to construct a multimodal network element digital twin model, including: M_f=∪_n=1^n=5(M_n) Among them, M_f is a multimodal network element model, n in M_n ranges from 1 to 5, M_1 is a fused appearance model, M_2 is a fused southbound protocol model, M_3 is a fused association information model, M_4 is a fused modal identification model, and M_5 is a fused modal control model.
9. The digital twin modeling method of a multimodal network element according to claim 1, characterized in that: The method further includes: Map the modes of the multimodal network element digital twin model to the modes of the physical network element device to obtain the corresponding mapping relationship: C_f=f_1(t,p)∩f_2(C_o,C_a,C_s) Among them, C_f is the modal set supported by the physical device, f_1(t,p) is the maximum modal set supported by the multimodal network element in the multimodal network, t is the network element device type, p is the southbound protocol, f_2(C_o,C_a,C_s) is the modal set supported by the multimodal network element in the multimodal network after the modal control strategy is implemented in the digital twin model of the multimodal network element, C_o is the original modal set before the implementation of the control strategy, C_a is the modal set added in the implementation of the control strategy, and C_s is the modal set reduced in the implementation of the control strategy.
10. A digital twin modeling device for a multimodal network element, characterized in that: The digital twin modeling device of the multimodal network element includes: A first construction module, which is used to construct the appearance model, southbound protocol model, association information model, modal identification model and corresponding constraint model of the corresponding type according to the appearance model information, southbound protocol information, association information, modal identification information and corresponding constraint information of various types of network element devices in the multimodal network; The first fusion module is used to fuse all types of appearance models, southbound protocol models, associated information models, modal recognition models and corresponding constraint models respectively to construct a fused appearance model, a fused southbound protocol model, a fused associated information model and a fused modal recognition model; A second construction module is used to construct a fusion modal control model based on the modal feature model output after the fusion modal recognition model is applied and combined with a path selection mechanism of a multimodal network; The second fusion module is used to fuse the fused appearance model, the fused southbound protocol model, the fused association information model, the fused modal identification model and the fused modal control model to construct a multimodal network element digital twin model.