WO2024011908A1 - 网络预测系统、方法、电子设备及存储介质 - Google Patents

网络预测系统、方法、电子设备及存储介质 Download PDF

Info

Publication number
WO2024011908A1
WO2024011908A1 PCT/CN2023/077491 CN2023077491W WO2024011908A1 WO 2024011908 A1 WO2024011908 A1 WO 2024011908A1 CN 2023077491 W CN2023077491 W CN 2023077491W WO 2024011908 A1 WO2024011908 A1 WO 2024011908A1
Authority
WO
WIPO (PCT)
Prior art keywords
network
information
model
prediction
digital twin
Prior art date
Application number
PCT/CN2023/077491
Other languages
English (en)
French (fr)
Inventor
黄卓垚
谭云生
Original Assignee
中兴通讯股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 中兴通讯股份有限公司 filed Critical 中兴通讯股份有限公司
Publication of WO2024011908A1 publication Critical patent/WO2024011908A1/zh

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W16/00—Network planning, e.g. coverage or traffic planning tools; Network deployment, e.g. resource partitioning or cells structures
    • H04W16/22—Traffic simulation tools or models
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W24/00—Supervisory, monitoring or testing arrangements
    • H04W24/06—Testing, supervising or monitoring using simulated traffic
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/50—Service provisioning or reconfiguring

Definitions

  • the present disclosure relates to the field of communications, and in particular to a network prediction system, method, electronic device and storage medium.
  • embodiments of the present disclosure provide a network prediction system including a digital twin and a network management controller.
  • the digital twin is composed of multiple models, each of which runs independently.
  • the digital twin is used to simulate the network performance of the communication network based on the network configuration information of the communication network.
  • the network management controller is used to provide network configuration information to the digital twin, obtain simulated network performance data from the digital twin, and make configuration changes to the physical network of the communication network based on the network performance data.
  • embodiments of the present disclosure also provide a network prediction method, which includes a digital twin acquiring network configuration information of a communication network from a network management controller.
  • the digital twin is composed of multiple models, and each of the multiple models runs independently;
  • the digital twin simulates the network performance of the communication network based on the network configuration information;
  • the network management controller obtains the simulated network performance data from the digital twin, and changes the configuration of the physical network of the communication network based on the network performance data.
  • an embodiment of the present disclosure also provides an electronic device, including at least one processor and a memory communicatively connected to the at least one processor.
  • the memory stores instructions that can be executed by at least one processor, and the instructions are executed by at least one processor, so that at least one processor can execute the above network prediction method.
  • embodiments of the present disclosure also provide a computer-readable storage medium that stores a computer program.
  • the computer program is executed by a processor, the above-mentioned network prediction method is implemented.
  • Figure 1 is a schematic structural diagram of a network prediction system according to some embodiments.
  • Figure 2 is a schematic diagram of the transmission relationship between the network prediction system and the physical network according to some embodiments
  • Figure 3 is an information diagram of a network prediction system according to some embodiments.
  • Figure 4 is a schematic structural diagram of another network prediction system according to some embodiments.
  • Figure 5 is a schematic structural diagram of yet another network prediction system according to some embodiments.
  • Figure 6 is a schematic diagram of the transmission relationship of each model in the network prediction system according to some embodiments.
  • Figure 7 is a flow chart of a network prediction method according to some embodiments.
  • Figure 8 is a schematic structural diagram of an electronic device according to some embodiments.
  • first and second are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the quantity of indicated technical features.
  • the network prediction system 10 includes a digital twin 101 and a network management controller 102 .
  • the digital twin 101 is composed of multiple models, and each of the multiple models operates independently.
  • the digital twin 101 is used to simulate the network performance of the communication network based on the network configuration information of the communication network.
  • a digital twin 101 is a virtual model used to accurately reflect the performance of a physical object.
  • the network management controller 102 is used to provide network configuration information to the digital twin 101, obtain simulated network performance data from the digital twin 101, and make configuration changes to the physical network of the communication network based on the network performance data.
  • the network prediction system 10 of the embodiment of the present disclosure can be applied to any communication network, such as optical transmission network, telephone switching network, etc.
  • the network prediction system 10 may serve as a digital twin model system of the physical network of the communication network.
  • each model can be a model, or a total model composed of multiple small models.
  • the technology used in each model can vary.
  • the model can use graph neural network, discrete event or recurrent neural network.
  • the physical objects simulated by each model can also be different.
  • the model can simulate the performance of the optical module, the stability of the power supply, or the delay of network data transmission.
  • Each model can run in a different environment, or each model can run in the same environment.
  • models can run in virtual machines, in cloud environments, or on physical machines and other electronic devices.
  • the input information obtained by each model can come from the physical network, other models, or from the network management controller 102.
  • the output information of the model can be transmitted to other models, or can be fed back to the network management controller 102 as a simulation result.
  • the network management controller 102 makes a decision, the network configuration is initiated to affect the operation of the physical network elements.
  • the physical object studied by Digital Twin 101 will be equipped with various sensors related to important functional areas. These sensors output data related to different aspects of the performance of the physical object, such as energy output, temperature, weather conditions, etc. This data is then forwarded to a processing system and applied to the digital copy. Once this data is obtained, virtual models can be used to run simulations, study performance issues, and generate possible improvements, ultimately yielding valuable insights. Valuable insights can in turn be applied to the original physical objects.
  • a digital twin network platform can be built. Through real-time interaction and mutual influence between the physical network and the digital twin network, the digital twin network platform can help the network achieve low-cost trial and error, intelligent decision-making and high-efficiency innovation.
  • the network prediction system of the embodiment of the present disclosure can predict various aspects of the entire communication network, so that the functions and performance of the entire communication network can be predicted from multiple dimensions. Perform predictive simulations.
  • different model structures can be selected for the digital twin according to different simulation prediction requirements. Therefore, the prediction system of the embodiment of the present disclosure has strong reusability and can be adapted to various communication networks, even if the network scale is expanded. , the complexity of the network increases, and the performance of the network can also be systematically simulated and predicted.
  • the network management controller provides network configuration information to the digital twin, obtains prediction data from the digital twin, and makes configuration changes to the physical network of the communication network based on the prediction data, so that the network management controller can operate between the physical network and the digital twin.
  • Information is transferred between devices, so that the physical network can be changed based on the network performance data predicted by the digital twin, so that the transmission performance of the physical network can be improved and simulation applications can be realized.
  • Each model in the digital twin 101 can have different trigger conditions. Only when the preset trigger conditions are met, the model will be simulated. Multiple models in the digital twin 101 can be selected according to the network functions or performance targeted by the network prediction system, and the type and number of models can be changed accordingly as needed. For example, for network performance prediction simulation, you can choose three different models; for network fault detection simulation, you can choose four different models. Models; 2 of each model. Simulation is realized through multi-model collaboration.
  • the network management controller 102 may be one device, or a controller composed of multiple devices.
  • the network management controller 102 may also be a device used by a network management control system. Therefore, the network management controller 102 is also called a network management control system.
  • the network management controller 102 and the digital twin 101 are deployed on the K8S cloud.
  • Each model of the digital twin 101 and the interactive information between the digital twin 101 and the network management controller 102 can be transmitted through messages.
  • the message transmission method can be any one of subscription, pull, and push.
  • Subscription means that the model subscribes to information in advance from the message source, and regularly obtains input messages carrying information sent by the message source (i.e., subscription messages); pull means that the model sends information query requests to the message source, and obtains input messages carrying information (i.e., response messages) );
  • Push means that the message source actively sends input messages carrying information to the model (i.e. push messages).
  • the transmission relationship between each model in the digital twin 101, the network management control system 104, and the physical network may be as shown in FIG. 2 .
  • the network prediction system 10 also includes an information controller 103 , which is configured to issue corresponding input configuration information to each model of the digital twin 101 .
  • Input configuration information is used to configure the model's input message acquisition method and data type.
  • Each model of the digital twin 101, the physical network, and the network management control system 104 are all sources or sinks of information, and they are collectively referred to as "services.”
  • the information controller 103 can define the information input and output relationships, information acquisition methods and information data types between each "service", and control the message collaboration between each "service”. For example, the information controller 103 issues the input configuration information based on the information map.
  • the infographic is a customized diagram tailored to the network functions predicted by the network prediction system 10 and the model selection of multiple models to indicate how the multiple models work together.
  • the model in the digital twin 101 changes, for example, a new model is registered, or an existing model is uninstalled
  • the information controller 103 issues the input configuration information based on the changed information map.
  • the network prediction system can form information graphs in the form of charts, so that information transmission relationships can be established between each "service” and the message collaboration between each "service” can be controlled. As shown in Figure 3, Model A to Model F in Figure 2 correspond to Service A to Service F respectively in the information diagram, the network management control system in Figure 2 corresponds to Service G, and the physical network in Figure 2 corresponds to Service H.
  • the information controller issues corresponding input configuration information to each model of the digital twin, and configures the input message acquisition method and data type of each model. Even if there are format differences between different models, The information controller can be used to coordinate information transmission, realize interaction and collaboration within the network prediction system, improve the compatibility of the network prediction system with the model, and increase the reusability of the network prediction system.
  • the information controller 103 can be set on an electronic device outside the network management control system 104 to control the flow of information and deliver configuration information to each model.
  • Configuration information may include input configuration information and output configuration information, as well as other configuration information.
  • the information controller can read the information graph, translate it into configuration information in each "service", and configure it to each "service”.
  • the information controller can define the configuration information configured to each "service” as follows.
  • the input message acquisition method includes any one or any combination of the following: subscription, pull, and push.
  • Each model performs simulation according to the input message and then outputs the simulation result as the output message.
  • Each input message and output message includes a timestamp that indicates when the message was sent.
  • each of the multiple models of the digital twin 101 starts running when a trigger condition for the model is met, which includes the model acquiring all required input messages.
  • the third timestamps carried by all required input messages are greater than the difference between the prediction time of the network prediction system and the simulation step size of the model, and the current time is earlier than the prediction time. That is, each model of the digital twin 101 needs to obtain all required input messages within one step time before the prediction time. In other words, after sufficient information required for simulation is obtained, the model is triggered to run and the simulation is performed.
  • digital twin 101 includes a traffic prediction model, a device degradation model, and a network performance model.
  • the traffic prediction model is used to obtain the traffic information of the communication network and generate traffic data at the predicted time based on the traffic information;
  • the equipment degradation model is used to obtain the equipment status information of the communication network and generate the link status at the predicted time based on the equipment status information.
  • the network performance model is used to obtain traffic data, link status data, and network configuration information of the communication network, and generate network performance data at the predicted time based on the traffic data, link status data, and network configuration information.
  • the traffic prediction model can be connected to the physical network and network performance model, obtain traffic information from the physical network, and transmit the traffic data generated by prediction to the network performance model.
  • Traffic information can be all traffic data generated by each network element in the network within a period of time, or all traffic data generated by each domain in the network within a period of time, etc.
  • the traffic prediction model can predict the traffic data at the prediction time based on the traffic information.
  • the predicted time is a preset time.
  • the predicted traffic data belongs to the same device as the obtained traffic information. For example, if the obtained traffic information comes from network element 1, the traffic data predicted based on the traffic information of network element 1 is the traffic size of network element 1 at the predicted time.
  • the network configuration information may be topology information of the network.
  • the device degradation model can be connected to the physical network and network performance model, obtain device status information from the physical network, and transmit the predicted link status data to the network performance model.
  • the device status information can be the device status information of each network element in the network within a period of time, such as current life, current, temperature, etc.
  • equipment degradation models can have corresponding model designs. For example, equipment degradation models suitable for telephone switching networks and equipment degradation models suitable for optical fiber networks.
  • the equipment degradation model can predict the link status data at the prediction time based on the equipment status information. For example, it can predict the transmission loss of the link based on the equipment status information of each network element in the link or transmission rate, etc.
  • the network performance model can be connected with the traffic prediction model, equipment degradation model and physical network, obtain network configuration information from the physical network, obtain traffic data from the traffic prediction model, obtain link status data from the equipment degradation model, and predict network performance at the moment Data predictions.
  • the network configuration information may be routing configuration information, division information of different domains in the network, etc.
  • Network performance data can be network delay, jitter, packet loss rate and other performance data.
  • the traffic prediction model can predict the traffic and obtain the traffic size at the predicted time
  • the equipment degradation model can predict the link status
  • the network performance model can be based on the traffic size, link status and network configuration. Predict network performance. That is, the system can use different models to predict the traffic and link status of the entire communication network, and then predict network performance based on the prediction results of these two aspects, so that it can predict network performance from multiple dimensions. Predict the performance of the entire communication network.
  • the communication network may be an optical transport network, in which case the device status information may be optical module status data, and the device degradation model may use an optical module degradation model corresponding to the optical transport network.
  • the time when the traffic prediction model obtains the traffic information is earlier than the prediction time, and the time interval between the time when the traffic prediction model obtains the traffic information and the prediction time is less than or equal to the first preset simulation step. That is, the acquisition time of the traffic information used for predicting the prediction time is relatively close to the prediction time. Therefore, the predicted traffic data is more accurate.
  • the time when the equipment degradation model obtains the equipment status information is earlier than the prediction time, and the time interval between the time when the equipment degradation model obtains the equipment status information and the prediction time is less than or equal to the second preset simulation step. That is, the acquisition time of the equipment status information used for predicting the prediction time is relatively close to the prediction time. Therefore, the predicted link status data is more accurate, and thus more accurate network performance data can be obtained.
  • the values of the first preset simulation step size and the second preset simulation step size can be preset according to requirements, and they can be the same or different. For either of the first preset simulation step size and the second preset simulation step size, different values may be preset according to different types of communication networks.
  • each model in the digital twin 101 can obtain the information and data they need through input messages.
  • the traffic prediction model is also used to obtain traffic information from the input message of the traffic prediction model. Before generating traffic data at the prediction time based on the traffic information, verify that the first timestamp carried by the input message of the traffic prediction model is greater than the prediction time and the third time. The difference between a preset simulation step size, and the first timestamp is earlier than the predicted time.
  • the device degradation model is also used to obtain device status information from the input message of the device degradation model.
  • the traffic prediction model and the device degradation model pass the verification.
  • the message timestamp carried in the respective input message can be used to verify the acquisition time of the traffic information and device status information before generating the traffic data at the predicted time and the link status data at the predicted time, ensuring the network performance data obtained by simulation. accuracy.
  • the input information of model N is the data at time T-t, and the output information is the data at time T inferred after simulation. Then the simulation step size of model N is t.
  • model N starts the simulation process at time T: 1.
  • the timestamps of all messages carrying input information in model N are greater than time T-t; 2.
  • time T is the moment closest to the present.
  • the network management control system 104 and the digital twin 101 are both deployed on the K8S cloud.
  • the digital twin 101 consists of a traffic prediction model, an optical module degradation model, and a network performance model. These models are It is a microservice running in K8S.
  • the information controller 103 is built into the network management control system 104. It reads the information graph, decomposes the input information into network performance model, traffic prediction model, and optical module degradation model and configures it as follows, and sends it to the microservice where these three models are located. .
  • the configuration of the traffic prediction model is as follows:
  • the optical module degradation model After receiving the configuration information of the information controller, the optical module degradation model completes the subscription of input messages such as current life, current, and temperature.
  • the network performance model is configured as follows:
  • the network performance model After receiving the configuration information sent by the information controller, the network performance model completes the microservice settings for pulling in traffic and link status input information.
  • the information transmission relationship of each model in the digital twin 101 can be shown in Figure 6.
  • the flow forecast model obtains flow information by subscribing to the flow data agent service (flowData Agent).
  • the flow data agent service It can be a service in the physical network; the network performance model (networkPerformance Model) pulls traffic data from the traffic prediction model by calling the preset data interface.
  • the optical fiber interface degradation model (optAgeing Model) obtains the status information of the optical fiber equipment by subscribing to the optical fiber interface data agent service (optData Agent).
  • the optical fiber interface data agent service can be a service in the physical network; the network performance model (networkPerformance Model) obtains the status information of the optical fiber equipment by calling the preset
  • the data interface is configured to pull link status data from the fiber interface degradation model.
  • An embodiment of the present disclosure provides a network prediction method. As shown in Figure 7, the network prediction method includes steps S701 to S703.
  • step S701 the digital twin 101 obtains the network configuration information of the communication network from the network management controller 102; the digital twin is composed of multiple models, and each of the multiple models runs independently.
  • Step S702 The digital twin 101 performs simulation predictions on the communication network based on the network configuration information.
  • step S703 the network management controller 102 obtains the prediction data obtained from the simulation from the digital twin 101, and changes the configuration of the physical network of the communication network based on the prediction data.
  • the network prediction system that implements the network prediction method includes, in addition to the digital twin 101 and the network management controller 102, an information controller 103. Before step S702, the method also includes the information controller 103 delivering corresponding input configuration information to each model of the digital twin 101 respectively. Input configuration information is used to configure the model's input message acquisition method and data type.
  • the information controller 103 issues the input configuration information based on the information map.
  • Infographics are custom charts that are customized for network functions predicted by a network prediction system and model selection for multiple models to indicate how multiple models work together.
  • the information controller issues the input configuration information based on the changed information map.
  • each of the multiple models in the digital twin 101 starts running when a trigger condition for the model is met, which includes the model acquiring all required input messages.
  • the third timestamps carried by all required input messages are greater than the difference between the prediction time of the network prediction system and the simulation step size of the model, and the current time is earlier than the prediction time.
  • digital twin 101 includes a traffic prediction model, a device degradation model, and a network performance model.
  • Step S702 includes the following.
  • the traffic prediction model obtains the traffic information of the communication network and generates traffic data at the predicted time based on the traffic information; the equipment degradation model obtains the equipment status information of the communication network and generates link status data at the predicted time based on the equipment status information; network performance The model obtains traffic data, link status data, and network configuration information of the communication network, and generates network performance data at the predicted time based on the traffic data, link status data, and network configuration information.
  • the network prediction system implementing the network prediction method includes a communication network:
  • Traffic prediction model The simulation step size is t.
  • the subscribed traffic data information is received at time T-t, triggers the simulation process, and generates traffic data at time T.
  • Optical module degradation model The simulation step size is t.
  • the subscribed optical module status data is received at time T-t, including the current life, current, temperature, etc., and the simulation process is triggered to generate the optical module status data at time T.
  • Network performance model The simulation step size is 0. At time T, the route at time T pushed by the network management control system is received, triggering the information pull process. The traffic data at time T is pulled from the traffic prediction model and pulled from the optical module degradation model. Obtaining the optical module status data at time T satisfies the acquisition of all input information at time T, triggers the simulation process, and generates time T Momentary network delay, jitter, and packet loss performance data are pushed to the network management control system.
  • the network management control system determines that the requirements are met and delivers the corresponding routing configuration to the physical network.
  • this embodiment is a system embodiment corresponding to the above-mentioned embodiment, and this embodiment can be implemented in cooperation with the above-mentioned embodiment.
  • the relevant technical details mentioned in the above embodiment are still valid in this embodiment, and will not be described again in order to reduce duplication.
  • the relevant technical details mentioned in this embodiment can also be applied to the above embodiments.
  • the embodiment of the present disclosure also relates to an electronic device.
  • the electronic device includes at least one processor 801 and a memory 802 communicatively connected with the at least one processor.
  • the memory 802 stores instructions that can be executed by at least one processor 801, and the instructions are executed by at least one processor 801 in the above method embodiment.
  • the memory 802 and the processor 801 are connected using a bus.
  • the bus may include any number of interconnected buses and bridges.
  • the bus connects various circuits of one or more processors 801 and the memory 802 together.
  • the bus may also connect various other circuits together, such as peripherals, voltage regulators, power management circuits, etc., which are not further described in this disclosure.
  • the bus interface provides the interface between the bus and the transceiver.
  • a transceiver may be one element or may include multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium.
  • the information processed by the processor 801 is transmitted on the wireless medium through the antenna.
  • the antenna can also receive the information transmitted on the wireless medium and transmit the information to the processor 801.
  • Processor 801 is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions.
  • Memory 802 may be used to store information used by the processor in performing operations.
  • Embodiments of the present disclosure also relate to a non-transitory computer-readable storage medium storing a computer program.
  • the above method embodiments are implemented when the computer program is executed by the processor.
  • the program is stored in a storage medium and includes a number of instructions to cause a device (which may be a microcontroller, a chip, etc.) or a processor to execute all or part of the steps of the method described in the embodiments of the present disclosure.
  • the aforementioned storage media include: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disk or optical disk and other media that can store program code. .

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)

Abstract

一种网络预测系统，包括数字孪生体和网络管理控制器。数字孪生体由多个模型构成，多个模型各自独立运行，用于基于通信网络的网络配置信息，对通信网络的网络性能进行仿真；网络管理控制器用于向数字孪生体提供网络配置信息，从数字孪生体获取仿真得到的预测数据，并根据预测数据，对通信网络的物理网络进行配置变更。

Description

网络预测系统、方法、电子设备及存储介质
本申请要求申请号为202210836171.2、2022年7月15日提交的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本公开涉及通信领域，特别涉及一种网络预测系统、方法、电子设备及存储介质。
背景技术
随着5G(5th generation)通信技术、物联网和云计算技术的发展以及层出不穷的网络新业务的涌现,网络负载不断增加,网络规模持续扩大。由此带来的网络复杂性使得网络的运行和维护变得越来越复杂。同时,由于网络运营的高可靠性要求,网络故障的高代价以及昂贵的试验成本,网络的变动往往牵一发而动全身，新技术的部署愈发困难。
发明内容
一方面，本公开实施例提供了一种网络预测系统，包括数字孪生体和网络管理控制器。数字孪生体由多个模型构成，多个模型各自独立运行，数字孪生体用于基于通信网络的网络配置信息，对通信网络的网络性能进行仿真。网络管理控制器用于向数字孪生体提供网络配置信息，从数字孪生体获取仿真得到的网络性能数据，并根据网络性能数据，对通信网络的物理网络进行配置变更。
另一方面，本公开实施例还提供了一种网络预测方法，包括数字孪生体从网络管理控制器获取通信网络的网络配置信息，数字孪生体由多个模型构成，多个模型各自独立运行；数字孪生体基于网络配置信息，对通信网络的网络性能进行仿真；网络管理控制器从数字孪生体获取仿真得到的网络性能数据，并根据网络性能数据，对通信网络的物理网络进行配置变更。
再一方面，本公开实施例还提供了一种电子设备，包括至少一个处理器以及与至少一个处理器通信连接的存储器。其中，存储器存储有可被至少一个处理器执行的指令，指令被至少一个处理器执行，以使至少一个处理器能够执行上述网络预测方法。
又一方面，本公开实施例还提供了一种计算机可读存储介质，存储有计算机程序，计算机程序被处理器执行时实现上述网络预测方法。
附图说明
为了更清楚地说明本公开中的技术方案，下面将对本公开一些实施例中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本公开的一些实施例的附图，对于本领域普通技术人员来讲，还可以根据这些附图获得其他的附图。此外，以下描述中的附图可以视作示意图，并非对本公开实施例所涉及的产品的实际尺寸、方法的实际流程、信号的实际时序等的限制。
图1为根据一些实施例的一种网络预测系统的结构示意图；
图2为根据一些实施例的网络预测系统与物理网络的传输关系示意图；
图3为根据一些实施例的网络预测系统的信息示意图；
图4为根据一些实施例的另一种网络预测系统的结构示意图；
图5为根据一些实施例的又一种网络预测系统的结构示意图；
图6为根据一些实施例的网络预测系统中各模型的传输关系示意图；
图7为根据一些实施例的网络预测方法的流程图；
图8为根据一些实施例的电子设备的结构示意图。
具体实施方式
为使本公开实施例的目的、技术方案和优点更加清楚，下面将结合附图对本公开的各实施例进行详细的阐述。然而，本领域的普通技术人员可以理解，在本公开各实施例中，为了使读者更好地理解本申请而提出了许多技术细节。但是，即使没有这些技术细节和基于以下各实施例的变化和修改，也可以实现本公开所要求保护的技术方案。
以下各个实施例的划分是为了描述方便，不应对本公开的具体实现方式构成任何限 定。各个实施例及实施例中的各特征在不矛盾的前提下可以相互结合相互引用。
以下，术语“第一”、“第二”仅用于描述目的，而不能理解为指示或暗示相对重要性或者隐含指明所指示的技术特征的数量。
本公开一些实施例提供一种网络预测系统。如图1所示，该网络预测系统10包括数字孪生体101和网络管理控制器102。
数字孪生体101由多个模型构成，多个模型各自独立运行。数字孪生体101用于基于通信网络的网络配置信息，对通信网络的网络性能进行仿真。数字孪生体101是一个虚拟模型，用于准确地反映物理对象的性能。
网络管理控制器102用于向数字孪生体101提供网络配置信息，从数字孪生体101获取仿真得到的网络性能数据，并根据网络性能数据，对通信网络的物理网络进行配置变更。
本公开实施例的网络预测系统10，可以应用于任何通信网络中，例如光传送网、电话交换网等。网络预测系统10可以作为通信网络的物理网络的数字孪生模型系统。在数字孪生体101中，每个模型可以是一个模型，或者是由多个小模型组成的一个总模型。各模型采用的技术可以各自不同。例如模型可以采用图神经网络、离散事件或者循环神经网络。各模型模拟的物理对象也可以各自不同。例如模型可以对光模块的性能、电源的稳定性，或者是网络数据传输的时延进行模拟。各模型的运行环境可以各不相同，或者各模型运行在相同的环境中。例如，模型可以运行在虚机中，或者运行在云环境或者物理机器与其他电子设备中。每个模型获取的输入信息可以来自物理网络，其他模型，或者来自于网络管理控制器102。模型的输出信息可以传输给其它模型，也可以作为模拟仿真的结果反馈给网络管理控制器102，经由网络管理控制器102判决后，发起网络配置影响物理网元的运行。
数字孪生体101所研究的物理对象会配备各种与重要功能领域相关的传感器，这些传感器输出与物理对象的不同方面的性能相关的数据，如能量输出、温度、天气条件等。然后，这些数据被转发到处理系统并应用于数字副本。获得这些数据后，虚拟模型可以用来运行模拟、研究性能问题，并生成可能的改进，最终产生有价值的见解。有价值的见解又可以应用于原始物理对象。
因此，通过数字孪生体构建物理网络的实时镜像,可增强物理网络所缺少的系统性仿真、优化、验证和控制能力,有利于网络新技术的部署,可以更加高效地应对网络的问题和挑战。将数字孪生技术应用于网络,创建物理网络的虚拟镜像,即可搭建数字孪生网络平台。通过物理网络和数字孪生网络实时交互,相互影响,数字孪生网络平台能够助力网络实现低成本试错、智能化决策和高效率创新。
在本公开实施例的网络预测系统中，通过数字孪生体中不同的模型，网络预测系统可以分别对整个通信网络的各个方面进行预测，从而可以从多个维度，对整个通信网络的功能、性能进行预测仿真。在该网络预测系统中，可以针对不同的仿真预测需求，为数字孪生体选择不同的模型结构，因此本公开实施例的预测系统的复用性强，可适应各种通信网络，即使网络规模扩大，网络的复杂性增加，也可以对网络的性能进行系统性的仿真预测。网络管理控制器为数字孪生体提供网络配置信息，从数字孪生体获取预测数据，并根据预测数据，对通信网络的物理网络进行配置变更，使得网络管理控制器可以在物理网络和数字孪生体之间进行信息传递，从而可以基于数字孪生体预测得到的网络性能数据对物理网络进行变更，使得物理网络的传输性能得到提升，实现仿真应用。
下面对本公开实施例提供的网络预测系统的实现细节进行说明，需要说明的是，以下内容仅为方便理解提供的实现细节，并非实施本方案的必须。
数字孪生体101中的各模型可以存在不同的触发条件，只有满足预设的触发条件后，模型才会进行仿真。数字孪生体101中的多个模型可以根据网络预测系统针对的网络功能或者性能进行选择，模型的种类和数量可以根据需要相应地进行变更。例如，对于网络性能的预测仿真，可以选择三种不同的模型；对于网络故障探测仿真，可以选择四种不同的 模型；每种模型各2个。通过多模型协同实现仿真。
网络管理控制器102可以是一个设备，或者，由多个设备组成的控制器。网络管理控制器102也可以是一个网络管理控制系统所使用的设备，所以，网络管理控制器102也称网络管理控制系统。
在一些实施例中，网络管理控制器102、数字孪生体101都部署于K8S云端。数字孪生体101的各个模型，以及数字孪生体101与网络管理控制器102之间的交互信息可以通过消息进行传输。消息的传输方式可以是订阅(Subscribe)、拉取(Pull)、推送(Push)中任意一种。订阅为模型预先向消息源进行信息订阅，定期获取消息源发送的携带信息的输入消息(即订阅消息)；拉取为模型向消息源发送信息查询请求，获取携带信息的输入消息(即应答消息)；推送为消息源主动向模型发送携带信息的输入消息(即推送消息)。
在数字孪生体101包括模型A至模型F的情况下，数字孪生体101中的各模型、网络管理控制系统104以及物理网络之间的传输关系可以如图2所示。
在一些实施例中，如图4所示，网络预测系统10还包括信息控制器103，信息控制器103用于分别向数字孪生体101的各模型下发对应的输入配置信息。输入配置信息用于配置模型的输入消息获取方式及数据类型。
数字孪生体101的各模型、物理网络、网络管理控制系统104，都是信息的源或者宿，他们统称为“服务”。
信息控制器103可以定义各个“服务”之间的信息输入输出关系、信息获取方式及信息数据类型，控制各“服务”间的消息协同。例如，信息控制器103基于信息图进行输入配置信息的下发。信息图为针对网络预测系统10所预测的网络功能和多个模型的模型选择进行定制的定制图表，用于指示多个模型的协同方式。当数字孪生体101中的模型发生变更，例如，有新模型注册，或者已有的模型卸载后，信息控制器103基于变更后的信息图进行输入配置信息的下发。网络预测系统可以通过图表的方式形成信息图，使各个“服务”之间建立信息传递的关系，控制各“服务”间的消息协同。如图3所示，图2中的模型A至模型F在信息图中分别对应Service(服务)A至Service F，图2中的网络管理控制系统对应Service G，图2中的物理网络对应Service H。
本公开实施例中，通过信息控制器分别向数字孪生体的各模型下发对应的输入配置信息，对各模型的输入消息获取方式及数据类型进行配置，即使不同模型之间存在格式差异，也可以通过信息控制器进行信息传递的协调，实现网络预测系统内的交互协同，并提高网络预测系统对模型的兼容性，增加网络预测系统的可复用度。
在一些实施例中，如图4所示，信息控制器103可以设置在网络管理控制系统104外部的电子设备上，以控制信息流，向各模型下发配置信息。配置信息可以包括输入配置信息和输出配置信息，以及其他配置信息。
信息控制器可以读取信息图，并将其翻译成各个“服务”中的配置信息，并配置到各个“服务”。信息控制器可以将配置到各个“服务”的配置信息定义如下。

在一些实施例中，输入消息获取方式包括以下任意一种或者任意组合：订阅、拉取和推送。
每个模型根据输入消息进行仿真后输出仿真的结果作为输出消息。每个输入消息以及输出消息都包括时间戳，用于表明消息发送的时间。
在一些实施例中，数字孪生体101的多个模型的每个在满足该模型的触发条件时开始运行，触发条件包括模型获取到全部所需的输入消息。全部所需的输入消息所携带的第三时间戳均大于网络预测系统的预测时刻与该模型的仿真步长的差值，且当前时刻早于预测时刻。即，数字孪生体101的各模型需要在预测时刻前的一个步长时间内，获取到所有需要的输入消息。也就是说，在获取到足够的仿真所需信息后，才触发该模型的运行，进行仿真。
在一些实施例中，数字孪生体101包括流量预测模型、设备劣化模型和网络性能模型。流量预测模型用于获取通信网络的流量信息，并根据流量信息，生成预测时刻的流量数据；设备劣化模型用于获取通信网络的设备状态信息，并根据设备状态信息，生成预测时刻的链路状态数据；网络性能模型用于获取流量数据、链路状态数据和通信网络的网络配置信息，并根据流量数据、链路状态数据和网络配置信息，生成预测时刻的网络性能数据。
流量预测模型可以与物理网络和网络性能模型进行连接，从物理网络获取流量信息，将预测生成的流量数据传输给网络性能模型。流量信息可以是一段时间内网络中各网元产生的所有流量数据，或者，是一段时间内网络中的各个域产生的所有流量数据等等。流量预测模型可以基于流量信息，进行预测时刻的流量数据的预测。预测时刻是预设好的一个时刻。预测的流量数据与获取的流量信息所属设备相同。例如，获取的流量信息来自于网元1，则基于网元1的流量信息预测得到的流量数据，就是网元1在预测时刻的流量大小。网络配置信息可以是网络的拓扑信息。
设备劣化模型可以与物理网络和网络性能模型进行连接，从物理网络获取设备状态信息，将预测生成的链路状态数据传输给网络性能模型。设备状态信息可以是一段时间内网络中各网元的设备状态信息，例如，当前寿命、电流、温度等。基于通信网络的类型不同，设备劣化模型可以有对应的模型设计，例如，适用于电话交换网络的设备劣化模型，适用于光纤网络的设备劣化模型。设备劣化模型可以基于设备状态信息进行预测时刻的链路状态数据的预测，例如，可以基于链路中的各网元的设备状态信息预测链路的传输损耗或者 传输速率等。
网络性能模型可以与流量预测模型、设备劣化模型和物理网络进行连接，从物理网络获取网络配置信息，从流量预测模型获取流量数据，从设备劣化模型获取链路状态数据，进行预测时刻的网络性能数据的预测。网络配置信息可以是路由配置信息、网络中不同域的划分信息等。网络性能数据可以是网络时延、抖动、丢包率等性能数据。
本公开实施例中，流量预测模型可以对流量进行预测，得到预测时刻的流量的大小，设备劣化模型可以对链路状态进行预测，网络性能模型可以基于流量的大小、链路状态和网络配置，预测网络的性能。即，系统可以使用不同的模型，分别对整个通信网络的流量、链路状态两个方面进行预测，再基于这两个方面的预测结果，对网络性能进行预测，从而可以从多个维度，对整个通信网络的性能进行预测。
在一些实施例中，通信网络可以是光传送网，在此情况下，设备状态信息可以是光模块状态数据，设备劣化模型可以使用与光传送网对应的光模块劣化模型。
在一些实施例中，流量预测模型获取流量信息的时刻早于预测时刻，且流量预测模型获取流量信息的时刻与预测时刻之间的时间间隔，小于等于第一预设仿真步长。即，对预测时刻进行预测使用的流量信息的获取时刻与预测时刻较为接近。因此，预测得到的流量数据准确度较高。
设备劣化模型获取设备状态信息的时刻早于预测时刻，且设备劣化模型获取设备状态信息的时刻与预测时刻之间的时间间隔，小于等于第二预设仿真步长。即，对预测时刻进行预测使用的设备状态信息的获取时刻与预测时刻较为接近。因此，预测得到的链路状态数据准确度较高，从而可以得到准确度较高的网络性能数据。
第一预设仿真步长和第二预设仿真步长的值可以根据需求预设，二者可以相同，也可以不同。针对第一预设仿真步长和第二预设仿真步长的任一个，可以根据通信网络的类型不同，为其预设不同的数值。
在一些实施例中，数字孪生体101中的各模型可以通过输入消息获取各自所需的信息和数据。流量预测模型，还用于从流量预测模型的输入消息中获取流量信息，在根据流量信息生成预测时刻的流量数据之前，验证流量预测模型的输入消息所携带的第一时间戳大于预测时刻与第一预设仿真步长的差值，且第一时间戳早于预测时刻。设备劣化模型，还用于从设备劣化模型的输入消息中获取设备状态信息，在根据设备状态信息生成预测时刻的链路状态数据之前，验证设备劣化模型的输入消息所携带的第二时间戳大于预测时刻与第二预设仿真步长的差值，且第二时间戳早于预测时刻。
本公开实施例中，通过以流量预测模型的输入消息承载流量信息，以设备劣化模型的输入消息承载设备状态信息，并在输入消息中携带消息时间戳，则流量预测模型和设备劣化模型通过验证各自的输入消息中携带的消息时间戳，就可以在生成预测时刻的流量数据和预测时刻的链路状态数据之前，对流量信息和设备状态信息的获取时间进行验证，保证仿真得到的网络性能数据的准确性。
模型N的输入信息是T-t时刻的数据，输出信息是仿真后推断的T时刻的数据，则模型N的仿真步长为t。
假设模型N的仿真时间步长为t，当模型N达到下面的要求时，模型N启动T时刻的仿真流程：1、模型N的所有携带输入信息的消息的时间戳都大于T-t时刻；2、在模型N获取到所有输入信息的前提下，T时刻是最接近当前的一个时刻。
在一些实施例中，如图5所示，网络管理控制系统104、数字孪生体101都部署于K8S云端，数字孪生体101由流量预测模型、光模块劣化模型、网络性能模型组成，这些模型都是运行在K8S中的微服务。信息控制器103内置于网络管理控制系统104中，读取信息图，分解成网络性能模型、流量预测模型、光模块劣化模型的输入信息配置如下，并下发到这三个模型所在的微服务。
流量预测模型的配置如下：
光模块劣化模型在接收到信息控制器的配置信息后，完成当前寿命、电流、温度等输入消息的订阅。
网络性能模型的配置如下：

网络性能模型在接收到信息控制器发送的配置信息后，完成流量和链路状态输入信息拉取的微服务设置。
在一些实施例中，数字孪生体101中的各模型的信息传输关系可以如图6所示，流量预测模型(flowForecast Model)从流量数据代理服务(flowData Agent)订阅得到流量信息，流量数据代理服务可以是物理网络中的一个服务；网络性能模型(networkPerformance Model)通过调用预设的数据接口，从流量预测模型中拉取流量数据。光纤接口劣化模型(optAgeing Model)从光纤接口数据代理服务(optData Agent)订阅得到光纤设备的状态信息，光纤接口数据代理服务可以是物理网络中的一个服务；网络性能模型(networkPerformance Model)通过调用预设的数据接口，从光纤接口劣化模型中拉取链路状态数据。
本公开的实施例提供一种网络预测方法，如图7所示，该网络预测方法包括步骤S701～S703。
步骤S701，数字孪生体101从网络管理控制器102获取通信网络的网络配置信息；数字孪生体由多个模型构成，多个模型各自独立运行。
步骤S702，数字孪生体101基于网络配置信息，对通信网络进行仿真预测。
步骤S703，网络管理控制器102从数字孪生体101获取仿真得到的预测数据，并根据预测数据，对通信网络的物理网络进行配置变更。
在一些实施例中，实现网络预测方法的网络预测系统除包括数字孪生体101和网络管理控制器102外，还包括信息控制器103。在步骤S702之前，所述方法还包括信息控制器103分别向数字孪生体101的各模型下发对应的输入配置信息。输入配置信息用于配置模型的输入消息获取方式及数据类型。
在一些实施例中，信息控制器103基于信息图进行输入配置信息的下发。信息图为针对网络预测系统所预测的网络功能和多个模型的模型选择进行定制的定制图表，用于指示多个模型的协同方式。
在一些实施例中，在数字孪生体101中的模型发生变更的情况下，信息控制器基于变更后的信息图进行输入配置信息的下发。
在一些实施例中，数字孪生体101中的多个模型的每个在满足该模型的触发条件时开始运行，触发条件包括模型获取到全部所需的输入消息。全部所需的输入消息所携带的第三时间戳均大于网络预测系统的预测时刻与该模型的仿真步长的差值，且当前时刻早于预测时刻。
在一些实施例中，数字孪生体101包括流量预测模型、设备劣化模型和网络性能模型。
步骤S702包括如下。
流量预测模型获取通信网络的流量信息，并根据流量信息，生成预测时刻的流量数据；设备劣化模型获取通信网络的设备状态信息，并根据设备状态信息，生成预测时刻的链路状态数据；网络性能模型获取流量数据、链路状态数据和通信网络的网络配置信息，并根据流量数据、链路状态数据和网络配置信息，生成预测时刻的网络性能数据。
在一个实施例中，实现网络预测方法的网络预测系统包括通信网络的：
流量预测模型：仿真步长为t，T-t时刻收到订阅的流量数据信息，触发仿真流程，生成T时刻的流量数据。
光模块劣化模型：仿真步长为t，T-t时刻收到订阅的光模块状态数据，包括了当前寿命、电流、温度等，触发仿真流程，生成T时刻光模块状态数据。
网络性能模型：仿真步长为0，T时刻收到网络管理控制系统推送的T时刻路由，触发信息拉取流程，从流量预测模型中拉取到T时刻流量数据，从光模块劣化模型中拉取到T时刻的光模块状态数据，满足了T时刻所有的输入信息获取，触发仿真流程，生成T时 刻的网络时延、抖动、丢包性能数据，并推送给网络管理控制系统。
在此基础上，网络管理控制系统在接收到网络性能模型推送的仿真结果后，判断满足要求，将相应的路由配置下发到物理网络中。
需要说明的是，本实施例为与上述实施例相对应的系统实施例，本实施例可与上述实施例互相配合实施。上述实施例中提到的相关技术细节在本实施例中依然有效，为了减少重复，这里不再赘述。相应地，本实施例中提到的相关技术细节也可应用在上述实施例中。
上面各种方法的步骤划分，只是为了描述清楚，在实现时可以合并为一个步骤或者对某些步骤进行拆分，分解为多个步骤，只要包括相同的逻辑关系，都在本公开的保护范围内；对算法中或者流程中添加无关紧要的修改或者引入无关紧要的设计，但不改变其算法和流程的核心设计都在本公开的保护范围内。
本公开实施例还涉及一种电子设备，如图8所示，该电子设备包括至少一个处理器801；与至少一个处理器通信连接的存储器802。存储器802存储有可被至少一个处理器801执行的指令，指令被至少一个处理器801执行上述的方法实施例。
存储器802和处理器801采用总线方式连接，总线可以包括任意数量的互联的总线和桥，总线将一个或多个处理器801和存储器802的各种电路连接在一起。总线还可以将诸如外围设备、稳压器和功率管理电路等之类的各种其他电路连接在一起，本公开不再对其进行进一步描述。总线接口在总线和收发机之间提供接口。收发机可以是一个元件，也可以包括多个元件，比如多个接收器和发送器，提供用于在传输介质上与各种其他装置通信的单元。经处理器801处理的信息通过天线在无线介质上进行传输，天线还可以接收无线介质上传输的信息并将信息传送给处理器801。
处理器801负责管理总线和通常的处理，还可以提供各种功能，包括定时，外围接口，电压调节、电源管理以及其他控制功能。存储器802可以被用于存储处理器在执行操作时所使用的信息。
本公开实施例还涉及一种非暂态计算机可读存储介质，存储有计算机程序。计算机程序被处理器执行时实现上述方法实施例。
本领域技术人员可以理解，实现上述实施例方法中的全部或部分步骤是可以通过程序来指令相关的硬件来完成。该程序存储在一个存储介质中，包括若干指令用以使得一个设备(可以是单片机，芯片等)或处理器(processor)执行本公开实施例所述方法的全部或部分步骤。而前述的存储介质包括：U盘、移动硬盘、只读存储器(ROM，Read-Only Memory)、随机存取存储器(RAM，Random Access Memory)、磁碟或者光盘等各种可以存储程序代码的介质。
本领域的技术人员将会理解，本发明的公开范围不限于上述具体实施例，并且可以在不脱离本申请的精神的情况下对实施例的某些要素进行修改和替换。本申请的范围受所附权利要求的限制。
本领域的普通技术人员可以理解，上述各实施方式是实现本公开的具体实施例，而在实际应用中，可以在不脱离本申请的精神的情况下对实施例的某些要素进行修改和替换，而不偏离本公开的精神和范围。

Claims (13)

  1. 一种网络预测系统，包括：
    数字孪生体，
    由多个模型构成，所述多个模型各自独立运行，所述数字孪生体用于基于通信网络的网络配置信息，对所述通信网络进行仿真预测；
    网络管理控制器，
    用于向所述数字孪生体提供所述网络配置信息，从所述数字孪生体获取仿真得到的预测数据，并根据所述预测数据，对所述通信网络的物理网络进行配置变更。
  2. 根据权利要求1所述的网络预测系统，还包括信息控制器，用于分别向所述数字孪生体中的所述多个模型下发对应的输入配置信息，其中，所述输入配置信息用于配置模型的输入消息获取方式及数据类型。
  3. 根据权利要求2所述的网络预测系统，其中，所述输入消息的获取方式，包括以下任意一种或者任意组合：
    订阅、拉取、推送；
    其中，所述订阅为所述模型预先向消息源进行信息订阅，定期获取所述消息源发送的携带信息的所述输入消息；
    所述拉取为所述模型向所述消息源发送信息查询请求，获取携带信息的所述输入消息；
    所述推送为所述消息源主动向所述模型发送携带信息的所述输入消息。
  4. 根据权利要求2所述的网络预测系统，其中，所述信息控制器用于基于信息图进行所述输入配置信息的下发；
    其中，所述信息图为针对所述网络预测系统所预测的网络功能和所述多个模型的模型选择进行定制的定制图表，用于指示所述多个模型的协同方式。
  5. 根据权利要求2至4中任一项所述的网络预测系统，其中，所述多个模型的每个在满足该模型的触发条件时开始运行；
    所述触发条件包括所述模型获取到全部所需的输入消息；
    其中，全部所需的所述输入消息所携带的时间戳均大于所述网络预测系统的预测时刻与所述模型的仿真步长的差值，且当前时刻早于所述预测时刻。
  6. 一种网络预测方法，包括：
    数字孪生体从网络管理控制器获取通信网络的网络配置信息；其中，所述数字孪生体由多个模型构成，所述多个模型各自独立运行；
    所述数字孪生体基于所述网络配置信息，对所述通信网络进行仿真预测；
    所述网络管理控制器从所述数字孪生体获取所述仿真得到的预测数据，并根据所述预测数据，对所述通信网络的物理网络进行配置变更。
  7. 根据权利要求6所述的网络预测方法，其中，在所述数字孪生体基于所述网络配置信息，对所述通信网络进行仿真预测之前，所述方法还包括：
    信息控制器分别向所述数字孪生体中的各模型下发对应的输入配置信息，其中，所述输入配置信息用于配置模型的输入消息获取方式及数据类型。
  8. 根据权利要求7所述的网络预测方法，其中，所述方法还包括：
    在所述数字孪生体中的模型发生变更的情况下，所述信息控制器基于变更后的信息图进行所述输入配置信息的下发；
    其中，所述信息图为针对网络预测系统所预测的网络功能和所述多个模型的模型选择进行定制的定制图表，用于指示所述多个模型的协同方式。
  9. 根据权利要求7所述的网络预测方法，其中，所述输入消息的获取方式，包括以下任意一种或者任意组合：
    订阅、拉取、推送；
    其中，所述订阅为所述模型预先向消息源进行信息订阅，定期获取所述消息源发送的 携带信息的所述输入消息；
    所述拉取为所述模型向所述消息源发送信息查询请求，获取携带信息的所述输入消息；
    所述推送为所述消息源主动向所述模型发送携带信息的所述输入消息。
  10. 根据权利要求6所述的网络预测方法，其中，所述数字孪生体包括流量预测模型、设备劣化模型和网络性能模型；
    所述数字孪生体基于所述网络配置信息，对所述通信网络进行仿真预测，包括：
    所述流量预测模型获取所述通信网络的流量信息，并根据所述流量信息，生成预测时刻的流量数据；
    所述设备劣化模型获取所述通信网络的设备状态信息，并根据所述设备状态信息，生成所述预测时刻的链路状态数据；
    所述网络性能模型获取所述流量数据、所述链路状态数据和所述通信网络的所述网络配置信息，并根据所述流量数据、所述链路状态数据和所述网络配置信息，生成所述预测时刻的网络性能数据。
  11. 根据权利要求10所述的网络预测方法，其中，所述流量预测模型获取所述流量信息的时刻早于所述预测时刻，且所述流量预测模型获取所述流量信息的时刻与所述预测时刻之间的时间间隔，小于等于第一预设仿真步长；
    所述设备劣化模型获取所述设备状态信息的时刻早于所述预测时刻，且所述设备劣化模型获取所述设备状态信息的时刻与所述预测时刻之间的时间间隔，小于等于第二预设仿真步长。
  12. 一种电子设备，包括：
    至少一个处理器；
    与所述至少一个处理器通信连接的存储器；
    所述存储器存储有被所述至少一个处理器执行的指令，所述指令被所述至少一个处理器执行，以使所述至少一个处理器执行如权利要求6至11中任一项所述的网络预测方法。
  13. 一种非暂态计算机可读存储介质，存储有计算机程序，其中，所述计算机程序被处理器执行时实现如权利要求6至11中任一项所述的网络预测方法。
PCT/CN2023/077491 2022-07-15 2023-02-21 网络预测系统、方法、电子设备及存储介质 WO2024011908A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210836171.2A CN117440417A (zh) 2022-07-15 2022-07-15 网络预测系统、方法、电子设备及存储介质
CN202210836171.2 2022-07-15

Publications (1)

Publication Number Publication Date
WO2024011908A1 true WO2024011908A1 (zh) 2024-01-18

Family

ID=89535361

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2023/077491 WO2024011908A1 (zh) 2022-07-15 2023-02-21 网络预测系统、方法、电子设备及存储介质

Country Status (2)

Country Link
CN (1) CN117440417A (zh)
WO (1) WO2024011908A1 (zh)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20190294975A1 (en) * 2018-03-21 2019-09-26 Swim.IT Inc Predicting using digital twins
CN113810953A (zh) * 2021-09-08 2021-12-17 重庆邮电大学 一种基于数字孪生的无线传感网络资源调度方法及系统
US20220014946A1 (en) * 2021-09-23 2022-01-13 Intel Corporation Methods and apparatus for digital twin aided resiliency
CN114189888A (zh) * 2021-11-29 2022-03-15 中国船舶重工集团公司第七一六研究所 基于数字孪生的5g融合网架构下多模终端接入系统及方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20190294975A1 (en) * 2018-03-21 2019-09-26 Swim.IT Inc Predicting using digital twins
CN113810953A (zh) * 2021-09-08 2021-12-17 重庆邮电大学 一种基于数字孪生的无线传感网络资源调度方法及系统
US20220014946A1 (en) * 2021-09-23 2022-01-13 Intel Corporation Methods and apparatus for digital twin aided resiliency
CN114189888A (zh) * 2021-11-29 2022-03-15 中国船舶重工集团公司第七一六研究所 基于数字孪生的5g融合网架构下多模终端接入系统及方法

Also Published As

Publication number Publication date
CN117440417A (zh) 2024-01-23

Similar Documents

Publication Publication Date Title
WO2020168761A1 (zh) 训练模型的方法和装置
US11669085B2 (en) Method and system for determining system settings for an industrial system
Xu et al. Dynamic resource provisioning for workflow scheduling under uncertainty in edge computing environment
CN113709048A (zh) 一种路由信息的发送、接收方法、网元及节点设备
EP3318976B1 (en) Data flow control device and data flow control method
CN111897300B (zh) 一种基于opc-ua的软/硬件联合仿真系统及其自适应协同方法
CN108306804A (zh) 一种Ethercat主站控制器及其通信方法和系统
CN109936473A (zh) 基于深度学习预测的分布计算系统及其运行方法
WO2023098374A1 (zh) 网络资源部署方法、装置、电子设备及存储介质
Ebeid et al. Model-driven design of network aspects of distributed embedded systems
Cintuglu et al. Real-time asynchronous information processing in distributed power systems control
US11556100B2 (en) Control method, related device, and system
WO2024011908A1 (zh) 网络预测系统、方法、电子设备及存储介质
CN109905435A (zh) 资源订阅方法及系统、存储介质、电子设备
TW201931227A (zh) 一種虛實網路資源拓樸之介面系統
CN109033483A (zh) 一种在yang模型中定义数据关系的方法、装置及系统
Pham et al. Towards an Elastic Fog‐Computing Framework for IoT Big Data Analytics Applications
CN111935026B (zh) 一种数据传输方法、装置、处理设备及介质
CN115001692A (zh) 模型更新方法及装置、计算机可读存储介质和电子设备
CN115460617A (zh) 基于联邦学习的网络负载预测方法、装置、电子设备及介质
CN112448833B (zh) 一种多管理域的通信方法和装置
CN113904946A (zh) 通信仿真中断控制方法及系统
CN112906745A (zh) 基于边缘协同的诚信智能网络训练方法
CN107241234B (zh) 一种as5643网络仿真方法及系统
Sghaier et al. Model Based Validation of Real Time QoS for NCDCLA Protocol in Wireless Sensor Networks

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 23838410

Country of ref document: EP

Kind code of ref document: A1