Network simulation method, device and equipment based on digital twinning and medium

Abstract

The invention discloses a network simulation method, device and equipment based on digital twinning and a medium. The method comprises the following steps: carrying out initial simulation deployment of a target network on a pre-constructed three-dimensional digital scene containing a virtual entity; acquiring multi-source state data of a corresponding physical entity in real time, so as to drive the virtual entity to move in the three-dimensional digital scene in real time through the multi-source state data, and acquiring virtual state data of the virtual entity, so as to update channel parameters of a target network in real time according to the virtual state data; and driving a target network to operate through the channel parameters, continuously monitoring performance data of the target network, and optimizing the target network according to the performance data to realize network simulation. According to the method, the accuracy of a network simulation result can be improved by constructing a network digital twin environment interacting with a real physical world in real time.

Description

Technical Field

[0001] This invention relates to the field of network simulation technology, and in particular to a network simulation method, apparatus, device, and medium based on digital twins. Background Technology

[0002] Network simulation is a key tool for the research and verification of 6G communication technology. As the vision of 6G expands to complex application scenarios such as holographic communication, the sensory internet, and collaborative unmanned systems, networks need to meet extremely high requirements for bandwidth, latency, reliability, and connection density. Through simulation, network performance can be predicted and evaluated before actual deployment, protocol algorithms can be optimized, and the robustness and adaptability of the network in dynamic physical environments can be ensured. This reduces R&D costs, accelerates technology implementation, and provides reliable support for the construction of future intelligent information infrastructure.

[0003] Existing technologies primarily rely on traditional network simulation tools (such as NS-3 and OMNeT++), which are based on event-driven and mathematical modeling and focus on the abstract simulation of network protocol stacks. However, current network simulation technologies remain at the static simulation stage, resulting in a severe disconnect between the simulation environment and physical reality. They cannot accurately simulate the complex impact of the real-world three-dimensional physical environment (such as building occlusion, terrain undulations, and dynamic obstacles) on wireless signal propagation, especially the dynamic impact of moving obstacles on wireless signal propagation. This makes it difficult for simulation results to accurately reflect network behavior and performance in actual deployment scenarios. Summary of the Invention

[0004] This invention provides a network simulation method, apparatus, device, and medium based on digital twins, which can improve the accuracy of network simulation results by constructing a network digital twin environment that interacts with the real physical world in real time.

[0005] In a first aspect, embodiments of the present invention provide a network simulation method based on digital twins, comprising:

[0006] An initial simulation deployment of a target network is performed on a pre-constructed 3D digital scene containing virtual entities; wherein the 3D digital scene is constructed based on a preset actual physical scene; and the virtual entities are dynamically constructed based on physical entities moving in the actual physical scene.

[0007] The system acquires multi-source state data of the corresponding physical entity in real time, and drives the virtual entity to move in the three-dimensional digital scene in real time using the multi-source state data. It also acquires virtual state data of the virtual entity, and updates the channel parameters of the target network in real time based on the virtual state data. The virtual state data includes the relative position and pose of the virtual entity and a preset virtual base station.

[0008] The target network is driven by the channel parameters, and its performance data is continuously monitored. Based on the performance data, the target network is optimized to achieve network simulation.

[0009] This invention establishes a high-fidelity digital twin scene and virtual network corresponding to the physical world by deploying simulations in a preset 3D digital scene. This ensures that the simulation environment originates from the real physical scene, solving the problems of simplified simulation environments and disconnection from physical reality in existing technologies. Through real-time bidirectional mapping between physical and virtual entities (i.e., driving the movement of virtual entities through the state data of physical entities in reality), the simulation input is ensured to be based on dynamic changes in the real world, improving simulation realism. Furthermore, since the propagation of network signals is affected by obstruction or reflection from obstacles, the dynamic changes in the physical world (such as position and attitude) are transformed into network channel parameters (such as path loss and delay) during network simulation, enabling the network simulation to respond to the physical environment and solving the problem of simplified channel models in existing technologies. Network simulation is performed based on real-time channel parameters, and network performance indicators (such as throughput and packet loss rate) are output, providing a data foundation for closed-loop feedback. Then, based on the network performance indicators, network configuration commands are dynamically generated and the network is optimized, realizing intelligent decision-making and adjustment based on network state, forming a dynamic closed loop of "physical-network-application," solving the problems of static simulation processes and lack of interactivity in existing technologies. Compared with existing technologies, this invention can improve the accuracy of network simulation results by constructing a network digital twin environment that interacts with the real physical world in real time.

[0010] Furthermore, before the initial simulation deployment of the target network on the pre-built 3D digital scene containing virtual entities, the process also includes constructing the 3D digital scene.

[0011] Specifically, constructing a three-dimensional digital scene involves:

[0012] Collect multi-source heterogeneous spatial data of the actual physical scene; wherein, the multi-source heterogeneous spatial data includes point cloud data, multi-angle image data and vector map data;

[0013] The point cloud data is registered to a unified coordinate system, and a preset surface reconstruction algorithm is used to transform the point cloud data into a triangular mesh model with a topological structure to generate the geometric skeleton of the three-dimensional digital scene.

[0014] By using a pre-defined multi-view stereo matching technology, the multi-angle image data is mapped as texture onto the surface of the geometric skeleton to obtain an initial three-dimensional digital scene.

[0015] Using a pre-defined 3D rendering engine, semantic and physical attributes of model elements in the initial 3D digital scene are marked based on the vector map data; wherein, the model elements include buildings and streets; and the physical attributes include materials and collision volumes.

[0016] This invention provides a foundation for high-fidelity scene construction by acquiring detailed data of the real environment, ensuring that the simulation environment is consistent with physical reality; it generates accurate three-dimensional geometric models based on point cloud data, solving the problem of scene topology simplification in the prior art; it increases the realism of the scene through texture mapping, making the virtual environment visually close to the physical world; and it supports interaction and simulation (such as occlusion detection in channel model calculation) through semantic and physical attribute labeling, improving simulation accuracy.

[0017] Furthermore, the real-time acquisition of multi-source state data corresponding to the physical entity, in order to drive the virtual entity to move in the three-dimensional digital scene in real time using the multi-source state data, specifically involves:

[0018] The system acquires multi-source state data of the physical entity in real time; wherein the multi-source state data includes the physical entity's geographical location, triaxial acceleration, and angular velocity.

[0019] The multi-source state data is fused to calculate the six-degree-of-freedom state vector of the physical entity; where the degree of freedom refers to the direction in which the physical entity can move independently.

[0020] The virtual entity is driven to move in the three-dimensional digital scene in real time using the six-degree-of-freedom state vector.

[0021] This invention collects real-time data from physical entities to ensure the authenticity of the data source; it solves the technical problem of unstable sensor data by processing data fusion to calculate the six-degree-of-freedom state vector; and it achieves virtual-real synchronization by mapping the physical state to virtual entities, ensuring that the virtual environment dynamically responds to changes in the physical world.

[0022] Furthermore, the virtual entity is driven to move in the three-dimensional digital scene in real time through the six-degree-of-freedom state vector, specifically as follows:

[0023] The six-degree-of-freedom state vector is integrated with the corresponding timestamp, and the geodetic coordinates in the six-degree-of-freedom state vector are converted into virtual coordinates in the three-dimensional digital scene to generate the final pose data.

[0024] Based on the final pose data, the transformation matrix of the virtual entity is updated so as to drive the synchronous refresh of the pose of the virtual entity in the three-dimensional digital scene, thereby realizing the movement of the virtual entity in the three-dimensional digital scene.

[0025] This invention integrates timestamps and coordinate transformations to ensure data time synchronization and spatial alignment, enabling virtual entities to update at the correct time and location, reducing mapping errors. By updating the transformation matrix, the virtual entity pose changes are driven in the rendering engine, achieving smooth and accurate motion rendering, improving user experience and simulation realism.

[0026] Furthermore, after converting the geodetic coordinates in the six-degree-of-freedom state vector into virtual coordinates in the three-dimensional digital scene, the coordinate sequence formed by the virtual coordinates is smoothed using a preset interpolation algorithm or extrapolation algorithm to compensate for network jitter.

[0027] The embodiments of the present invention smooth the coordinate sequence to reduce the discontinuity or jitter of virtual entity movement caused by network latency or jitter, ensuring smooth virtual entity movement even under non-ideal network conditions and improving simulation quality.

[0028] Furthermore, based on the virtual state data, the channel parameters of the target network are updated in real time, specifically as follows:

[0029] By using a pre-defined ray tracing-based channel model and combining it with the virtual state data, the influence of the virtual entity's movement on the network signal is simulated, and the channel parameters of the target network are updated based on the influence.

[0030] This invention uses a ray-tracing-based channel model to simulate wireless signal propagation with high precision, taking into account physical factors such as obstruction and reflection, making the channel parameters more realistic and solving the problem of simplification in existing channel models. Furthermore, it updates the channel parameters based on virtual state data, transforming the dynamic changes of virtual entities into changes in channel conditions, enabling the network simulation to respond dynamically to the physical environment and improving simulation accuracy.

[0031] Furthermore, based on the performance data, new network configuration commands are dynamically generated, specifically as follows:

[0032] The performance data is fused with pre-acquired context data to construct a comprehensive state vector; wherein, the performance data includes the current speed and service type of the virtual entity;

[0033] The integrated state vector is input into a preset reinforcement learning model so that the reinforcement learning model generates action instructions for adjusting the network state according to a preset optimization objective function.

[0034] The action instructions are parsed into one or more network configuration commands for the target network.

[0035] This invention provides comprehensive decision input by integrating network performance (such as bandwidth and latency) and contextual information (such as entity speed and service type); it generates action instructions by using a reinforcement learning model to achieve intelligent decision-making, automatically optimize network configuration based on real-time status, adapt to dynamic changes, and solve the problem of static simulation in the prior art.

[0036] Secondly, embodiments of the present invention provide a network simulation device based on digital twins, including a simulation deployment module, a channel parameter acquisition module, and a simulation network optimization module, wherein...

[0037] The simulation deployment module is used to perform initial simulation deployment of the target network on a pre-constructed 3D digital scene containing virtual entities; wherein, the 3D digital scene is constructed according to a preset actual physical scene; and the virtual entities are dynamically constructed according to physical entities moving in the actual physical scene.

[0038] The channel parameter acquisition module is used to acquire multi-source state data of the corresponding physical entity in real time, so as to drive the virtual entity to move in the three-dimensional digital scene in real time through the multi-source state data, and acquire the virtual state data of the virtual entity, so as to update the channel parameters of the target network in real time according to the virtual state data; wherein, the virtual state data includes the relative position and pose of the virtual entity and the preset virtual base station;

[0039] The simulation network optimization module is used to drive the operation of the target network through the channel parameters, continuously monitor the performance data of the target network, and optimize the target network based on the performance data to achieve network simulation.

[0040] This invention, through a simulation deployment module, performs simulation deployment in a preset 3D digital scene, thereby establishing a high-fidelity digital twin scene and virtual network corresponding to the physical world. This ensures that the simulation environment originates from the real physical scene, solving the problems of simplified simulation environments and disconnection from physical reality in existing technologies. Through a channel parameter acquisition module, a real-time bidirectional mapping between physical and virtual entities is established (i.e., the movement of virtual entities is driven by the state data of physical entities in reality), ensuring that the simulation input is based on dynamic changes in the real world, improving simulation realism. Furthermore, since the propagation of network signals can be affected by obstacles or reflections, the network simulation... In the simulation process, dynamic changes in the physical world (such as position and attitude) are transformed into network channel parameters (such as path loss and delay), enabling the network simulation to respond to the physical environment and solving the problem of simplified channel models in existing technologies. Through the simulation network optimization module, network simulation is performed based on real-time channel parameters, and network performance indicators (such as throughput and packet loss rate) are output, providing a data foundation for closed-loop feedback. Then, based on the network performance indicators, network configuration commands are dynamically generated and the network is optimized, realizing intelligent decision-making and adjustment based on network status, forming a dynamic closed loop of "physical-network-application", and solving the problems of static simulation processes and lack of interactive capabilities in existing technologies.

[0041] Thirdly, embodiments of the present invention provide a terminal device, including: a processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other through the communication bus;

[0042] The memory is used to store at least one executable instruction that causes the processor to perform operations as described in any of the above-described digital twin-based network simulation methods.

[0043] Fourthly, embodiments of the present invention provide a computer-readable storage medium, the computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device or apparatus where the computer-readable storage medium is located to perform the network simulation method based on digital twin as described in any of the above.

[0044] The above description is merely an overview of the technical solutions of the embodiments of the present invention. In order to better understand the technical means of the embodiments of the present invention and to implement them in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the embodiments of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0045] Figure 1 A schematic diagram of a network simulation method based on digital twins provided in an embodiment of the present invention;

[0046] Figure 2 A schematic diagram of a digital twin scene model for a simulation platform;

[0047] Figure 3 This invention provides a schematic diagram of a network simulation system based on digital twins.

[0048] Figure 4 This is a structural diagram of a network simulation device based on digital twins provided in an embodiment of the present invention. Detailed Implementation

[0049] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] Example 1:

[0051] like Figure 1 As shown in the figure, a network simulation method based on digital twins provided by an embodiment of the present invention includes the following steps:

[0052] S101, Initial simulation deployment of the target network is performed on a pre-constructed 3D digital scene containing virtual entities; wherein, the 3D digital scene is constructed according to a preset actual physical scene; the virtual entities are dynamically constructed according to physical entities moving in the actual physical scene;

[0053] In one specific embodiment, by utilizing Unreal Engine to construct centimeter-level 3D digital scenes such as cities and indoor environments, and combining this with spectrum generation tools like Sionna to simulate real channel propagation, the realism of the simulation environment is greatly enhanced. Entities in the physical world (vehicles, drones, etc.) can be mapped into the virtual scene in real time, ensuring that the simulation input originates from the real world. Figure 2 As shown, Figure 2 This is a schematic diagram of a digital twin scene model for a simulation platform.

[0054] In this embodiment, before the initial simulation deployment of the target network on the pre-constructed 3D digital scene containing virtual entities, the method further includes constructing the 3D digital scene. Specifically, constructing the 3D digital scene involves: collecting multi-source heterogeneous spatial data of the actual physical scene; wherein the multi-source heterogeneous spatial data includes point cloud data, multi-angle image data, and vector map data; registering the point cloud data to a unified coordinate system, and using a preset surface reconstruction algorithm to convert the point cloud data into a triangular mesh model with a topological structure to generate the geometric skeleton of the 3D digital scene; using a preset multi-view stereo matching technology, mapping the multi-angle image data as texture onto the surface of the geometric skeleton to obtain the initial 3D digital scene; and using a preset 3D rendering engine, marking the semantic and physical attributes of the model elements in the initial 3D digital scene based on the vector map data; wherein the model elements include buildings and streets; and the physical attributes include material and collision volume.

[0055] In one specific embodiment, multi-source heterogeneous spatial data of the physical environment is collected. The data includes, but is not limited to, high-precision point cloud data obtained by lidar scanning, multi-angle high-definition image data obtained by oblique photography, and vector map data obtained from a geographic information system (GIS).

[0056] In one specific embodiment, denoising, filtering, and thinning operations are performed on the point cloud data; and iterative nearest point (ICP) or its improved algorithm is used to register point cloud data from different stations to a unified coordinate system. Color correction and distortion correction are performed on the image data.

[0057] In one specific embodiment, a surface reconstruction algorithm (e.g., Poisson reconstruction algorithm) is used to transform the processed point cloud data into a triangular mesh model with a topological structure to generate the geometric skeleton of the scene. Subsequently, using multi-view stereo matching technology, the high-resolution image is accurately mapped as a texture and baked onto the surface of the triangular mesh model.

[0058] In one specific embodiment, the generated 3D model is imported into a 3D rendering engine (such as Unreal Engine), and combined with the GIS data, the model elements (such as buildings and roads) in the scene are semantically marked and given physical attributes (such as material and collision volume), and finally an interactive, highly realistic digital twin scene is constructed.

[0059] In one specific embodiment, the initial simulation deployment of the target network is carried out on a pre-constructed 3D digital scene containing virtual entities. Specifically, based on containerization or virtualization technology, a complete software-defined 6G network simulation environment is deployed, including a core network (such as Open5GS) and a radio access network (such as UERANSIM). Key parameters of this virtual network, such as topology, base station locations, network slicing, and QoS policies, can be defined and dynamically adjusted through configuration files (such as YAML format).

[0060] S102, real-time acquisition of multi-source state data of the corresponding physical entity, so as to drive the virtual entity to move in the three-dimensional digital scene in real time through the multi-source state data, and acquisition of virtual state data of the virtual entity, so as to update the channel parameters of the target network in real time according to the virtual state data; wherein, the virtual state data includes the relative position and pose of the virtual entity and the preset virtual base station;

[0061] In this embodiment, the real-time acquisition of multi-source state data of the corresponding physical entity, so as to drive the virtual entity to move in the three-dimensional digital scene in real time through the multi-source state data, specifically involves: acquiring multi-source state data of the physical entity in real time; wherein, the multi-source state data includes the physical entity's geographical location, three-axis acceleration, and angular velocity; performing fusion processing on the multi-source state data to calculate the six-degree-of-freedom state vector of the physical entity; wherein, the degree of freedom refers to the direction in which the physical entity can move independently; and driving the virtual entity to move in the three-dimensional digital scene in real time through the six-degree-of-freedom state vector.

[0062] In one specific embodiment, raw data streams containing geographic location, triaxial acceleration, and angular velocity are acquired at a preset frequency using onboard sensors such as a Global Positioning System (GPS) and an Inertial Measurement Unit (IMU).

[0063] Furthermore, the original data stream is input to the state calculation unit, which runs a filtering algorithm (e.g., extended Kalman filter) to fuse the multi-source data and calculate a stable and frequently updated six-degree-of-freedom state vector for the physical entity.

[0064] In this embodiment, the virtual entity is driven to move in the three-dimensional digital scene in real time through the six-degree-of-freedom state vector. Specifically, the six-degree-of-freedom state vector is integrated with the corresponding timestamp, and the geodetic coordinates in the six-degree-of-freedom state vector are converted into virtual coordinates in the three-dimensional digital scene to generate final pose data. Based on the final pose data, the transformation matrix of the virtual entity is updated so as to drive the synchronous refresh of the pose of the virtual entity in the three-dimensional digital scene through the transformation matrix, thereby realizing the movement of the virtual entity in the three-dimensional digital scene.

[0065] In this embodiment, after converting the geodetic coordinates in the six-degree-of-freedom state vector into virtual coordinates in the three-dimensional digital scene, the coordinate sequence formed by the virtual coordinates is smoothed by a preset interpolation algorithm or extrapolation algorithm to compensate for network jitter.

[0066] In one specific embodiment, the calculated state vector and timestamp are serialized together using a predefined data protocol (such as Protobuf) to form a data packet, and then sent to the simulation control back-end subsystem via a low-latency communication protocol (such as UDP).

[0067] In one specific embodiment, after receiving the data packet, the simulation control back-end subsystem deserializes it and performs a coordinate system transformation, converting the geodetic coordinates in the state vector into world coordinates in the 3D visualization scene. To compensate for network jitter, interpolation or extrapolation algorithms can be further used to smooth the coordinate sequence.

[0068] In one specific embodiment, the transformation matrix of the corresponding virtual entity in the digital twin scene is updated using the pose data obtained from the final processing, thereby driving the synchronous refresh of its pose in the 3D rendering engine and completing a precise mapping of the virtual and real states.

[0069] In this embodiment, the channel parameters of the target network are updated in real time based on the virtual state data. Specifically, by using a preset ray tracing-based channel model and combining it with the virtual state data, the influence of the virtual entity's movement on the network signal is simulated, and the channel parameters of the target network are updated based on the influence.

[0070] In one specific embodiment, a parameter mapping interface is constructed from the simulation control backend to the virtual network simulator. This interface transforms the state of virtual entities in the scene (such as their relative position and attitude with respect to the base station) into precise channel state parameters (such as path loss, latency, and bandwidth), and updates these parameters in the virtual network in real time. The channel model preferably employs a method based on ray tracing or combined with AI channel generation tools such as Sionna to more accurately simulate complex electromagnetic environments.

[0071] Specifically, the process of converting the state of virtual entities in the scene (such as their relative position and attitude with respect to the base station) into precise channel state parameters (such as path loss, delay, and bandwidth) involves calling the channel model to calculate real-time parameters such as signal path loss and shadow fading based on the relative position and distance between the virtual drone and each model, as well as the building occlusion relationship.

[0072] S103, using the channel parameters, drive the operation of the target network and continuously monitor the performance data of the target network. Based on the performance data, optimize the target network to achieve network simulation.

[0073] In one specific embodiment, the simulation process is initiated, and the virtual network operates based on real-time updated channel parameters. Its performance indicators (such as throughput and packet loss rate) are continuously monitored and fed back to the simulation control backend. The backend pushes this performance data to the 3D visualization frontend, where it is presented intuitively by changing visual elements such as link color and particle velocity.

[0074] In this embodiment, new network configuration commands are dynamically generated based on the performance data. Specifically, the performance data is fused with pre-acquired context data to construct a comprehensive state vector. The performance data includes the current speed and service type of the virtual entity. The comprehensive state vector is input into a preset reinforcement learning model so that the reinforcement learning model generates action instructions for adjusting the network state according to a preset optimization objective function. The action instructions are parsed into one or more network configuration commands for the target network.

[0075] In one specific embodiment, the computer data processing flow for implementing the simulation closed loop specifically includes:

[0076] a. The simulation control backend obtains network performance index data in real time from the virtualized 6G network subsystem through API or log parsing, and normalizes it into a structured performance data vector.

[0077] b. Construction of the comprehensive state vector: The performance data vector is fused with contextual information extracted from the digital twin scenario (e.g., the current speed of the virtual entity, business type) to construct a comprehensive state vector, which serves as the input to the intelligent decision agent.

[0078] c. Decision reasoning and action generation: The intelligent decision agent (e.g., a reinforcement learning-based model) receives the comprehensive state vector and, based on a preset optimization objective function, infers and outputs an action instruction for adjusting the network state.

[0079] d. Command parsing and conversion: The simulation control backend receives and parses the action command, and converts it into one or more specific configuration commands for the virtualized 6G network subsystem that conform to its interface specifications (e.g., modifying parameters in the YAML configuration file or generating a gRPC request).

[0080] e. Configuration Distribution and Closed-Loop Implementation: The configuration commands are distributed to the virtualized 6G network subsystem via a standardized communication interface, causing changes to its network behavior. This change affects the network performance metrics at the next time step, thus forming an automated control closed loop encompassing state observation, decision reasoning, action execution, and state feedback, effectively solving the problem of static simulation processes in existing systems.

[0081] Preferred, such as Figure 3 As shown, this embodiment of the invention provides a network simulation system based on digital twins, which establishes a data link between physical entities, virtual networks, and upper-layer intelligent applications. The real-time state of the physical entity (e.g., drone location) is continuously fed back to the digital entity through virtual-physical mapping (interface 6). Dynamic changes in the physical world directly affect the channel conditions of the digital twin network (UERANSIM and Open5gs). The UE backend collects these network parameters and converts them into standardized channel states (interface 3), which are then fed back to the Configurator module, which acts as the upper-layer application. The LangGraph agent within this module generates or optimizes the network configuration (interface 4) based on the real-time network state and user requirements. Finally, the generated configuration (*.yaml file) is imported into the simulator to adjust network behavior (interface 5). Changes in network performance, in turn, affect the decisions of the upper-layer application and drive the physical entity to make corresponding behavioral adjustments through virtual-physical mapping (interface 6), thus forming a complete dynamic closed loop of "physical state affecting the network -> network performance feeding back to the application -> application decision adjusting physical behavior and the network." This closed-loop mechanism makes the simulation process more closely resemble the dynamic interaction logic of the real world.

[0082] This invention establishes a high-fidelity digital twin scene and virtual network corresponding to the physical world by deploying simulations in a preset 3D digital scene. This ensures that the simulation environment originates from the real physical scene, solving the problems of simplified simulation environments and disconnection from physical reality in existing technologies. Through real-time bidirectional mapping between physical and virtual entities (i.e., driving the movement of virtual entities through the state data of physical entities in reality), the simulation input is ensured to be based on dynamic changes in the real world, improving simulation realism. Furthermore, since the propagation of network signals is affected by obstruction or reflection from obstacles, the dynamic changes in the physical world (such as position and attitude) are transformed into network channel parameters (such as path loss and delay) during network simulation, enabling the network simulation to respond to the physical environment and solving the problem of simplified channel models in existing technologies. Network simulation is performed based on real-time channel parameters, and network performance indicators (such as throughput and packet loss rate) are output, providing a data foundation for closed-loop feedback. Then, based on the network performance indicators, network configuration commands are dynamically generated and the network is optimized, realizing intelligent decision-making and adjustment based on network state, forming a dynamic closed loop of "physical-network-application," solving the problems of static simulation processes and lack of interactivity in existing technologies. Compared with existing technologies, this invention can improve the accuracy of network simulation results by constructing a network digital twin environment that interacts with the real physical world in real time.

[0083] Example 2:

[0084] like Figure 4 As shown, this embodiment provides a network simulation device based on digital twins, including a simulation deployment module 201, a channel parameter acquisition module 202, and a simulation network optimization module 203, wherein...

[0085] The simulation deployment module 201 is used to perform initial simulation deployment of the target network on a pre-constructed 3D digital scene containing virtual entities; wherein, the 3D digital scene is constructed according to a preset actual physical scene; and the virtual entities are dynamically constructed according to physical entities moving in the actual physical scene.

[0086] The channel parameter acquisition module 202 is used to acquire multi-source state data of the corresponding physical entity in real time, so as to drive the virtual entity to move in the three-dimensional digital scene in real time through the multi-source state data, and acquire the virtual state data of the virtual entity, so as to update the channel parameters of the target network in real time according to the virtual state data; wherein, the virtual state data includes the relative position and pose of the virtual entity and the preset virtual base station;

[0087] In this embodiment, the channel parameter acquisition module 202 acquires multi-source state data of the corresponding physical entity in real time, so as to drive the virtual entity to move in the three-dimensional digital scene in real time through the multi-source state data. Specifically, the channel parameter acquisition module 202 acquires multi-source state data of the physical entity in real time; wherein, the multi-source state data includes the geographical location, three-axis acceleration and angular velocity of the physical entity; the multi-source state data is fused to calculate the six-degree-of-freedom state vector of the physical entity; wherein, the degree of freedom refers to the direction in which the physical entity can move independently; the virtual entity is driven to move in the three-dimensional digital scene in real time through the six-degree-of-freedom state vector.

[0088] In this embodiment, the channel parameter acquisition module 202 drives the virtual entity to move in the three-dimensional digital scene in real time through the six-degree-of-freedom state vector. Specifically, the channel parameter acquisition module 202 integrates the six-degree-of-freedom state vector with the corresponding timestamp, and converts the ground coordinates in the six-degree-of-freedom state vector into virtual coordinates in the three-dimensional digital scene to generate final pose data. Based on the final pose data, the transformation matrix of the virtual entity is updated so as to drive the synchronous refresh of the pose of the virtual entity in the three-dimensional digital scene through the transformation matrix, thereby realizing the movement of the virtual entity in the three-dimensional digital scene.

[0089] In this embodiment, the channel parameter acquisition module 202 updates the channel parameters of the target network in real time based on the virtual state data. Specifically, the channel parameter acquisition module 202 simulates the impact of the virtual entity's movement on the network signal by combining the virtual state data with a preset ray tracing-based channel model, and updates the channel parameters of the target network based on the impact.

[0090] The simulation network optimization module 203 is used to drive the operation of the target network through the channel parameters, continuously monitor the performance data of the target network, and optimize the target network based on the performance data to achieve network simulation.

[0091] In this embodiment, the simulation network optimization module 203 dynamically generates new network configuration commands based on the performance data. Specifically, the simulation network optimization module 203 fuses the performance data with pre-acquired context data to construct a comprehensive state vector. The performance data includes the current speed and service type of the virtual entity. The comprehensive state vector is input into a preset reinforcement learning model so that the reinforcement learning model generates action instructions for adjusting the network state according to a preset optimization objective function. The action instructions are then parsed into one or more network configuration commands for the target network.

[0092] For a more detailed explanation of the working principle and procedures of this embodiment, please refer to the relevant description in Embodiment 1.

[0093] This invention, through a simulation deployment module 201, performs simulation deployment in a preset 3D digital scene, thereby establishing a high-fidelity digital twin scene and virtual network corresponding to the physical world. This ensures that the simulation environment originates from the real physical scene, solving the problems of simplified simulation environments and disconnection from physical reality in existing technologies. Through a channel parameter acquisition module 202, a real-time bidirectional mapping between physical and virtual entities is established (i.e., the movement of virtual entities is driven by the state data of physical entities in reality), ensuring that the simulation input is based on the dynamic changes of the real world, improving the simulation realism. Furthermore, since the propagation of network signals can be affected by obstacles or reflections, in the network... In the network simulation process, dynamic changes in the physical world (such as position and attitude) are transformed into network channel parameters (such as path loss and delay), enabling the network simulation to respond to the physical environment and solving the problem of simplified channel models in existing technologies. Through the simulation network optimization module 203, network simulation is performed based on real-time channel parameters, and network performance indicators (such as throughput and packet loss rate) are output, providing a data foundation for closed-loop feedback. Then, based on the network performance indicators, network configuration commands are dynamically generated and the network is optimized, realizing intelligent decision-making and adjustment based on network status, forming a dynamic closed loop of "physical-network-application", and solving the problems of static simulation process and lack of interactive capabilities in existing technologies.

[0094] Example 3:

[0095] This embodiment provides a terminal device, including: a processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other through the communication bus;

[0096] The memory is used to store at least one executable instruction that causes the processor to perform operations as described in any of the above-described digital twin-based network simulation methods.

[0097] Example 4:

[0098] This invention provides a computer-readable storage medium including a stored computer program, wherein the computer program, when running, controls the device or apparatus containing the computer-readable storage medium to execute the network simulation method based on digital twins as described above.

[0099] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0100] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A network simulation method based on digital twins, characterized in that, include: An initial simulation deployment of a target network is performed on a pre-constructed 3D digital scene containing virtual entities; wherein the 3D digital scene is constructed based on a preset actual physical scene; and the virtual entities are dynamically constructed based on physical entities moving in the actual physical scene. The system acquires multi-source state data of the corresponding physical entity in real time, and drives the virtual entity to move in the three-dimensional digital scene in real time using the multi-source state data. It also acquires virtual state data of the virtual entity, and updates the channel parameters of the target network in real time based on the virtual state data. The virtual state data includes the relative position and pose of the virtual entity and a preset virtual base station. The target network is driven by the channel parameters, and its performance data is continuously monitored. Based on the performance data, the target network is optimized to achieve network simulation.

2. The network simulation method based on digital twins as described in claim 1, characterized in that, Before the initial simulation deployment of the target network in the pre-built 3D digital scene containing virtual entities, the process also includes constructing the 3D digital scene. Specifically, constructing a three-dimensional digital scene involves: Collect multi-source heterogeneous spatial data of the actual physical scene; wherein, the multi-source heterogeneous spatial data includes point cloud data, multi-angle image data and vector map data; The point cloud data is registered to a unified coordinate system, and a preset surface reconstruction algorithm is used to transform the point cloud data into a triangular mesh model with a topological structure to generate the geometric skeleton of the three-dimensional digital scene. By using a pre-defined multi-view stereo matching technology, the multi-angle image data is mapped as texture onto the surface of the geometric skeleton to obtain an initial three-dimensional digital scene. Using a pre-defined 3D rendering engine, semantic and physical attributes of model elements in the initial 3D digital scene are marked based on the vector map data; wherein, the model elements include buildings and streets; and the physical attributes include materials and collision volumes.

3. The network simulation method based on digital twins as described in claim 1, characterized in that, The real-time acquisition of multi-source state data of the corresponding physical entity, and the use of this multi-source state data to drive the virtual entity to move in the 3D digital scene in real time, specifically includes: The system acquires multi-source state data of the physical entity in real time; wherein the multi-source state data includes the physical entity's geographical location, triaxial acceleration, and angular velocity. The multi-source state data is fused to calculate the six-degree-of-freedom state vector of the physical entity; where the degree of freedom refers to the direction in which the physical entity can move independently. The virtual entity is driven to move in the three-dimensional digital scene in real time using the six-degree-of-freedom state vector.

4. The network simulation method based on digital twins as described in claim 3, characterized in that, The virtual entity is driven to move in the 3D digital scene in real time using the six-degree-of-freedom state vector, specifically as follows: The six-degree-of-freedom state vector is integrated with the corresponding timestamp, and the geodetic coordinates in the six-degree-of-freedom state vector are converted into virtual coordinates in the three-dimensional digital scene to generate the final pose data. Based on the final pose data, the transformation matrix of the virtual entity is updated so as to drive the synchronous refresh of the pose of the virtual entity in the three-dimensional digital scene, thereby realizing the movement of the virtual entity in the three-dimensional digital scene.

5. The network simulation method based on digital twins as described in claim 4, characterized in that, After converting the geodetic coordinates in the six-degree-of-freedom state vector into virtual coordinates in the three-dimensional digital scene, the method further includes smoothing the coordinate sequence formed by the virtual coordinates using a preset interpolation algorithm or extrapolation algorithm to compensate for network jitter.

6. The network simulation method based on digital twins as described in claim 1, characterized in that, Based on the virtual state data, the channel parameters of the target network are updated in real time, specifically as follows: By using a pre-defined ray tracing-based channel model and combining it with the virtual state data, the influence of the virtual entity's movement on the network signal is simulated, and the channel parameters of the target network are updated based on the influence.

7. The network simulation method based on digital twins as described in claim 1, characterized in that, Based on the performance data, a new network configuration command is dynamically generated, specifically as follows: The performance data is fused with pre-acquired context data to construct a comprehensive state vector; wherein, the performance data includes the current speed and service type of the virtual entity; The integrated state vector is input into a preset reinforcement learning model so that the reinforcement learning model generates action instructions for adjusting the network state according to a preset optimization objective function. The action instructions are parsed into one or more network configuration commands for the target network.

8. A network simulation device based on digital twins, characterized in that, It includes a simulation deployment module, a channel parameter acquisition module, and a simulation network optimization module, among which, The simulation deployment module is used to perform initial simulation deployment of the target network on a pre-constructed 3D digital scene containing virtual entities; wherein, the 3D digital scene is constructed according to a preset actual physical scene; and the virtual entities are dynamically constructed according to physical entities moving in the actual physical scene. The channel parameter acquisition module is used to acquire multi-source state data of the corresponding physical entity in real time, so as to drive the virtual entity to move in the three-dimensional digital scene in real time through the multi-source state data, and acquire the virtual state data of the virtual entity, so as to update the channel parameters of the target network in real time according to the virtual state data; wherein, the virtual state data includes the relative position and pose of the virtual entity and the preset virtual base station; The simulation network optimization module is used to drive the operation of the target network through the channel parameters, continuously monitor the performance data of the target network, and optimize the target network based on the performance data to achieve network simulation.

9. A terminal device, characterized in that, include: The processor, memory, communication interface, and communication bus are provided, wherein the processor, memory, and communication interface communicate with each other via the communication bus. The memory is used to store at least one executable instruction that causes the processor to perform the operation of the network simulation method based on digital twins as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein, when the computer program is executed, it controls the device or apparatus containing the computer-readable storage medium to perform the network simulation method based on digital twins as described in any one of claims 1 to 7.

Images