Digital twin entity for replicating a physical wireless network

Abstract

The invention relates to a digital twin entity (110) for replicating a physical wireless network (120), which can be represented as a plurality of network nodes (121a, 121b) and interactions (122) between the network nodes (121a, 121b) in a heterogeneous structure (123) with different types of nodes and edges. The digital twin entity comprises an interface (130) to the physical wireless network for receiving information about the network nodes and the interactions between the network nodes of the physical wireless network. The information can be represented as a plurality of heterogeneous structures derived from measurements of the network nodes, each heterogeneous structure representing a structure of the physical wireless network taken by a respective network node. The digital twin entity comprises a service mapping model (SMM) subsystem (140) for providing a replica of the physical wireless network based on an artificial intelligence / machine learning (AI / ML) model (141). The AI / ML model is trained using the heterogeneous structures derived from the measurements of the network nodes. The AI / ML model is used to generate a digital representation (702) for each network node of the physical wireless network.

Description

Technical Field

[0001] This invention relates to the fields of artificial intelligence (AI) and machine learning (ML) applied to mobile networks such as 3GPP systems. Specifically, this invention relates to a digital twin entity for replicating a physical wireless network, and a method for realizing a digital twin model of a wireless access network based on heterogeneous structural information. Background Technology

[0002] Due to the high mobility of users, mobile networks (such as 3GPP systems) are inherently dynamic. To achieve the aforementioned network applications, the use of digital twin models for network management has been proposed to reduce OAM costs. Such digital twin models for wireless networks need to be accurate and efficient, ensuring real-time synchronization between the model and events occurring within the wireless network. To quickly replicate wireless network behavior and thus successfully build a digital twin, the digital twin model needs to be accurate, fast, and adaptable. Furthermore, wireless nodes may appear suddenly. In such cases, the digital twin model should accommodate these different network scales, reducing the number of model updates while handling increases in wireless network scale, for example, when the mobile network experiences high load during peak hours. The digital twin instance should be scalable to adapt to the ever-changing complexity of the physical wireless network. Currently, there is a lack of digital twin models for radio access networks that are both accurate and computationally efficient. Digital twin models need to be computationally efficient to securely and quickly evaluate various what-if scenarios and configurations in a real-time closed-loop manner. Summary of the Invention

[0003] This invention provides a scheme for a fast, accurate, and scalable digital representation for physical wireless networks.

[0004] The above and other objectives are achieved through the features of the independent claim. Other implementations will be apparent from the dependent claims, the description, and the drawings.

[0005] This invention proposes a digital twin entity for replicating a physical wireless network, which can be represented as multiple network nodes and interactions between network nodes in a heterogeneous structure with different types of nodes and edges.

[0006] Digital twin entities are fast (i.e., computationally efficient) in replicating network behavior and generating digital representations of wireless networks. These capabilities enable the evaluation of numerous what-if wireless network scenarios / configurations in a short time. Due to computational efficiency, multiple what-if wireless network scenarios / configurations can be rapidly evaluated. Real-time closed-loop network optimization can be achieved using digital twin models, allowing more optimization solutions to be transmitted to the physical wireless network.

[0007] The digital twin entity is accurate in reflecting wireless system behavior at both the network and network node levels because it has the ability to accurately estimate key performance indicators (KPIs) for each wireless node (e.g., downlink data rate per user, cell throughput per BS, etc.). Furthermore, the model can accurately replicate network behavior under various dynamics and characteristics of the wireless network (e.g., changes in UE state, different functions, different radio phenomena). Some examples of wireless network dynamics include, but are not limited to, handover (HO), changes in UE scheduling frequency resources, changes in radio channels, changes in inter-cell interference, and cell load fluctuations due to mobility.

[0008] Digital twin entities are scalable to account for different wireless network topologies / scenarios (e.g., wireless network deployments / loads not visible during training) while reducing model updates and performance degradation. They are also capable of handling increased wireless network scale (e.g., dense small cell deployments). In terms of implementation in real wireless systems (e.g., 5G systems), digital twin instances can scale to multiple requests from multiple network applications in the RAN, CORE (e.g., 5G SBA), and / or end users for seamless integration in service-based architectures.

[0009] The digital twin entity proposed in this invention can be deployed as one or more digital twin instances in existing or new network functions (NFs) within the RAN / Core of a wireless network. The digital twin entity or model can be based on a 5G service-based architecture (SBA) by exposing public digital twin services to other NFs. In this way, the digital twin instance can be self-contained and loosely coupled with other NFs, thereby facilitating its deployment without affecting existing NFs in the RAN and Core. The digital twin instance is scalable, allowing multiple network applications (via the northbound interface) to request the digital twin service.

[0010] The digital twin entity proposed in this invention can integrate AI / ML for in-network optimization within the RAN. The digital twin can provide services for two general network applications: (i) what-if performance analysis, and (ii) generating synthetic data for AI / ML model training to support envisioned use cases. Other currently undeveloped network applications are expected to benefit from the digital twin of the wireless network.

[0011] The proposed solution in this invention is based on the idea that wireless network nodes and the underlying wireless phenomena between them are represented as different types of nodes and interactions in a heterogeneous structure (e.g., a heterogeneous graph). The heterogeneous structure is fed as samples into a heterogeneous graph neural network (HGNN) model so that the model learns to simulate the underlying wireless phenomena. The topology of the input heterogeneous structure is constructed based on wireless measurements of network nodes. The input heterogeneous structure defines the computational graph of the HGNN-based model. By monitoring structural changes in the physical wireless network, the computational graph of the digital twin model is synchronized with the structure of wireless phenomena in the physical wireless network. Structural changes refer to changes in the interactions between network nodes (e.g., user handover (HO), changes in inter-cell interference, etc.). In the HGNN model, different types of wireless interactions are used in the computation of the input heterogeneous structure signals, each with different independent trainable parameters. These trainable parameters are shared among nodes interacting with the same wireless phenomena (e.g., wireless communication and inter-cell interference).

[0012] These characteristics of HGNN-based digital twin models enable them to achieve significantly higher computational efficiency and accuracy than unstructured or isomorphic structured models in inferring network behavior, and they can generalize better to different network topologies. After training the HGNN-based digital twin model using a specific learning process (e.g., self-supervised / contrastive graph representation learning), the model generates a general wireless vector representation (GWVR) for each network node across all node types. GWVR is inspired by word embeddings in natural language processing. GWVR can be provided as input for any network application (e.g., performance in what-if scenarios, network prediction, QoS prediction, network planning, AI / ML model training).

[0013] To realize this concept, a mechanism and several functionalities are proposed to create and tune wireless network models based on heterogeneous graph neural networks that generate universal wireless vector representations. The features of this implementation are as follows: The logical data instruction entity (DIE) functions as described below. These configurable instructions, known as local neighborhood construction rules (LNCRs), are provided to all wireless nodes in the physical wireless network to monitor relevant changes in the local wireless situation. Instructions can be configured by the network or, for example, by OAM. The heterogeneous local wireless structure (HLWS) is constructed based on the local wireless situation for each type of wireless interaction. An HLWS consists of a given wireless node's set of neighbors and the wireless interactions between the node and its neighbors. Wireless nodes register with the DIE, and this entity manages the registration information types defined below.

[0014] A set of local neighborhood construction rules (LNCRs) for each node type, as described below, is used to monitor relevant changes in the physical wireless network and notify these changes to enable the creation / adjustment of the digital twin model residing in the digital twin instance. Since different types of local information are available on each type of wireless node, the LCNR is defined for each node type, and a rule entry is defined for each type of wireless interaction. For each rule entry in the LCNR, after an event is triggered, the action is to notify the digital twin instance of the wireless network, as described in Section 2.2. The wireless node notifies the digital twin instance of relevant wireless interaction changes in the wireless network.

[0015] The Structure Assemble Entity (SAE) function, as described below, aggregates local structural changes reported by all wireless nodes in a physical wireless network for each type of wireless interaction to construct a heterogeneous global wireless structure (HGWS), representing the structure of the wireless network at a given instance. The aggregated HGWS is provided as input samples to one or more functional models (e.g., HGNN models) within one or more service mapping models (described below). Using the HGNN-based model as the functional model, the structure of the input HGWS aggregated by the SAE defines the computational graph of the HGNN-based model.

[0016] Messages used to adapt the digital twin model to changes in the wireless network and provide a digital representation for network applications are required. To achieve this, messages need to be exchanged through the southbound interface (SI) (digital twin instance and physical wireless network) and the northbound interface (NI) (digital twin instance and network application). Wireless nodes register with the DIE, providing local information, including but not limited to one or more available local measurements for each node, node ID, and node type. Subsequently, the DIE transmits the LNCR to the wireless node based on its node type and available local measurements. The node begins measurements, and once an event from the associated LCNR is triggered, the wireless node transmits HLWS to the SAE to specify localized changes regarding the parties to the radio interaction (with source and destination node IDs, e.g., source BS ID and destination UE ID in downlink communication) and the radio interaction itself (with radio interaction type ID, as described below). The SAE aggregates all HLWS transmitted through the wireless nodes in the network to create the HGWS. The HGWS (connected by the feature information of nodes and edges) is provided as input samples to one or more functional models (e.g., HGNN models) within one or more service mapping models (described below). Based on the HGWS provided as input samples, the SMM generates a general wireless vector representation (GWVR) for each node in the HGWS (e.g., two different GWVR types for UE a and UE b). UE and a GWVR type for BS x BS These GWVRs should be provided to network applications via NI upon request. GWVRs can be used in any network application. Network applications include, but are not limited to: performance in what-if network scenarios, network prediction, network optimization, network planning, AI / ML model training, etc.

[0017] To describe the invention in detail, the following terms, abbreviations and symbols will be used: 5G (fifth-generation mobile network) LTE Long Term Evolution NR New Radio User Equipment (UE) BS base station RAN (Radio Access Network) NF network function AF application function HGNN (Heterogeneous Graph Neural Network) GNN (Graph Neural Network) HG heterogeneous graph DIE Data Instruction Entity LNCR constructs local neighborhood construction rules for each node type. HLWS (Heterogeneous Local Wireless Structure) HGWS (Heterogeneous Global Wireless Structure) SAE structure assemble entity SMM service mapping model DTEM Digital Twin Entity Management DR data repository NI Northbound Interface SI Southbound Interface GWVR General Wireless Vector Representation OAM operation, administration and management Deep reinforcement learning (DRL) Heterogeneous graph neural network (HGNN): as defined in "T. Kipf: Deep Learning with Graph Structured Representations," PhD dissertation, University of Amsterdam, 2020. The relation (in the reference) refers to heterogeneity.

[0018] Heterogeneous graph (HG): A data structure consisting of nodes of different types and edges (also of different types) between said nodes, as well as feature information connecting these nodes and edges. In an HG, different types of information (i.e., different data types and dimensions) are connected to nodes and edges. In this invention, user equipment (UE) and base station (BS) are represented in the HG by nodes of different types, which have different types of local information (e.g., location information of the UE and resource block (RB) utilization of the base station). Furthermore, these wireless nodes interact with each other in different ways; for example, the BS communicates with the UE on the downlink while also interfering with other UEs connected to neighboring BSs. This is caused by different types of edges ( Wireless communication interaction type and No interference Line interaction type This reflects the direction of edges in HG, meaning that nodes between edges can be either source or destination nodes, depending on the edge's direction. These edges indicate that different independent trainable parameters are computed in the learning model for each type of wireless interaction (see below for details).

[0019] Data Instruction Entity (DIE): The DIE is a logical entity whose function is to define configurable rules, called local neighborhood construction rules (defined below), for each node type (e.g., UE and BS in a wireless network). These rules are sent to the nodes. One rule entry is defined for each type of wireless interaction (see the Implementation section for details). The goal of these rules is to track relevant structural changes in the physical wireless network (e.g., handover (HO), changes in inter-cell interference, etc.) and trigger notifications to the digital twin instance of the wireless network to send these changes. The DIE manages registration information from wireless nodes (e.g., ID, node type, one or more available local node measurements for each node, etc.), which constitutes part of the digital twin process of the wireless network. Wireless nodes register with the DIE, and after registration, the DIE transmits a set of configurable instructions for each node type, which is called local neighborhood construction.

[0020] Wireless Node Type: This refers to the classification of wireless nodes based on their functions (e.g., a UE allows users to access radio access network services, a BS is the central connection point for wireless devices to communicate in the radio access network, a RIS is a programmable structure that controls the wireless propagation of electromagnetic waves, a network repeater is a repeater that extracts data from received signals, applies noise correction techniques, and retransmits the signal, etc.). Classification can also include finer-grained distinctions between wireless nodes. Different types of wireless nodes have different available local node measurements (e.g., a UE measures downlink channel quality from a nearby BS, such as received signal reference power (RSRP) and signal-to-interference ratio (SINR), and can obtain location information from end-user applications; a UE measures sidelink channel quality, such as received signal strength indicator (RSSI); a BS measures resource block (RB) utilization, cell load (based on the number of connected UEs), uplink channel quality, etc.).

[0021] For each node type, the Local Neighborhood Construction Rule (LNCR) is a set of rules for input network parameters (e.g., as mathematical expressions, event-based triggers, conditional statements, etc.) that enables the detection of relevant interaction changes between wireless nodes based on local measurements or local network events (see the Implementations section for details). Since different types of local information are available for each type of wireless node, the LCNR is defined for each node type, and a rule entry is defined for each wireless interaction type. For each rule entry in the LCNR, upon triggering an event, the action is to notify the digital twin instance of the wireless network of relevant changes regarding which wireless node interacted with which other wireless node (i.e., with source and target node IDs) and the form of the wireless interaction (i.e., with a wireless interaction type ID (defined below)).

[0022] Wireless interaction type: This refers to the classification of wireless interactions between wireless nodes based on their characteristics (e.g., downlink communication refers to the transmission of data from the BS to the UE, where the transmitting BS and the receiving UE have specific configurations; downlink inter-cell interference refers to unintended interference caused by the BS to the wireless signals received by a nearby UE connected to an adjacent BS; uplink communication refers to the transmission of data from the UE to the BS, where the transmitting UE and the receiving BS have specific configurations; sidelink communication refers to the transmission of data from the transmitting UE to the receiving UE, etc.). The classification can also include finer-grained distinctions between wireless interactions.

[0023] Heterogeneous Local Wireless Structure (HLWS) Changes: Notifications reported by wireless nodes in a wireless network regarding changes to the local structure. A structure change refers to a change in the wireless interactions between wireless nodes. The change in wireless interaction is specified by the source and destination wireless node IDs and the type of wireless interaction that is being changed. Examples of changes in wireless interaction include, but are not limited to: changes in received interference, handover from a source BS to a destination BS, changes in frequency bands or wireless access technologies, and the addition of a secondary wireless link to a secondary service BS. These notifications from wireless nodes in the wireless network are reported to the structure aggregation entity residing in the digital twin instance (defined below).

[0024] Heterogeneous Global Wireless Structure (HGWS): A data structure representing the structure of a wireless network at a given moment. An HGWS is formed by the aggregation of multiple HLWS at the same moment. An HGWS captures wireless interactions between wireless nodes that extend beyond their one-hop local neighborhood. An HGWS represents the structure of a physical wireless network at a given moment. An HWGS includes characteristic information and wireless interactions connected to the wireless nodes.

[0025] Structure assemble entity (SAE): SAE is a logical entity whose function is to aggregate HLWS reported by wireless nodes in a wireless network to construct HGWS (e.g., heterogeneous graph).

[0026] General Wireless Vector Representation (GWVR): GWVR is a low-dimensional, learned, continuous vector representation (i.e., a real-valued number) of a wireless network, generated for each wireless node by a Digital Twin Functional Model (i.e., HGNN). GWVR captures local structural information and local and neighborhood feature information for each type of wireless interaction. GWVR is defined for each wireless node, for each edge (as a function of the GWVR for the source and target nodes), and for each graph (as a function of the GWVR for all wireless nodes within the HGWS). Different types of GWVR (e.g., with different dimensions) are generated for each node type and edge type (e.g., UE GWVR and BS GVR). GWVR is applicable to any network application (defined in Section 3.5). For example, GWVR can be used as input to other functions to calculate wireless network KPIs (e.g., downlink rate of UE wireless nodes or cell throughput of BS wireless nodes).

[0027] A digital twin process refers to the process by which a wireless node periodically sends local and structural information to a digital twin instance (defined in Section 3.5), and the instance internally replicates the behavior of the wireless node. Network applications (defined in Section 3.5) can request digital representations of different parts of the physical wireless network in the form of GWVR from the digital twin instance. The instance efficiently provides GWVR in real time, including alternative what-if scenarios. In the case of closed-loop control, the digital twin instance can send control information to the wireless node to execute control.

[0028] Functional mapping to phenomena refers to a mapping that takes GWVR as input and outputs real-world phenomena parameters (e.g., downlink rate, latency jitter, uplink packet error rate, uplink spectral efficiency, etc.). These phenomena are measurable in the physical wireless network. Therefore, the functional mapping allows network applications to use GWVR to evaluate the results of different what-if network configurations.

[0029] According to a first aspect, the present invention relates to a digital twin entity for replicating a physical wireless network, the physical wireless network being representable as a plurality of network nodes in a heterogeneous structure having different types of nodes and edges and interactions between the network nodes, the digital twin entity comprising: an interface to the physical wireless network for receiving information about the network nodes of the physical wireless network and the interactions between the network nodes, the information being representable as a plurality of heterogeneous structures derived from measurements of the network nodes, each heterogeneous structure representing a structure of the physical wireless network acquired by a corresponding network node; and a service mapping model (SMM) subsystem for providing a copy of the physical wireless network based on an artificial intelligence and / or machine learning (AI / ML) model trained using the heterogeneous structures derived from the measurements of the network nodes; wherein the AI / ML model is used to generate a digital representation for each network node of the physical wireless network.

[0030] This digital twin entity enables a fast, accurate, and scalable digital representation of physical wireless networks.

[0031] In one exemplary implementation of the digital twin entity, the AI / ML model includes a heterogeneous graph neural network (HGNN).

[0032] By using this HGNN, models with different types of nodes and edges can be applied, which can accurately replicate the structure of a physical wireless network with different types of base stations (BS) and user equipment (UE) and the connections between them.

[0033] In one exemplary implementation of the digital twin entity, an AI / ML model is used to generate a generic wireless vector representation for each network node of the physical wireless network.

[0034] Such GWVRs are easy to compute because they are low-dimensional, learned continuous vector representations (i.e., real-valued numbers) of wireless networks, where the GWVR is generated for each wireless node by a digital twin functional model (i.e., HGNN). Through such GWVRs, local structural information as well as local and neighborhood feature information can be efficiently captured for each type of wireless interaction.

[0035] In one exemplary implementation of the digital twin entity, the digital twin entity includes: a digital twin entity management subsystem for monitoring the performance and resource consumption of the AI / ML model.

[0036] This digital twin entity management subsystem allows for efficient monitoring of the performance and resource consumption of AI / ML models.

[0037] In one exemplary implementation of the digital twin entity, the digital twin entity includes: a data repository for collecting and storing information about the network nodes of the physical wireless network and the interactions between the network nodes by collecting and updating real-time data of the network nodes through the interface with the physical wireless network.

[0038] This data repository can store all relevant information used to replicate the physical wireless network, and this information can be used for further processing.

[0039] In one exemplary implementation of the digital twin entity, the digital twin entity includes: an interface with a network application system for receiving digital twin service requests from multiple network applications and providing the services of the digital twin entity to the multiple network applications.

[0040] Through this interface, web applications can easily request services from digital twin entities.

[0041] In one exemplary implementation of the digital twin entity, the digital twin service request includes at least one of the following: a request for at least one network node identifying the physical wireless network, the digital representation of which is requested by a network application; a request for a timestamp of the digital representation; optional information including at least one of a phenomenon ID and one or more what-if network configuration IDs; wherein the service of the digital twin entity includes at least one of the following: a digital representation of the at least one network node of the physical wireless network, specified in the digital twin service request; a timestamp of the digital representation; optional information including at least one of the following: a mapping from the representation of the phenomenon ID to the phenomenon function; and optimal network configuration.

[0042] The phenomenon ID is optional information in a digital twin service request. The phenomenon ID is associated with the representation-to-phenomenon functionality mapping defined above (see "Representation-to-phenomenon functionality mapping"), as detailed below. Figure 1 and Figure 2 As described.

[0043] Another optional piece of information in a digital twin service request is one or more What-if network configuration IDs and the expected What-if network configuration / state to be evaluated for each What-if network configuration ID, including the expected future state / configuration (e.g., one or more alternative UE locations, alternative BS deployments and / or resource scheduling policies, alternative cell load, etc.), as described below. Figure 1 and Figure 2 As described.

[0044] Optional information in digital twin service messages includes: a mapping from the phenomenon ID in the service request to the phenomenon function, for example, as described below. Figure 1 and Figure 2 As described. Another optional piece of information in the digital twin service message includes the optimal network configuration for the optimization actions specified in the service request, for example, as described below regarding... Figure 1 and Figure 2 As described.

[0045] This information allows for the easy and efficient identification of services that use digital twin entities.

[0046] In one exemplary implementation of the digital twin entity, the digital twin entity includes: a data instruction entity (DIE) for defining a set of configurable rules for each node type of the network node; wherein the DIE is used to register the network node of the physical wireless network, and transmit the set of configurable rules to the network node after registration.

[0047] This DIE entity or function can define a set of configurable instructions to collect local structure information from the wireless node as HLWS.

[0048] In one exemplary implementation of the digital twin entity, the configurable rule set is a local neighborhood construction rule (LNCR), which enables the detection of changes in interactions between network nodes based on local measurements or local network events of the network nodes.

[0049] This LNRC approach enables efficient monitoring of relevant structural changes in physical wireless networks for each type of wireless interaction, and can notify these HLWS changes to facilitate the creation / adjustment of digital twin models residing in digital twin instances. Compared to unstructured models, model adaptation allows for more efficient and accurate computation to infer network behavior.

[0050] In one exemplary implementation of the digital twin entity, the DIE is configured to receive local information from the network node, the local information including at least one of the following: an identifier of the corresponding network node; the node type of the corresponding network node; available measurements of the corresponding network node; and the communication functions of the corresponding network node.

[0051] This enables efficient aggregation of changes in HLWS reported by wireless nodes in a wireless network.

[0052] In one exemplary implementation of the digital twin entity, the digital twin entity includes: a structure assemble entity (SAE) for constructing a heterogeneous global wireless structure (HGWS) representing the structure of the physical wireless network at a given time, based on notifications of local structural changes reported by the network nodes. The SAE is used to provide the HGWS to the SMM subsystem to enable the AI / ML model to adapt to the local structural changes of the physical wireless network.

[0053] This enables the aggregation of changes in the HLWS reported by wireless nodes in a wireless network to construct the HGWS.

[0054] Changes in the local structure of a physical wireless network are also known as changes in heterogeneous local wireless structure (HLWS).

[0055] In one exemplary implementation of the digital twin entity, the notification by which the network node reports the local structure change of the physical wireless network includes at least one of the following: a wireless interaction type ID, a source wireless node, a target wireless node, and an optional trigger signal that triggers the local structure change report of the network node.

[0056] The trigger signal indicates a trigger measurement (e.g., a downlink RSRP received at the target UE from a new source serving BS during an HO event, an inter-cell interference measurement received at the target UE from a source interfering BS, a side link RSSI received at the target UE from a source UE, etc.).

[0057] The local structural changes of the physical wireless network, namely the changes in HLWS, are transmitted from the network node to the SAE (residing in the digital twin entity).

[0058] In one exemplary implementation of the digital twin entity, the digital twin entity is implemented in the 3GPP radio access network (RAN) entity of the physical wireless network.

[0059] Digital twin entities can be flexibly implemented in different entities within the 3GPP radio access network.

[0060] In one exemplary implementation of the digital twin entity, the SMM subsystem and the SAE entity are implemented within the data analysis function component of the 3GPP radio access network (RAN) entity of the physical wireless network. The DIE entity is implemented within the operation and maintenance (OAM) entity of the 3GPP core network.

[0061] This allows for flexible implementation. Parts of the digital twin entity can be implemented in the RAN, while other parts can be implemented in the core network.

[0062] In one exemplary implementation of the digital twin entity, the SMM subsystem and the SAE entity are implemented in the data analysis function component of the 3GPP core network; the DIE entity is implemented in the operation and maintenance (OAM) entity of the 3GPP core network.

[0063] This supports flexible implementation. Digital twin entities can be implemented in the core network, specifically distributed across different entities within the core network.

[0064] In one exemplary implementation of the digital twin entity, the digital twin entity is used in a mobile edge computing (MEC) server connected to the physical wireless network.

[0065] This allows for flexible implementation. Digital twin entities can be implemented within a single MEC server connected to a physical wireless network.

[0066] According to a second aspect, the present invention relates to a method for replicating a physical wireless network via a digital twin entity, wherein the physical wireless network can be represented as a plurality of network nodes in a heterogeneous structure having different types of nodes and edges and interactions between the network nodes, the method comprising: receiving information about the network nodes of the physical wireless network and the interactions between the network nodes, the information being represented as a plurality of heterogeneous structures derived from measurements of the network nodes, each heterogeneous structure representing a structure of the physical wireless network acquired by a corresponding network node; providing a copy of the physical wireless network based on an artificial intelligence and / or machine learning (AI / ML) model trained using the heterogeneous structures derived from measurements of the network nodes; and generating a digital representation for each network node of the physical wireless network using the AI / ML model.

[0067] This method of replicating physical wireless networks through digital twin entities enables a fast, accurate, and scalable digital representation of physical wireless networks. Attached Figure Description

[0068] Other embodiments of the present invention will be described in conjunction with the following drawings, wherein: Figure 1 A schematic diagram of the system architecture of the wireless network system 100 with digital twin entity 110 provided by the present invention is shown; Figure 2 A schematic diagram showing the system architecture of the wireless network system 100 and the digital twin entity 110 in more detail is provided. Figure 3 An exemplary message diagram 300 is shown, illustrating message exchange between entities involved in the wireless network system 100; Figure 4a , Figure 4b and Figure 4c An exemplary network diagram is shown, illustrating the creation of a new wireless node in an HGWS converged in SAE; Figure 5 An exemplary message diagram 500 provided in the first embodiment is shown, illustrating message exchange with a digital twin instance in a gNB; Figure 6 An exemplary 3GPP wireless network system 600 provided in the first embodiment is shown, wherein a complete digital twin instance is implemented in the RAN entity; Figure 7 An implementation example of a physical wireless network 120, represented as a heterogeneous architecture 123, is shown; Figure 8An implementation example of the heterogeneous structure 123 of the digital twin model and its associated heterogeneous computation graph 123b is shown; Figure 9 An exemplary message diagram 900 provided in a sub-implementation of the first embodiment is shown, which illustrates message exchange with the SMM / SAE at RANDAF in the RAN and the DIE at OAM in the core network; Figure 10 An exemplary 3GPP wireless network system 1000 provided in a sub-implementation of the first embodiment is shown, wherein a portion of the digital twin instance is implemented in the RAN entity, and the DIE is implemented at the OAM function in the core network; Figure 11 An exemplary message diagram 1100 provided in the second embodiment is shown, which illustrates the message exchange between the SMM / SAE at the NWDAF in the core network and the DIE at the OAM in the core network; Figure 12 An exemplary 3GPP wireless network system 1200 provided in the second embodiment is shown, wherein a portion of the digital twin instance is implemented in NWDAF, and the DIE is implemented at the OAM function in the core network; Figure 13 An exemplary message diagram 1300 provided in the third embodiment is shown, illustrating message exchange with the SMM / SAE / DIE at mobile edge computing (MEC); Figure 14 An exemplary 3GPP wireless network system 1400 provided in a third embodiment is shown, wherein a digital twin instance is implemented at the MEC. Detailed Implementation

[0069] In the following detailed description, reference is made to the accompanying drawings, which form a part of this specification, illustrating specific aspects in which the invention can be practiced. It should be understood that other aspects may be utilized and structural or logical changes may be made without departing from the scope of the invention. Therefore, the following detailed description should not be construed as limiting, and the scope of the invention is defined by the appended claims.

[0070] It should be understood that the notes relating to the described methods also apply to the corresponding devices or systems used to perform the methods, and vice versa. For example, if a specific method step is described, the corresponding device may include a unit to perform the described method step, even if such a unit is not explicitly described or shown in the figures. Furthermore, it should be understood that features of the various exemplary aspects described herein can be combined with each other unless otherwise explicitly stated.

[0071] Figure 1A schematic diagram of the system architecture of a wireless network system 100 with a digital twin entity 110 provided by the present invention is shown.

[0072] The wireless network system 100 consists of three main system components: a physical wireless network 120, a wireless network digital twin instance 110 (also referred to herein as digital twin entity 110 or digital twin instance), and a network application 180. These system components are... Figure 1 (Also in Figure 2 The details are explained in the Chinese text and described below.

[0073] Physical wireless network: A wireless system operating using multiple radio access technologies (RATs) consists of multiple types of radio access nodes (RATs). Figure 1 The network access nodes consist of network nodes 121a. These wireless access nodes include, but are not limited to, macro and micro cells (BS) belonging to different RATs, non-terrestrial networks (NTNs) such as satellites, remote radio heads (RRHs), and intelligent reflecting surfaces (IRS). These wireless access nodes serve UEs with different functions / capabilities (e.g., traditional smartphones, automotive communication systems, Internet of Things (IoT) devices, robot communication systems, actuator communication systems, etc.), different communication functions (e.g., multi-element antenna topology, beamforming, support for different frequency bands, etc.), and running different types of services (e.g., enhanced mobile broadband, extended reality (XR), ultra-reliable and low-latency communication, etc.). Figure 1 Network node 121b in the middle.

[0074] Wireless access node ( Figure 1 Network node 121a) and user equipment ( Figure 1 Network node 121b) can communicate in downlink, uplink, and sidelink. Wireless communication can be performed using different frequency bands. When wireless access nodes use the same resources, they may cause interference to adjacent UEs (or adjacent BSs).

[0075] For the digital twin process, the physical wireless network 120 communicates with the wireless network digital twin instance 110 through the southbound interface (SI) 130. SI 130 is responsible for the information exchange between the wireless network digital twin instance 110 and the physical wireless network 120.

[0076] Examples of digital twins for wireless networks: The wireless network digital twin instance 110 includes three key system sub-components: service mapping model (SMM) system sub-component 140, digital twin entity management (DTEM) system sub-component 150, and data repository (DR) system sub-component 160.

[0077] Service Mapping Model (SMM) 140: SMM 140 is a subsystem within the digital twin instance 110, providing data modeling and data model instances for various network applications. An HGNN-based model (also known as AI / ML model 141) resides within SMM subcomponent 140, and it generates a generic vector representation for each network node 121a, 121b in the wireless network 120 (see below for details). A functional model and a basic model reside within SMM subcomponent 140. The functional model refers to the model of the physical wireless network 120 used for specific functions (e.g., status monitoring, traffic analysis, fault diagnosis, performance assurance, etc.). The basic model refers to the network topology, environmental information, and operational status of network nodes 121a, 121b of the wireless network 120.

[0078] Digital twin entity management (DTEM) 150: DTEM 150 is a subsystem used to monitor the performance and resource consumption of the individual models residing in SMM 140, visualize the various components of the network digital twin 110, and can be enabled to control the various components of the network digital twin 110.

[0079] Data repository (DR) 160: DR 160 is a subsystem that collects and stores network data to build a digital twin model residing in SMM subcomponent 140 by collecting and updating real-time data from network nodes 121a and 121b in the physical wireless network 120.

[0080] Northbound interface (NI) 170: The NI 170 is a network application-oriented interface between the wireless network digital twin 110 and the network application 180. The NI 170 is responsible for information exchange between the wireless network digital twin 110 and the network application 180.

[0081] Southbound interface (SI) 130: SI 130 is the physical twin-oriented interface between the wireless network digital twin 110 and the physical wireless network 120. SI 130 is responsible for information exchange between the wireless network digital twin 110 and the physical wireless network 120.

[0082] Network Applications Network application system component 180 refers to various applications, including but not limited to network optimization, network prediction, performance in what-if wireless scenarios, network planning, network management / coordination, and AI / ML model training. These network applications can be requested by different entities in the network, including but not limited to operation, administration and maintenance (OAM) functions, application functions (AF), NF, BS, AMF, SMF, UPF, etc.

[0083] Names, functions, and structures related to this invention The above defines new functions and structures related to this invention.

[0084] The process of creating and adapting one or more digital twin models of wireless network 120 and sending one or more representations to network application 180 may include the following seven steps: Creation process ( Figure 4a (An example of creating a new wireless node in HGWS in SAE is shown) Step 1: A new wireless node 121b appears in wireless network 120. Wireless nodes in the network are participating in the digital twin process as described above. New wireless nodes 121a and 121b in the physical wireless network 120 register with the DIE via SI 130 to form part of the digital twin process. Wireless nodes 121a and 121b provide local information to the DIE, including but not limited to: Wireless Node ID The wireless node types defined above As defined above, measurements can be performed using local nodes. Node communication functionality (optional: if it is a UE, provide one or more configured QoS requirements associated with its PDU session. Other node-related information available in other NFs can also be provided to the DIE).

[0085] Step 2: During the initial registration of wireless nodes during the initial creation of one or more digital twin functional models, DIE210 (see...) Figure 2The LNCR for each node type is specified using user input (e.g., OAM). Alternatively, the LCNR can be dynamically provided by DTEM 150. DIE 210 specifies the number of LNCRs (i.e., one LNCR per wireless node type) and the wireless interaction type (i.e., one LCNR rule entry per wireless interaction type) based on available local node measurements and different wireless node types. Using the provided available local node measurements, DIE 210 specifies configurable events that track the topology of the local wireless situation for each wireless interaction type (i.e., rule entries for a specific LNCR). DIE 210 transmits the LNCR for each node type to wireless nodes 121a and 121b via SI 130. The LCNR defined in DIE 210 can be defined dynamically or using user input with OAM functionality.

[0086] Adjustment process ( Figure 4b The HGWS in SAE 220 is shown when the wireless node changes state (see [link]). Figure 2 Adjustments to ( ). Step 3: Wireless nodes 121a and 121b configure internal processes, begin measurement, and track events according to the provided LNCR.

[0087] Step 4: Network events trigger changes to the HLWS on the wireless nodes based on rule entries / conditional statements specified in the LNCR. Wireless nodes 121a and 121b transmit data to SAE 220 (see...). Figure 2 Send one or more HLWS changes, such as, but not limited to: As defined above, the wireless interaction type ID, Source wireless node, Target wireless node, Optional: Triggering measurements (e.g., the downlink reference signal received power (RSRP) received by the target user equipment (UE) from the new source serving base station (BS) during a handover (HO) event, the inter-cell interference measurement received by the target UE from the source interfering BS, the received signal strength indicator (RSSI) received by the target UE from the source UE, etc.).

[0088] If multiple events occur simultaneously (e.g., an increase in downlink inter-cell interference caused by other adjacent BSs and a simultaneous HO event), the events can be multiplexed and reported in the same HLWS change. The HLWS change is reported to SAE220.

[0089] Step 5: SAE 220 aggregates the HLWS received from wireless nodes 121a and 121b on wireless network 120 and constructs the HGWS of the physical wireless network. SAE 220 transmits the HGWS to the SMM function. Upon receiving the HGWS, the functional model residing in SMM 140, as defined above, updates its computation graph based on the HGWS. When the HGWS is provided as an input sample, the functional model generates a GWVR for each node in the input HGWS. The digital twin process for the new wireless nodes is active.

[0090] Step 6: Digital twin instance 110 is replicating the physical wireless network 120 and awaits a digital twin service request 171 from network application 180 via NI 170. Network application 180 includes, but is not limited to, network optimization, network prediction, performance in what-if wireless scenarios, network planning, network management and orchestration, and AI / ML model training run by other NF / AFs (such as network data analytics function (NWDAF), gNB, UPF, etc.). Network application 180 requests the digital twin service 171 from digital twin instance 110. The digital twin service request 171 includes, but is not limited to: One or more wireless node IDs, for which the network application requests a digital representation (GWVR). GWVR timestamps Optional: The phenomenon ID associated with the representation-to-phenomenon function mapping defined above. Optional: One or more What-if network configuration IDs and the expected What-if network configuration / state to be evaluated for each What-if network configuration ID, including the expected future state / configuration (e.g., one or more alternative UE locations, alternative BS deployments and / or resource scheduling policies, alternative cell loads, etc.). Digital twin instance 110 may also have the ability to generate What-if network configurations / states.

[0091] Optional: Network application 180, such as trial-and-error network optimization (e.g., resource allocation based on reinforcement learning, Bayesian optimization, exhaustive search, etc.), may choose to directly request the optimal network configuration (i.e., the optimization variable) to avoid communication overhead (multiple message requests / responses or subscriptions to exchange representations and optimization actions). The digital twin entity management (DTEM) 150 residing in digital twin instance 110 can perform optimization (referred to as virtual control) using one or more SMM functional models. DTEM 150 provides the optimal network configuration to network application 180. Subsequently, network application 180 can use the digital twin model to control the physical wireless network 120 using the found network configuration.

[0092] Step 7: The SMM 140 residing in digital twin instance 110 provides GWVR to network application 180 in accordance with digital twin service request 171. Digital twin service response 172 includes, but is not limited to: The GWVR specified in Service Request 171 for each radio node ID, optionally: a GWVR for each network configuration ID can be provided based on each radio node ID specified in Service Request 171. By replicating historical information of the physical wireless network based on a digital twin instance, a GWVR for a what-if scenario can be generated.

[0093] GWVR timestamps Optional: Map the phenomenon ID to the phenomenon function based on the phenomenon ID representation in service request 171.

[0094] Optional: DTEM 150 residing in digital twin instance 110 can provide optimal network configuration for the optimization actions specified in service request 171.

[0095] Figure 2 A schematic diagram showing the system architecture of the wireless network system 100 and the digital twin entity 110 in more detail is provided.

[0096] Specifically, Figure 2 The wireless network system 100 in the middle corresponds to Figure 1 The wireless network system 100 has an additional data instruction entity (DIE) 210 and a structure assemble entity (SAE) 220 residing in a wireless network digital twin instance 110 (also referred to herein as digital twin entity 110).

[0097] As mentioned above Figure 1As described, the wireless network system 100 consists of three main system components: a physical wireless network 120, a wireless network digital twin instance 110 (also referred to herein as digital twin entity 110 or digital twin instance), and a network application 180.

[0098] Digital twin entity 110 can be used to replicate physical wireless network 120, which can be represented as multiple network nodes 121a, 121b and interactions 122 between network nodes 121a, 121b in a heterogeneous structure 123 with different types of nodes and edges.

[0099] The digital twin entity 110 includes an interface 130 (also denoted as SI) to the physical wireless network 120 for receiving information about network nodes 121a and 121b of the physical wireless network 120 and interactions 122 between network nodes 121a and 121b. This information can be represented as multiple heterogeneous structures 123 derived from measurements of network nodes 121a and 121b, each heterogeneous structure 123 representing the structure of the physical wireless network 120 as obtained by the corresponding network node 121a and 121b.

[0100] The digital twin entity 110 includes a service mapping model (SMM) subsystem 140 for providing a copy of the physical wireless network 120 based on an artificial intelligence and / or machine learning (AI / ML) model 141. This AI / ML model 141 is trained using a heterogeneous structure 123 derived from measurements of network nodes 121a and 121b.

[0101] AI / ML model 141 is used to generate digital representations 702 for each network node 121a, 121b of the physical wireless network 120 (see details below). Figure 7 ).

[0102] AI / ML model 141 may include heterogeneous graph neural network (HGNN) as described above.

[0103] AI / ML model 141 can be used to generate a general wireless vector representation (GWVR) 702 as defined above for each network node 121a, 121b of physical wireless network 120.

[0104] The digital twin entity 110 may include a digital twin entity management subsystem 150 for monitoring the performance and resource consumption of the AI / ML model 141.

[0105] The digital twin entity 110 may include a data repository 160 for collecting and storing information about the network nodes 121a and 121b of the physical wireless network 120 and the interactions 122 between the network nodes 121a and 121b, by collecting and updating real-time data through an interface 130 with the physical wireless network 120.

[0106] Digital twin entity 110 may include an interface 170 (also referred to as NI) with network application system 180 for receiving digital twin service requests 171 from multiple network applications 181, 182 and providing services 172 of digital twin entity 110 to multiple network applications 181, 182.

[0107] The digital twin service request 171 may include at least one of the following: a request for at least one network node 121a, 121b that identifies the physical wireless network 120, the digital representation 702 of the at least one network node 121a, 121b being requested by the network application; and a request for a timestamp of the digital representation 702.

[0108] The service 172 of the digital twin entity 110 may include at least one of the following: a digital representation 702 of at least one network node 121a, 121b of the physical wireless network 120 specified in the digital twin service request 171; and a timestamp of the digital representation 702.

[0109] The digital twin entity 110 may include a data instruction entity (DIE) 210 for defining a set of configurable rules 211 for each node type of network nodes 121a and 121b. The DIE 210 can be used to register network nodes 121a and 121b of the physical wireless network 120 and, after registration, transmit the set of configurable rules 211 to network nodes 121a and 121b.

[0110] The configurable rule set 211 can be a local neighborhood construction rule (LNCR), which can detect changes in the interaction between network nodes 121a and 121b based on local measurements or local network events of network nodes 121a and 121b.

[0111] DIE 210 can be used to receive local information from network nodes 121a and 121b, the local information including at least one of the following: the identifier of the corresponding network node; the node type of the corresponding network node; the available measurements of the corresponding network node; and the communication functions of the corresponding network node.

[0112] The digital twin entity 110 may include a structure assemble entity (SAE) 220, which is used to construct a heterogeneous global wireless structure (HGWS) representing the structure of the physical wireless network 120 at a given time, based on the notification of local structural changes of the physical wireless network 120 reported by network nodes 121a and 121b.

[0113] SAE 220 can be used to provide HGWS to SMM subsystem 140 to adapt AI / ML model 141 to local structural changes in physical wireless network 120.

[0114] Changes in the local structure of a physical wireless network are also known as changes in heterogeneous local wireless structure (HLWS).

[0115] The notification of local structural changes of physical wireless network 120 reported by network nodes 121a and 121b may include at least one of the following: wireless interaction type ID, source wireless node, target wireless node, and optional trigger signal that triggers the local structural change report of network nodes 121a and 121b.

[0116] The trigger signal indicates a trigger measurement (e.g., a downlink RSRP received at the target UE from a new source serving BS during an HO event, an inter-cell interference measurement received at the target UE from a source interfering BS, a side link RSSI received at the target UE from a source UE, etc.).

[0117] The local structural changes of the physical wireless network, namely the changes in HLWS, are transmitted from the network node to the SAE (residing in the digital twin entity).

[0118] Digital twin entity 110 can be used in the 3GPP radio access network (RAN) entity 610 of physical wireless network 120 (such as... Figure 6 Implemented as shown in the figure.

[0119] SMM subsystem 140 and SAE 220 entities can be used in the data analysis function component 1010 of the 3GPP radio access network (RAN) entity of the physical wireless network 120 (such as... Figure 10 DIE entity 210 can be implemented in the 3GPP core network 1001 operation and maintenance (OAM) entity 1020 (as shown). Figure 10 Implemented as shown in the figure.

[0120] SMM subsystem 140 and SAE 220 entities can be used in the data analysis function component 1030 of the 3GPP core network 1001 (such as... Figure 12 Implemented in (as shown). DIE entity 210 can be used in the operation and maintenance (OAM) entity 1020 of the 3GPP core network 1001 (as shown). Figure 12 Implemented as shown in the figure.

[0121] Digital twin entity 110 can be used in a mobile edge computing (MEC) server 1410 connected to a physical wireless network 120 (such as... Figure 14 Implemented as shown in the figure.

[0122] The present invention also introduces a method for replicating a physical wireless network 120 via a digital twin entity 110 as described above, wherein the physical wireless network can be represented as a plurality of network nodes 121a, 121b and interactions 122 between network nodes in a heterogeneous structure 123 having different types of nodes and edges.

[0123] This method includes receiving information about network nodes and interactions between network nodes in a physical wireless network. This information can be represented as multiple heterogeneous structures derived from measurements of the network nodes, each heterogeneous structure representing the structure of the physical wireless network as obtained by the corresponding network node, as described above.

[0124] This approach includes providing a copy of the physical wireless network based on an artificial intelligence and / or machine learning (AI / ML) model, which is trained using a heterogeneous structure derived from measurements of network nodes as described above.

[0125] This approach involves generating a digital representation for each network node of a physical wireless network using an AI / ML model, as described above.

[0126] Figure 3An exemplary message diagram 300 is shown, illustrating message exchange between entities involved in the wireless network system 100.

[0127] Message diagram 300 shows, for example Figure 2 The digital twin entity 110, including SMM 140, SAE 220 and DIE 210, is shown communicating with network nodes 121a and 121b of a physical wireless network 120 including UE1 to UEn, BS1, BS2 and BS3 and NF / AF.

[0128] Following the initial registration process 301 between DIE 210 and the physical radio networks 120, including UE1 to UEn, BS1, BS2, and BS3, and NF / AF, message 302 is sent from DIE 210 to UE1 to transmit local neighborhood construction rules for each node type. UE1 performs the configuration process 302a and begins measurement 303 using the digital twin entity 110. UE1 then sends message 304 to SAE 220 to trigger a topology update of the local graph for each interaction type. SAE 220 sends message 305 to SMM 140 to notify of the updated global graph. SMM 140 then updates the computation graph 305a.

[0129] After the state in UE1 has changed 305b (e.g., by handover execution), UE1 sends message 305c to SAE 220 to trigger a local neighborhood update for each interaction type. SAE 220 sends message 305d to SMM 140 to notify of the updated global graph. SMM 140 then updates the computation graph 305e.

[0130] NF / AF sends message 306 to SMM 140 to request digital twin services, such as what-if network scenario performance. SMM 140 replies with message 307 to transmit a generic vector representation and an optional output function. NF / AF can then evaluate the performance 307a of the what-if scenario.

[0131] Figure 4a , Figure 4b and Figure 4c An exemplary network diagram is shown, illustrating the creation of a new wireless node in an HGWS that converges in SAE.

[0132] exist Figure 4aIn this context, a new wireless node 121b (UE1) appears in the wireless network 120. The wireless nodes in the network are participating in the digital twin process as described above. The new wireless node 121b of UE1 registers with DIE 210 via SI 130 to form part of the digital twin process. As described above, wireless nodes 121a and 121b provide local information to DIE 210.

[0133] exist Figure 4b The diagram illustrates the adjustment of the HGWS in SAE 220 when the radio node changes state. A state change event occurs here, for example, when UE1 performs a handover. State change events can also be applied to other node types, such as base station 121a. SAE 220 aggregates the HLWS received from radio nodes 121a and 121b on radio network 120 and constructs the HGWS of the physical radio network. SAE 220 transmits the HGWS to the SMM function. Upon receiving the HGWS, the functional model residing in SMM 140, as defined above, updates its computation graph based on the HGWS. When the HGWS is provided as an input sample, the functional model generates a GWVR for each node in the input HGWS. The digital twin process for the new radio node is active.

[0134] Digital twin instance 110 is being based on Figure 4c The physical wireless network 120 is replicated, and the NI 170 waits for a digital twin service request 171 from the network application 180.

[0135] Figure 5 An exemplary message diagram 500 provided in the first embodiment is shown, illustrating message exchange with a digital twin instance in a gNB.

[0136] This first embodiment is based on the implementation of the entire digital twin instance 110, namely, including SMM / SAE / DIE functions 140, 220, and 210 in the RAN entity, such as the RAN data analysis function (e.g., RANDAF) of the access network (AN) (e.g., gNB 1) or the RAN of the 3GPP RAN. Here, another AN in the RAN (e.g., gNB 2) requests digital twin services from digital twin instance 110 to consume digital twin services for specific network applications (e.g., what-if performance analysis or AI-based network optimization). Other network applications include AI / ML model training (e.g., QoS prediction). The message exchange in this first embodiment is as follows: Figure 5 As shown.

[0137] Message Figure 500 illustrates the communication between a digital twin entity 110, including SMM 140, SAE 220, DIE 210 and AN1, and network nodes 121a and 121b of a physical wireless network 120 including UE1 to UEn, AN2 and AN3.

[0138] Following the initial registration process 501 between DIE 210, AN1, and the physical radio network 120 including UE1 to UEn, AN2, and AN3, message 502 is sent from DIE 210 to UE1 to transmit local neighborhood construction rules for each node type. UE1 performs configuration process 502a and begins measurement 503 using digital twin entity 110. UE1 then sends message 504 to SAE 220 to trigger a topology update of the local graph for each edge type. SAE 220 sends message 505 to SMM 140 to notify of the updated global graph. SMM 140 then updates the computation graph 505a.

[0139] After the state in UE1 has changed 505b (e.g., by handover execution), UE1 sends message 505c to SAE 220 to trigger a local neighborhood update for each edge type. SAE 220 sends message 505d to SMM 140 to notify of the updated global graph. SMM 140 then updates the computation graph 505e.

[0140] AN2 sends message 506 to SMM 140 to request digital twin services, such as what-if network scenario performance. SMM 140 replies with message 507 to transmit a generic vector representation and an optional output function. AN2 can then evaluate the performance 507a of the what-if scenario.

[0141] Steps (1) to (2) are included in the above regarding Figure 1 The general process described involves LNCR being transmitted via SI to the wireless nodes involved in the digital twin process, which is implemented through the Uu and Xn interfaces.

[0142] Step (3) includes the above regarding Figure 1 The general process described below. Implementation examples of LNCR for node-type UE and node-type BS are described below. Here, the radio interaction type is implemented as a side type.

[0143] Step (4) includes the above regarding Figure 1 In the general process described, the wireless node transmits changes from one or more HLWS to the SAE implemented within the gNB via the Uu and Xn interfaces.

[0144] Step (5) includes the above regarding Figure 1In the general process described, the SAE notifies the SMM, which is also implemented within gNB 1, of the updated HGWS. An example of an implementation of HGWS as a heterogeneous graph is shown below. Figure 7 As shown (for simplicity, additional node and edge feature information has been omitted). The SMM functionality is implemented, but not limited to, model classes of HGNN models. An example of the implementation of the heterogeneous computation graph of the HGNN model in the SMM functionality is shown below. Figure 8 As shown.

[0145] Step (6) includes the above regarding Figure 1 The general process described is as follows: gNB 2 requests a digital twin service, including a phenomenon ID, from the digital twin instance in gNB 1. In this embodiment, the phenomenon ID is an analysis ID. gNB 2, as a network application, runs AI-based trial-and-error optimization (e.g., RL-based or exhaustive search for resource allocation).

[0146] Step (7) includes the above regarding Figure 1 The general process described involves a response that maps a representation of a phenomenon ID to a phenomenon function based on the request. In a closed loop, the network application can request additional digital twin representations (e.g., updated optimized network actions) in what-if scenarios until convergence, as... Figure 5 As shown.

[0147] After gNB 2 uses digital twin instance 110 to find the optimal configuration, gNB 2 applications use the resource allocation configuration found by the digital twin to achieve even better performance.

[0148] The SI interface can be used for: (a) gNB-gNB pairs, (b) gNB-UE pairs, (c) RANDAF-gNB pairs, and (d) RANDAF-UE pairs. For gNB-gNB pairs, it is implemented via the Xn interface; for gNB-UE pairs, it is implemented via the Uu interface; for RANDAF-gNB pairs, it is implemented via a new interface defined to connect a new RAN entity (such as RANFAF) to the gNB; for UE-RANDAF pairs, it is implemented using the gNB as a relay via Uu and a new interface for RANDAF-gNB. The NI interface can be implemented via the Xn, N1, and / or N2 interfaces. Figure 6 The message flow of a 3GPP network is shown.

[0149] Figure 6 An exemplary 3GPP wireless network system 600 provided in the first embodiment is shown, wherein a complete digital twin instance is implemented in the RAN entity.

[0150] The 3GPP Radio Network System 600 includes: a control plane with the following entities: AF, AMF, OAM, and SMF; and a data plane with the following entities: AN1, UE1, AN2, UE2, AN3, UE3, UE4, UPF, and DN, as defined by 3GPP.

[0151] The digital twin entity 110 with SMM 140, SAE 220 and DIE 210 is implemented in 3GPP RAN entity 610 (AN1 entity here).

[0152] As mentioned above Figure 1 and Figure 2 The local neighborhood construction rule (LNCR) described for each node type is transmitted from AN1 to UE1 and AN2, from AN2 to UE2 and AN3, and from AN3 to UE3 and UE4 via the first path 601.

[0153] As mentioned above Figure 1 and Figure 2 The described heterogeneous local wireless structure (HWLS) update is transmitted from UE4 and UE3 to AN3, from AN3 and UE2 to AN2, and from AN2 and UE1 to AN1 via the second path 602.

[0154] As mentioned above Figure 1 and Figure 2 The general wireless vector representation (GWVR) described is transmitted from AN1 to AN2 via the third path 603.

[0155] Table 1 shows an implementation example of LNCR for node-type UEs. The radio interaction type is implemented as a side type. Rule entry 1 corresponds to side type 1, its description is associated with "Uu DL communication", its measurement trigger event corresponds to the HO execution event, and the action corresponds to step (4), as mentioned above regarding Figure 1 As described. Rule entry 2 corresponds to edge type 2, and its description is associated with "Uu DL inter-cell interference". If RSRP is greater than a certain threshold, the event of rule 2 is triggered.

[0156] Table 1: Examples of LNCR Implementation for Node-Type UEs

[0157] Table 2 shows an implementation example of LNCR for node type BS. The wireless interaction type is implemented as an edge type. Rule entry 1 corresponds to edge type 1, its description is associated with "Uu Uu communication", its measurement trigger event corresponds to the HO execution event, and the action corresponds to step (4), as mentioned above. Figure 1 As described. Rule entry 2 corresponds to edge type 2, and its description is associated with "Uu UL inter-cell interference". If the SRS is greater than a certain threshold, the event of rule 2 is triggered.

[0158] Table 2: Example of LNCR implementation for node type BS

[0159] Figure 7 An implementation example of a physical wireless network 120, represented as a heterogeneous structure 123, is shown.

[0160] Figure 7 The physical wireless network 120 shown on the left includes three base stations BS1, BS2, and BS3 as network nodes 121a and four user equipments UE1, UE2, UE3, and UE4 as network nodes 121b. The interaction 122 between base station 121a and UE 121b is represented by the connection between different network nodes 121a and 121b.

[0161] Physical wireless network 120 can be represented as a heterogeneous structure 123 with different types of nodes and edges (see [link]). Figure 7 (on the right side), namely, having a first type of network node 121a (BS node type) representing BS1, BS2, and BS3 and a second type of network node 121b (UE node type) representing UE1, UE2, UE3, and UE4, and interaction 122 between the two types of network nodes.

[0162] For each UE or each UE node type, a general wireless vector representation (GWVR) can be determined and associated with the corresponding UE.

[0163] therefore, Figure 7 An implementation example of a physical wireless network 120, represented as a heterogeneous graph or heterogeneous structure 123 (i.e., HGWS), is shown. The SMM can generate a GWVR for each wireless node (for simplicity, only the GWVR for the UE wireless node is shown).

[0164] Figure 8 An implementation example of the heterogeneous structure 123 of the digital twin model and its associated heterogeneous computation graph 123b is shown.

[0165] exist Figure 8On the left, wireless networks and wireless phenomena are shown as heterogeneous graphs, i.e., graphs with multiple types of nodes and edges.

[0166] In this example, three BS node type nodes communicate with one UE node type node. The heterogeneous structure 123 includes a communication side-type path 801 from BS1 to UE1 and two interference side-type paths 802 from BS2 to UE1 and from BS3 to UE1.

[0167] The SMM functionality is implemented, but not limited to, the HGNN model class. Figure 8 This is an example of the implementation of the heterogeneous computation graph of the HGNN model in the SMM function.

[0168] exist Figure 8 On the right, as an example, the associated heterogeneous computation diagram 123b of the digital twin model is shown.

[0169] The general radio vector representation 702 of UE1 is calculated by weighting each input of the corresponding base stations BS1, BS2, and BS3, comparing each input with a threshold, summing and normalizing each input.

[0170] Figure 9 An exemplary message diagram 900 provided by a sub-implementation of the first embodiment is shown, illustrating message exchange with the SMM / SAE at RANDAF in the RAN and the DIE at OAM in the core network.

[0171] This sub-implementation of the first embodiment 1 is based on a partial implementation of the digital twin instance 110 in the RAN, wherein SMM / SAE functions 140 and 220 are jointly located at the RAN data analytics function (RANDAF), and DIE function 210 is located at the OAM function in the CORE. In this sub-implementation, network applications (e.g., AI / ML model training) are run by NWDAF to train a QoS prediction model and deploy one or more models for inference. The goal of the network application is to predict the QoS, such as downlink throughput, that the UE will typically experience a few seconds in advance. The QoS prediction model is trained on synthetic data generated by the digital twin instance 110 (i.e., the GWVR generated by SMM 140 for each radio node). The digital twin instance 110 not only allows the generation of multiple synthetic data for corner scenarios, but its generated GWVR captures the relevant radio structure, thereby reducing the sample complexity of downstream network applications (such as QoS prediction). Another benefit is the privacy of RAN information relative to the CORE, and the 3GPP compliance of information exchange between the RAN and CORE.

[0172] Since DIE function 210 resides at the OAM function within the CORE, DIE 210 transmits the LCNR to the corresponding radio node via the SBA message bus, N1 / N2, Uu, and Xn interfaces of the 5G Core (5GC). HLWS changes are transmitted via the Uu and Xn interfaces for the newly defined interfaces of the RANDAF-UE and RANDAF-gNB. A Digital Twin Service Request 906 from the NWDAF is sent to the RAN entity via the common bus connecting all NFs in the CORE (e.g., the SBA message bus of the 5G Core (5GC)) and the N2 interface. A Digital Twin representation (i.e., the GWVR for each node) is sent to the NWDAF via the N2 interface and the 5GC's SBA message bus 907. The NWDAF can use the GWVR as input to train a QoS prediction model. Figure 10 The message flow of a 3GPP network is shown.

[0173] Message Figure 900 illustrates the communication between a digital twin entity 110, including SMM 140, SAE 220, and DIE 210, and network nodes 121a and 121b of a physical wireless network 120 including UE1 to UEn, AN2, AN3, and AN1.

[0174] Following the initial registration process 901 between DIE 210, AN1, and the physical wireless network 120 including UE1 to UEn, AN2, and AN3, message 902 is sent from DIE 210 to UE1 to transmit local neighborhood construction rules for each node type. UE1 performs configuration process 902a and begins measurement 903 using digital twin entity 110. UE1 then sends message 904 to SAE 220 to trigger a topology update of the local graph for each edge type. SAE 220 sends message 905 to SMM 140 to notify of the updated global graph. SMM 140 then updates the computation graph 905a.

[0175] After the state in UE1 has changed 905b (e.g., by handover execution), UE1 sends message 905c to SAE 220 to trigger a local neighborhood update for each edge type. SAE 220 sends message 905d to SMM 140 to notify of the updated global graph. SMM 140 then updates the computation graph 905e.

[0176] The NF / AF (e.g., NWDAF) sends message 906 to SMM 140 to request digital twin services, such as what-if network scenario performance. SMM 140 replies with message 907 to transmit a generic vector representation and an optional output function. The NF / AF (e.g., NWDAF) can then perform AL / ML model training (e.g., QoS prediction) 907a.

[0177] Figure 10 An exemplary 3GPP wireless network system 1000 provided in a sub-implementation of the first embodiment is shown, wherein a portion of the digital twin instance is implemented in the RAN entity, and the DIE is implemented at the OAM function in the core network.

[0178] The 3GPP wireless network system 1000 includes: a control plane 1001, having the following entities: AF, AMF, OAM and SMF; and a data plane, having the following entities: AN1, UE1, AN2, UE2, AN3, UE3, UE4, UPF and DN, as defined by 3GPP.

[0179] A portion of digital twin entity 110, having SMM 140 and SAE 220, is implemented in RANDAF 1010 of 3GPP RAN entity 610 (AN1 entity here). Another portion of digital twin entity 110, having DIE 210, is implemented in control plane 1001 or OAM 1020 of the core network.

[0180] As mentioned above Figure 1 and Figure 2 The described local neighborhood construction rule (LNCR) for each node type is transmitted from OAM 1020 to AMF via the first path 601, from AMF to AN1, AN2 and AN3, from AN1 to UE1, from AN2 to UE2 and from AN3 to UE3 and UE4.

[0181] As mentioned above Figure 1 and Figure 2 The described heterogeneous local wireless structure (HWLS) update is transmitted from UE4 and UE3 to AN3, from AN3 and UE2 to AN2, from AN2 and UE1 to AN1, and from AN1 to RANDAF 1010 via the second path 602.

[0182] As mentioned above Figure 1 and Figure 2 The general wireless vector representation (GWVR) described is transmitted from RANDAF 1010 to AN1, from AN1 to AMF, and from AMF to AF via a third path 603.

[0183] Figure 11An exemplary message diagram 1100 provided in the second embodiment is shown, illustrating message exchange between the SMM / SAE at the NWDAF in the core network and the DIE at the OAM in the core network.

[0184] This second embodiment is based on a partial implementation of digital twin instance 110, specifically SMM / SAE functions 140 and 220 at the NWDAF of the 3GPP core network and DIE 210 at the OAM function. In this implementation, another entity, referred to as the repository, such as the network repository function (NRF), collects registration information at the 5GC of the radio nodes involved in the digital twin process and discovers other NFs present in the 3GPP core network. The involved NFs / AFs perform standard procedures for capability and service openness according to TS.23.502.

[0185] In this embodiment, a network application (e.g., AI / ML model training) is run by another NF / AF (e.g., NWDAF 2) to train one or more QoS prediction models. The QoS prediction models are trained on synthetic data generated by the digital twin instance 110 (i.e., GWVR generated by SMM 140 for each wireless node).

[0186] Since all functions now associated with digital twin instances reside in the 3GPP core network, these functions discover services and information from other NF / AFs through a repository (i.e., NRF).

[0187] Step (1) The wireless nodes involved in the digital twin process register with the NRF.

[0188] Step (1.1) DIE requests the repository to perform wireless node discovery.

[0189] Step (1.2) The repository responds with a wireless node discovery response, which includes information related to the wireless node ID, node type, available local measurements for each node, and communication capabilities of the wireless node involved in the digital twin process.

[0190] Other steps are included in the above regarding Figure 1 and Figure 2 The general process is described below. Since DIE function 210 is located at the OAM function within the CORE, DIE 210 transmits LCNR to the corresponding radio node via the SBA message bus of the 5G Core (5GC), N1 / N2, Uu, and Xn interfaces. HLWS changes are transmitted to NWDAF 1 via the Uu and Xn interfaces, N1, and the SBA message bus of the 5G Core (5GC).

[0191] Digital twin service requests from NWDAF 2 are sent to NWDAF 2 via a common bus connecting all NFs in the CORE (e.g., the SBA message bus of the 5GCore (5GC)). The digital twin representation (i.e., the GWVR for each node) is sent to the other NFs / AFs, namely NWDAF 2, via the 5GC's SBA message bus. NWDAF 2 can then use the GWVR as input to train a QoS prediction model. Figure 12 The message flow of a 3GPP network is shown.

[0192] Message Figure 1100 illustrates communication between a digital twin entity 110 (e.g., implemented as NWDAF) including SMM 140, SAE 220, DIE 210 and a repository (e.g., NRF) and network nodes 121a, 121b of a physical wireless network 120 including UE1 to UEn, AN2, AN3 and AN1 and NF / AF (e.g., NWDAF2).

[0193] After the initial registration process 1101 between the repository, AN1, and the physical radio networks 120 including UE1 to UEn, AN2, and AN3, message 1101a is sent from the DIE to the repository for radio node discovery. The repository responds with message radio node response 1101b. Then, message 1102 is sent from the DIE 210 to UE1 to transmit local neighborhood construction rules for each node type. UE1 configures process 1102a and begins measurement 1103. Then, UE1 sends message 1104 to the SAE 220 to trigger a topology update of the local graph for each edge type. The SAE 220 sends message 1105 to the SMM 140 to notify of the updated global graph. The SMM 140 then updates the computation graph 1105a.

[0194] After the state in UE1 has changed 1105b (e.g., by handover execution), UE1 sends message 1105c to SAE 220 to trigger a local neighborhood update for each edge type. SAE 220 sends message 1105d to SMM 140 to notify of the updated global graph. SMM 140 then updates the computation graph 1105e.

[0195] The NF / AF (e.g., NWDAF2) sends message 1106 to SMM 140 to request digital twin services, such as what-if network scenario performance. SMM 140 replies with message 1107 to transmit a generic vector representation and an optional output function. The NF / AF (e.g., NWDAF2) can then perform AL / ML model training (e.g., QoS prediction) 1107a.

[0196] Figure 12An exemplary 3GPP wireless network system 1200 provided in the second embodiment is shown, wherein a portion of the digital twin instance is implemented in the NWDAF, and the DIE is implemented at the OAM function in the core network.

[0197] The 3GPP wireless network system 1200 includes: a control plane 1001 with the following entities: AF, AMF, OAM, SMF, NRF and NWDAF; and a data plane with the following entities: AN1, UE1, AN2, UE2, AN3, UE3, UE4, UPF and DN, as defined by 3GPP.

[0198] A portion of the digital twin entity 110, having SMM 140 and SAE 220, is implemented in NWDAF 1030 in the core network. Another portion of the digital twin entity 110, having DIE 210, is implemented in control plane 1001 or OAM 1020 in the core network.

[0199] As mentioned above Figure 1 and Figure 2 The described local neighborhood construction rule (LNCR) for each node type is transmitted from OAM 1020 to AMF via the first path 601, from AMF to AN1, AN2 and AN3, from AN1 to UE1, from AN2 to UE2 and from AN3 to UE3 and UE4.

[0200] As mentioned above Figure 1 and Figure 2 The described heterogeneous local wireless structure (HWLS) update is transmitted via second path 602 from UE4 and UE3 to AN3, from UE2 to AN2, from UE1 to AN1, from AN1, AN2 and AN3 to AMF and from AMF to NWDAF 1030.

[0201] As mentioned above Figure 1 and Figure 2 The general wireless vector representation (GWVR) described is transmitted from the NWDAF 1030 to the AF via a third path 603.

[0202] Figure 13 An exemplary message diagram 1300 provided in the third embodiment is shown, illustrating message exchange with the SMM / SAE / DIE at mobile edge computing (MEC).

[0203] This embodiment is based on a complete implementation of digital twin instance 110 (i.e., SMM / SAE / DIE functions 140, 220, 210) on a MEC server connected to the 3GPP system. In this implementation, another entity, called a repository, such as a network repository function (NRF), collects registration information at the 5GC of the radio nodes involved in the digital twin process, and discovers other NFs present in the 3GPP network. The involved NFs / AFs perform standard procedures for capability and service openness according to TS. 23.502.

[0204] MEC applications may or may not use the network exposure function (NEF) to access services from untrusted entities outside the domain.

[0205] In this embodiment, network applications (e.g., AI / ML model training) are run by another NF / AF (e.g., NWDAF 1) to train one or more QoS prediction models. The QoS prediction models are trained on synthetic data generated by digital twin instance 110 (i.e., GWVR generated by SMM for each radio node). For the current implementation, the MEC server logically resides in the data network (DN).

[0206] When SMM / SAE / DIE functionality is implemented in the MEC server, the DIE 210 transmits the LCNR to the corresponding radio nodes via the N6, N3, Xn, and Uu interfaces. HLWS changes communicate with the MEC server residing in the DN via the Uu and Xn interfaces, and the N3 and N6 interfaces. Digital twin service requests from NWDAF 1 are sent to the MEC server via a common bus connecting all NFs in the CORE (e.g., the SBA message bus of the 5G Core (5GC), and the N4 and N6 interfaces). The digital twin representation (i.e., the GWVR for each node) is sent to other NFs / AFs, namely NWDAF 1, via the N4 / N6 interfaces and the 5GC's SBA message bus. NWDAF1 can use the GWVR as input to train AI / ML models (e.g., QoS prediction models).

[0207] Message Figure 1300 illustrates communication between a digital twin entity 110 (e.g., implemented as MEC) including SMM 140, SAE 220, and DIE 210 and a repository (e.g., NRF) and network nodes 121a, 121b of a physical wireless network 120 including UE1 to UEn, AN2, AN3, and AN1 and NF / AF (e.g., NWDAF1).

[0208] After initial process registration 1301 between the repository, AN1, and the physical radio networks 120 including UE1 to UEn, AN2, and AN3, message 1301a is sent from the DIE to the repository for radio node discovery. The repository responds with message radio node response 1301b. Then, message 1302 is sent from the DIE 210 to UE1 to transmit local neighborhood construction rules for each node type. UE1 configures process 1302a and begins measurement 1303. Then, UE1 sends message 1304 to the SAE 220 to trigger a topology update of the local graph for each edge type. The SAE 220 sends message 1305 to the SMM 140 to notify of the updated global graph. The SMM 140 then updates the computation graph 1305a.

[0209] After the state in UE1 has changed 1305b (e.g., by handover execution), UE1 sends message 1305c to SAE 220 to trigger a local neighborhood update for each edge type. SAE 220 sends message 1305d to SMM 140 to notify of the updated global graph. SMM 140 then updates the computation graph 1305e.

[0210] The NF / AF (e.g., NWDAF1) sends message 1306 to SMM 140 to request digital twin services, such as what-if network scenario performance. SMM 140 replies with message 1307 to transmit a generic vector representation and an optional output function. The NF / AF (e.g., NWDAF1) can then perform AL / ML model training (e.g., QoS prediction) 1307a.

[0211] Figure 14 An exemplary 3GPP wireless network system 1400 provided in a third embodiment is shown, wherein a digital twin instance is implemented at the MEC.

[0212] The 3GPP Radio Network System 1400 includes: a control plane 1001 with the following entities: AF, AMF, OAM, SMF, NEF, NRF and NWDAF; and a data plane with the following entities: AN1, UE1, AN2, UE2, AN3, UE3, UE4, UPF and DN and MEC 1410, as defined by 3GPP.

[0213] Implement a digital twin entity 110 with SMM 140, SAE 220 and DIE 210 in MEC server 1410.

[0214] As mentioned above Figure 1 and Figure 2The local neighborhood construction rule (LNCR) described for each node type is transmitted from MEC server 1410 to UPF via the first path 601, from UPF to AN3, AN2 and AN1, from AN3 to UE3 and UE4, from AN2 to UE2 and from AN1 to UE1.

[0215] As mentioned above Figure 1 and Figure 2 The described heterogeneous local wireless structure (HWLS) update is transmitted via a second path 602 from UE4 and UE3 to AN3, from UE2 to AN2, from UE1 to AN1, from AN1, AN2 and AN3 to UPF, and from UPF to MEC server 1410.

[0216] As mentioned above Figure 1 and Figure 2 The general wireless vector representation (GWVR) described is transmitted from MEC server 1410 to AF via UPF and SMF through a third path 603.

[0217] In this invention, a logical DIE function or entity 210 is introduced. This enables the definition of a configurable set of instructions to collect local structural information from the wireless node as HLWS.

[0218] The novel LNCR set introduced in this invention supports monitoring relevant structural changes in the physical wireless network for each type of wireless interaction and notifies these HLWS changes to enable the creation / adjustment of the digital twin model residing in the digital twin instance. Compared to unstructured models, model adaptation allows for more efficient and accurate computation to infer network behavior.

[0219] The logical SAE function or entity 220, along with local and global structures, is introduced to enable the HGNN model residing in the SMM function or entity 140 to adjust the changes in HLWS reported by wireless nodes in the aggregated wireless network to construct the HGWS.

[0220] Introducing messages to adapt digital twin models to changes in wireless networks and provide digital representations (i.e., GWVR) enables digital twin instances to generate GWVRs based on each node defined in the HGWS as input samples and transmit the generated GWVRs to any network application.

[0221] While certain features or aspects of the invention may have been disclosed in combination with only one of several implementations, such features or aspects may be combined with one or more other features or aspects of other implementations, provided that they are necessary or advantageous for any given or particular application. Furthermore, to a certain extent, the terms “comprising,” “having,” “possessing,” or other variations of these words are used in the detailed description or claims; such terms, like the term “comprising,” are similar and both indicate inclusion. Similarly, the terms “exemplary” and “for example” are used only as examples and not as best or most preferred. The terms “coupled” and “connected,” as well as their derivatives, may be used. It should be understood that these terms may be used to indicate that two elements cooperate or interact with each other, whether they are in direct physical or electrical contact, or whether they are not in direct contact with each other.

[0222] While specific aspects have been illustrated and described herein, those skilled in the art will understand that various alternatives and / or equivalent implementations may replace the specific aspects shown and described without departing from the scope of the invention. This application is intended to cover any adaptations or variations of the specific aspects discussed herein.

[0223] Although the elements in the following claims are listed in a particular order using corresponding labels, these elements are not necessarily limited to being implemented in that particular order unless the description of the claims otherwise implies a particular order for implementing some or all of these elements.

[0224] Based on the above guidance, many alternatives, modifications, and variations will be apparent to those skilled in the art. Of course, those skilled in the art will readily recognize that the invention has many applications beyond those described herein. Although the invention has been described with reference to one or more specific embodiments, those skilled in the art will recognize that many changes can be made to the invention without departing from its scope. Therefore, it should be understood that the invention can be implemented in ways different from those specifically described herein, as long as it remains within the scope of the appended claims and their equivalents.

Claims

1. A digital twin entity (110) for replicating a physical wireless network (120), characterized in that, The physical wireless network (120) can be represented as multiple network nodes (121a, 121b) in a heterogeneous structure (123) with different types of nodes and edges, and the interaction (122) between the network nodes (121a, 121b), wherein the digital twin entity (110) includes: An interface (130) to the physical wireless network (120) is used to receive information about the interaction (122) between the network nodes (121a, 121b) of the physical wireless network (120), the information being represented as a plurality of heterogeneous structures (123) derived from measurements of the network nodes (121a, 121b), each heterogeneous structure (123) representing the structure of the physical wireless network (120) obtained by the corresponding network node (121a, 121b); Service Mapping Model (SMM) subsystem (140) for providing a copy of the physical wireless network (120) based on an artificial intelligence and / or machine learning AI / ML model (141), the AI / ML model (141) being trained using the heterogeneous structure (123) derived from the measurements of the network nodes (121a, 121b); The AI / ML model (141) is used to generate a digital representation (702) for each network node (121a, 121b) of the physical wireless network (120).

2. The digital twin entity (110) according to claim 1, characterized in that, The AI / ML model (141) includes a heterogeneous graph neural network (HGNN).

3. The digital twin entity (110) according to claim 1 or 2, characterized in that, The AI / ML model (141) is used to generate a general wireless vector representation (702) for each network node (121a, 121b) of the physical wireless network (120).

4. The digital twin entity (110) according to any one of the preceding claims, characterized in that, include: A digital twin entity management subsystem (150) is used to monitor the performance and resource consumption of the AI / ML model (141).

5. The digital twin entity (110) according to any one of the preceding claims, characterized in that, include: A data repository (160) is used to collect and store information about the interactions (122) between the network nodes (121a, 121b) of the physical wireless network (120) and the network nodes (121a, 121b) by collecting and updating real-time data of the network nodes (121a, 121b) through the interface (130) of the physical wireless network (120).

6. The digital twin entity (110) according to any one of the preceding claims, characterized in that, include: An interface (170) with a network application system (180) is used to receive digital twin service requests (171) from multiple network applications (181, 182) and to provide services (172) of the digital twin entity (110) to the multiple network applications (181, 182).

7. The digital twin entity (110) according to claim 6, characterized in that, The digital twin service request (171) includes at least one of the following: A request for at least one network node (121a, 121b) that identifies the physical wireless network (120), wherein the digital representation (702) of the at least one network node (121a, 121b) is requested by the network application; A request for the timestamp of the digital representation (702); Optional information includes the phenomenon ID and at least one of one or more what-if network configuration IDs; The service (172) of the digital twin entity (110) includes at least one of the following: The digital representation (702) of the at least one network node (121a, 121b) of the physical wireless network (120) is specified in the digital twin service request (171); The number represents the timestamp of (702); Optional information includes at least one of the following: a mapping from the phenomenon ID to the phenomenon function; optimal network configuration.

8. The digital twin entity (110) according to any one of the preceding claims, characterized in that, include: Data instruction entity DIE (210) is used to define a set of configurable rules (211) for each node type of the network nodes (121a, 121b). The DIE (210) is used to register the network nodes (121a, 121b) of the physical wireless network (120), and after registration, it transmits the configurable rule set (211) to the network nodes (121a, 121b).

9. The digital twin entity (110) according to claim 8, characterized in that, The configurable rule set (211) is a Local Neighborhood Construction Rule (LNCR), which can detect the interaction changes between the network nodes (121a, 121b) based on local measurements or local network events of the network nodes (121a, 121b).

10. The digital twin entity (110) according to claim 8 or 9, characterized in that, The DIE (210) is used to receive local information from the network nodes (121a, 121b), the local information including at least one of the following: The identifier of the corresponding network node; The node type of the corresponding network node; Available measurements for the corresponding network nodes; The communication function of the corresponding network node.

11. The digital twin entity (110) according to any one of claims 8 to 10, characterized in that, include: The Structural Convergence Entity (SAE) (220) is used to construct a heterogeneous global wireless data structure (HGWS) representing the structure of the physical wireless network (120) at a given time, based on the notification of local structural changes of the physical wireless network (121a, 121b) reported by the network nodes (121a, 121b). The SAE (220) is used to provide the HGWS to the SMM subsystem (140) so that the AI / ML model (141) can adapt to the local structural changes of the physical wireless network (120).

12. The digital twin entity (110) according to claim 11, characterized in that, The notification from the network nodes (121a, 121b) regarding the change in the local structure of the physical wireless network (120) includes at least one of the following: Wireless interaction type ID, Source wireless node, Target wireless node, An optional trigger signal can be used to trigger the local structural change report of the network nodes (121a, 121b).

13. The digital twin entity (110) according to any one of the preceding claims, characterized in that, The digital twin entity (110) is implemented in the 3GPP Radio Access Network (RAN) entity (610) of the physical wireless network (120).

14. The digital twin entity (110) according to any one of claims 8 to 12, characterized in that, The SMM subsystem (140) and the SAE (220) entities are used to implement the data analysis function component (1010) of the 3GPP radio access network RAN ​​entity of the physical wireless network (120); The DIE entity (210) is used to implement the operation and maintenance OAM entity (1020) in the 3GPP core network (1001).

15. The digital twin entity (110) according to any one of claims 8 to 12, characterized in that, The SMM subsystem (140) and the SAE (220) entities are used to implement the data analysis function component (1030) of the 3GPP core network (1001); The DIE entity (210) is used to implement the operation and maintenance OAM entity (1020) of the 3GPP core network (1001).

16. The digital twin entity (110) according to any one of claims 8 to 12, characterized in that, The digital twin entity (110) is used to implement a mobile edge computing MEC server (1410) connected to the physical wireless network (120).

Images