Data acquisition method and apparatus, device, storage medium, and computer program product

Abstract

The present application discloses a data acquisition method and apparatus, a device, a storage medium, and a computer program product. The method comprises: an access network device receiving first information sent by a first function, the first information being used for indicating acquisition of data related to a first task, the first task being related to a digital twin, and the first function being at least used for processing user plane data; sending second information to one or more terminals, the second information being used for indicating acquisition of the data related to the first task; receiving first data reported by the one or more terminals, the first data comprising one or more first vectors, and the first vector including a vector obtained by the terminal performing feature representation on the acquired data; performing fusion representation on received one piece or more pieces of the first data and second data to obtain a second vector, the second data including the data locally stored in the access network device and related to the first task; and sending third data to the first function, the third data comprising the second vector.

Description

Data acquisition methods, devices, equipment, storage media, and computer program products

[0001] Cross-reference to related applications

[0002] This application claims priority and benefits to patent application No. 202510198962.0, filed with the China National Intellectual Property Administration on February 21, 2025, the entire contents of which are incorporated herein by reference. Technical Field

[0003] This application relates to the field of communications, and more particularly to a data acquisition method, apparatus, device, storage medium, and computer program product. Background Technology

[0004] In related technologies, data storage centers deployed in physical networks can centrally collect data related to digital twin tasks when pre-configured instructions are triggered, process the collected datasets, and send the processed data to the twin network for the construction of digital twins.

[0005] However, using solutions in related technologies to process data for digital twin tasks results in significant consumption of communication resources. Summary of the Invention

[0006] To address the related technical issues, embodiments of this application provide a data acquisition method, apparatus, device, storage medium, and computer program product.

[0007] The technical solution of this application embodiment is implemented as follows:

[0008] This application provides a data acquisition method applied to an access network device, the method comprising:

[0009] Receive first information sent by a first function, the first information being used to instruct the collection of data related to a first task, the first task being related to a digital twin, and the first function being used to process user plane data at least;

[0010] Send a second message to one or more terminals, the second message being used to instruct the collection of data related to the first task;

[0011] Receive first data reported by one or more terminals, the first data containing one or more first vectors, the first vectors including vectors obtained by the terminals performing feature representation on the collected data;

[0012] The received one or more of the first data and the second data are fused and represented to obtain a second vector, wherein the second data includes local data of the access network device related to the first task;

[0013] Send third data to the first function, the third data containing the second vector.

[0014] This application embodiment also provides a data acquisition method applied to a terminal, the method comprising:

[0015] Receive second information sent by the access network device, the second information being used to instruct the collection of data related to the first task, the first task being related to digital twin;

[0016] Collect data related to the first task;

[0017] The collected data is characterized to obtain one or more first vectors;

[0018] Send first data to the access network device, the first data including the one or more first vectors.

[0019] This application embodiment also provides a data acquisition method applied to a first function, the first function being used to process user plane data, the method comprising:

[0020] The system receives a fifth message sent by the second function, the fifth message being used to instruct the collection of data related to the first task, the first task being related to digital twins, and the second function being used at least for network data analysis.

[0021] Send a first message to one or more access network devices, the first message being used to instruct the collection of data related to the first task;

[0022] Receive third data sent by one or more access network devices, the third data including a second vector, the second vector including a vector obtained by the access network device by fusing and representing the data related to the first task;

[0023] Relation extraction and joint feature learning are performed on one or more third data and fourth data received to obtain a first tensor, wherein the fourth data includes data related to the first function and the first task locally;

[0024] Send sixth data to the second function, the sixth data containing the first tensor.

[0025] This application embodiment also provides a data acquisition method applied to a second function, the second function being at least used for network data analysis, the method comprising:

[0026] Send a fifth message to one or more first functions, the fifth message being used to instruct the collection of data related to a first task, the first task being related to a digital twin, and the first function being used to process user plane data at least;

[0027] Receive sixth data reported by one or more first functions, the sixth data containing a first tensor, the first tensor containing a tensor obtained by the first function from relation extraction and joint feature learning of the data related to the first task;

[0028] Using all of the first tensors, construct the digital twin corresponding to the first task.

[0029] This application embodiment also provides a data acquisition device, installed in an access network device, including:

[0030] The first receiving unit is configured to receive first information sent by a first function, the first information being used to indicate the collection of data related to a first task, the first task being related to a digital twin, and the first function being used to process user plane data at least; and to receive first data reported by one or more terminals, the first data containing one or more first vectors, the first vectors including vectors obtained by the terminals performing feature representation on the collected data.

[0031] The first processing unit is configured to fuse and characterize one or more of the received first data and second data to obtain a second vector, wherein the second data includes local data of the access network device related to the first task.

[0032] A first sending unit is configured to send second information to one or more terminals, the second information being used to instruct the collection of data related to the first task; and to send third data to the first function, the third data including the second vector.

[0033] This application embodiment also provides a data acquisition device, including:

[0034] The second receiving unit is used to receive second information sent by the access network device. The second information is used to indicate the collection of data related to the first task, which is related to the digital twin.

[0035] The second processing unit is used to collect data related to the first task; and to perform feature characterization on the collected data to obtain one or more first vectors.

[0036] The second sending unit is configured to send first data to the access network device, the first data including the one or more first vectors.

[0037] This application embodiment also provides a data acquisition device, configured with a first function, the first function being used at least to process user plane data, including:

[0038] The third receiving unit is configured to receive fifth information sent by the second function, the fifth information being used to indicate the collection of data related to the first task, the first task being related to a digital twin, and the second function being used at least for network data analysis; and to receive third data sent by one or more access network devices, the third data containing a second vector, the second vector including a vector obtained by the access network device through fusion representation of the data associated with the first task;

[0039] The third processing unit is used to perform relation extraction and joint feature learning on one or more received third data and fourth data to obtain a first tensor, wherein the fourth data includes data related to the first function and the first task.

[0040] The third sending unit is configured to send first information to one or more access network devices, the first information being used to instruct the collection of data related to the first task; and to send sixth data to the second function, the sixth data containing the first tensor.

[0041] This application embodiment also provides a data acquisition device, including:

[0042] The fourth sending unit is used to send fifth information to one or more first functions, the fifth information being used to indicate the collection of data related to a first task, the first task being related to a digital twin, and the first function being used to process user plane data at least.

[0043] The fourth receiving unit is configured to receive the sixth data reported by the one or more first functions, the sixth data including a first tensor, the first tensor including a tensor obtained by the first function performing relation extraction and joint feature learning on the data related to the first task.

[0044] The fourth processing unit is used to construct a digital twin corresponding to the first task using all of the first tensors.

[0045] This application embodiment also provides an access network device, including: a first processor and a first communication interface; wherein,

[0046] The first communication interface is configured to receive first information sent by a first function, the first information indicating the collection of data related to a first task, the first task being related to a digital twin, and the first function being at least used to process user plane data; send second information to one or more terminals, the second information indicating the collection of data related to the first task; receive first data reported by the one or more terminals, the first data containing one or more first vectors, the first vectors including vectors obtained by the terminals performing feature representation on the collected data; and send third data to the first function, the third data containing the second vectors.

[0047] The first processor is configured to fuse and characterize one or more of the received first data and second data to obtain a second vector, wherein the second data includes local data of the access network device related to the first task.

[0048] This application also provides a terminal, including: a second processor and a second communication interface; wherein,

[0049] The second communication interface is used to receive second information sent by the access network device, the second information being used to instruct the collection of data related to a first task, the first task being related to a digital twin; and to send first data to the access network device, the first data containing one or more first vectors;

[0050] The second processor is used to collect data related to the first task; and to perform feature characterization on the collected data to obtain one or more first vectors.

[0051] This application embodiment also provides a first function, which is at least used for processing user plane data, including: a third processor and a third communication interface; wherein...

[0052] The third processor is configured to receive fifth information sent by the second function, the fifth information indicating the collection of data related to a first task, the first task being related to a digital twin, and the second function being at least used for network data analysis; send first information to one or more access network devices, the first information indicating the collection of data related to the first task; receive third data sent by the one or more access network devices, the third data containing a second vector, the second vector including a vector obtained by the access network devices through fusion representation of the data related to the first task; and send sixth data to the second function, the sixth data containing a first tensor.

[0053] The third processor is used to perform relation extraction and joint feature learning on one or more received third data and fourth data to obtain a first tensor, wherein the fourth data includes data related to the first function locally and the first task.

[0054] This application embodiment also provides a second function, which is at least used for network data analysis, including: a fourth processor and a fourth communication interface; wherein...

[0055] The fourth communication interface is used to send fifth information to one or more first functions, the fifth information being used to instruct the collection of data related to a first task, the first task being related to a digital twin, and the first function being used to process user plane data at least; and to receive sixth data reported by the one or more first functions, the sixth data containing a first tensor, the first tensor containing a tensor obtained by the first function from relation extraction and joint feature learning of the data related to the first task.

[0056] The fourth processor is used to construct a digital twin corresponding to the first task using all the first tensors.

[0057] This application also provides an access network device, including: a first processor and a first memory for storing a computer program capable of running on the processor.

[0058] Wherein, when the first processor is used to run the computer program, it executes the steps of any of the methods described above on the access network device side.

[0059] This application also provides a terminal, including: a second processor and a second memory for storing computer programs capable of running on the processor.

[0060] Wherein, when the second processor is running the computer program, it executes the steps of any of the methods described above on the terminal side.

[0061] This application embodiment also provides a first function, including: a third processor and a third memory for storing a computer program capable of running on the processor.

[0062] When the third processor runs the computer program, it executes any of the steps of the first functional side method described above.

[0063] This application also provides a second function, including: a fourth processor and a fourth memory for storing computer programs capable of running on the processor.

[0064] The fourth processor is used to execute any of the steps of the second functional side method when running the computer program.

[0065] This application embodiment also provides a storage medium storing a computer program thereon. When the computer program is executed by a processor, it implements the steps of any of the above-described access network device-side methods, or the steps of any of the above-described terminal-side methods, or the steps of any of the above-described first functional-side methods, or the steps of any of the above-described second functional-side methods.

[0066] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of any of the above-described access network device-side methods, or the steps of any of the above-described terminal-side methods, or the steps of any of the above-described first functional-side methods, or the steps of any of the above-described second functional-side methods.

[0067] The data acquisition method, apparatus, device, storage medium, and computer program product provided in this application embodiment include: a second function sending fifth information to one or more first functions, the fifth information indicating the acquisition of data related to a first task, the first task being related to digital twins; the first function being used to process user plane data at least, and the second function being used to perform network data analysis at least; the first function receiving the fifth information sent by the second function; the first function sending first information to one or more access network devices, the first information indicating the acquisition of data related to the first task; the access network devices receiving the first information sent by the first function; the access network devices sending second information to one or more terminals, the second information indicating the acquisition of data related to the first task; the terminals receiving the second information sent by the access network devices; the terminals acquiring data related to the first task; the terminals performing feature characterization on the acquired data to obtain one or more first vectors; and the terminals sending first data to the access network devices. The first data includes one or more first vectors; the access network device receives the first data reported by one or more terminals; the access network device performs fusion representation on the received one or more first data and second data to obtain a second vector, the second data including local data related to the first task of the access network device; the access network device sends third data to the first function, the third data including the second vector; the first function receives the third data sent by one or more access network devices; the first function performs relation extraction and joint feature learning on the received one or more third data and fourth data to obtain a first tensor, the fourth data including local data related to the first task of the first function; the first function sends sixth data to the second function, the sixth data including the first tensor; the second function receives the sixth data reported by one or more first functions; the second function uses all the first tensors to construct a digital twin corresponding to the first task.

[0068] The solution provided in this application embodiment involves the following steps: When a second function needs to construct a digital twin corresponding to a first task, the second function sends a data collection request to the first function associated with the first task (e.g., a User Plane Function (UPF)). The first function then sends a data collection request to the access network device associated with the first task, and the access network device sends a data collection request to the terminal associated with the first task. The terminal then collects data related to the first task, performs feature representation on the collected data to obtain a feature vector for the terminal, and reports the feature vector to the access network device. The access network device performs fusion representation on the received feature vector of the terminal and the local data related to the first task to obtain a feature vector for the access network device, and reports the feature vector to the first function. The first function performs relation extraction and joint feature learning on the received feature vector of the access network device and the local data related to the first task to obtain a tensor for constructing the digital twin, and reports the tensor to the second function. The second function then uses the received tensor to construct the digital twin corresponding to the first task. In this process, the terminal, access network equipment, and the first functional side all process the collected data as needed (which can also be understood as on-network data representation or on-network data processing), and report the processed data layer by layer (the layers can include the terminals, access network equipment, the first function, and the second function, respectively) (which can also be understood as multi-level data transmission). Thus, the processed data (such as tensors or feature vectors) is smaller in size compared to the unprocessed data. Therefore, transmitting the processed data consumes fewer communication resources (such as bandwidth) compared to transmitting unprocessed data (which can also be understood as full metadata transmission). The overhead is lower; at the same time, compared with the scheme of centralized collection and centralized processing of digital twin-related data, when the network element nodes (such as terminals, access network equipment, and first functions) in the transmission path perform multi-level hierarchical representation of the collected data (i.e., multi-level on-network data representation is performed separately), the computing power requirement is distributed on multiple levels of devices, which is easier to meet, the processing latency is lower, and there is no need to deploy an additional storage center for centralized collection and processing of data, resulting in lower costs; in addition, the scheme provided in this application embodiment does not need to rely on a pre-built instruction library for data collection, and can flexibly collect data according to needs. Attached Figure Description

[0069] Figure 1 is a flowchart illustrating a data acquisition method in related technologies;

[0070] Figure 2 is a flowchart illustrating the first data acquisition method according to an embodiment of this application;

[0071] Figure 3 is a flowchart illustrating the second data acquisition method according to an embodiment of this application;

[0072] Figure 4 is a flowchart illustrating the third data acquisition method according to an embodiment of this application;

[0073] Figure 5 is a flowchart illustrating the fourth data acquisition method according to an embodiment of this application;

[0074] Figure 6 is a schematic diagram of the multi-level data transmission architecture in the application example of this application;

[0075] Figure 7 is a flowchart illustrating an application example of this application: a multi-level efficient data acquisition method for digital twins.

[0076] Figure 8 is a flowchart illustrating a method for configuring user equipment (UE) data collection and reporting via RRC (Radio Resource Control) signaling according to an embodiment of this application.

[0077] Figure 9 is a schematic diagram of the structure of a first type of data acquisition device according to an embodiment of this application;

[0078] Figure 10 is a schematic diagram of the structure of a second type of data acquisition device according to an embodiment of this application;

[0079] Figure 11 is a schematic diagram of the structure of a third type of data acquisition device according to an embodiment of this application;

[0080] Figure 12 is a schematic diagram of the structure of the fourth data acquisition device according to an embodiment of this application;

[0081] Figure 13 is a schematic diagram of the access network device structure according to an embodiment of this application;

[0082] Figure 14 is a schematic diagram of the terminal structure according to an embodiment of this application;

[0083] Figure 15 is a schematic diagram of the first functional structure of an embodiment of this application;

[0084] Figure 16 is a schematic diagram of the second functional structure of an embodiment of this application;

[0085] Figure 17 is a schematic diagram of the data acquisition system structure according to an embodiment of this application. Detailed Implementation

[0086] The present application will now be described in further detail with reference to the accompanying drawings and embodiments.

[0087] In the network architecture provided by related technologies, Network Data Analytics Function (NWDAF) is typically used for intelligent decision-making and data analysis (which can also be understood as intelligent decision-making and analysis). However, NWDAF may introduce illusions (such as incorrect inferences or predictions) and uncertainties during the intelligent decision-making and analysis process, thereby affecting the accuracy and reliability of the decision-making and analysis results.

[0088] In related technologies, constructing a digital twin in NWDAF (which can also be understood as digital twin modeling, or a digital twin network model or digital twin model) enhances the accuracy and integrity of network data, thereby ensuring the accuracy and reliability of NWDAF's intelligent decision-making and analysis results. Specifically, the steps for constructing a digital twin network model typically include: collecting data from various network entities and / or functions, processing the collected data to obtain processed data, and using the processed data for parametric modeling to obtain a digital model of the entire network in virtual space. The accuracy of digital twin modeling is closely related to the real-time nature and diversity (diversity can also be understood as richness, the quantity and quality of data) of the collected data; in other words, data is the foundation for constructing the digital twin.

[0089] However, the currently defined data service types in NWDAF are limited, typically only supporting application scenarios and use cases such as network optimization and customer experience improvement. Therefore, NWDAF can usually only perceive relevant data from the core network side (i.e., only achieve single-domain data intelligent perception), lacking the ability to perceive relevant data from the terminal's domain, the access network device's domain, etc. (which can also be understood as multimodal data from the terminal side and the access network device side). In other words, the types of data that NWDAF can collect are limited. Currently, there is no effective solution to ensure that NWDAF can collect diverse and rich data (i.e., ensure the diversity and richness of the collected data) and to build a more accurate digital twin in real time based on the collected data to better meet business needs.

[0090] Meanwhile, in practical applications, the amount of data related to the construction of the digital twin on the terminal side and the access network equipment side may be large. When NWDAF collects data for the construction of the digital twin, if all data (which can also be understood as full metadata) on the terminal side and the access network equipment side is transmitted, there may be a problem of excessive bandwidth overhead.

[0091] As can be seen from the above description, there is an urgent need for a data acquisition scheme that can ensure the diversity of collected data while minimizing the bandwidth overhead of transmitting the collected data, so that NWDAF can efficiently collect relevant multimodal data from multiple domains (specifically, the domains corresponding to the core network, access network equipment, and terminals) when collecting data for building digital twins.

[0092] In related technologies, in order to achieve efficient transmission of digital twin-related data (which can also be understood as digital twin data, or data for constructing a digital twin) in the physical network, i.e., to reduce the overhead of transmitting digital twin-related data, as shown in Figure 1, an efficient data acquisition method for digital twin networks is proposed, which specifically includes the following steps:

[0093] Step 101: The data storage center in the physical network sends registration information to the command management center in the twin network to register; the registration information may include the address of the data storage information, the data type of the stored data, the size of the stored data, etc.

[0094] Step 102: The data storage center of the twin network sends a data acquisition request to the command management center;

[0095] Step 103: The command management center of the twin network configures data acquisition commands based on the data acquisition request and addresses the data storage center in the physical network related to the data acquisition request;

[0096] Step 104: The command management center of the twin network sends the configured data acquisition command to the data storage center in the physical network related to the data acquisition request;

[0097] Step 105: After receiving the data acquisition instruction, the data storage center in the physical network parses the data acquisition instruction and executes data acquisition according to the instruction. Then, the data storage center in the physical network uses the instruction model to perform data cleaning, knowledge representation and other processing on the acquired data to obtain the processed data (which can also be understood as represented knowledge).

[0098] Step 106: The data storage center in the physical network sends the processed data to the data storage center in the twin network so that the data storage center in the twin network can use the processed data to build a digital twin.

[0099] As described above, the data storage center in the physical network first processes the collected data (such as data cleaning and knowledge representation) to obtain processed data, and then transmits the processed data to the data storage center in the twin network to achieve data acquisition. In this process, because the processed data is smaller in size than the original data before processing, the communication resource consumption required for transmitting the processed data is lower, enabling efficient data transmission.

[0100] However, in practical applications, the above data acquisition methods have the following drawbacks:

[0101] 1) It heavily relies on pre-built instruction libraries;

[0102] 2) A data storage center needs to be deployed to centrally collect and process data related to digital twin tasks, which is costly;

[0103] 3) Data storage centers in physical networks process all collected datasets (such as data cleaning and knowledge representation), which requires significant computing power. When the computing resources of the data storage center are limited, there will be a large processing delay.

[0104] Based on this, in various embodiments of this application, when a second function (e.g., NWDAF) needs to construct a digital twin corresponding to a first task, the second function sends a data acquisition request to the first function (e.g., UPF) associated with the first task, the first function sends a data acquisition request to the access network device associated with the first task, and the access network device sends a data acquisition request to the terminal associated with the first task. Then, the terminal collects data related to the first task, performs feature characterization on the collected data to obtain the terminal's feature vector, and reports the terminal's feature vector to the access network device. The access network device performs fusion characterization on the received terminal's feature vector and the access network device's local data related to the first task to obtain the access network device's feature vector, and reports the access network device's feature vector to the first function. The first function performs relation extraction and joint feature learning on the received access network device's feature vector and the first function's local data related to the first task to obtain a tensor used to construct the digital twin, and reports the tensor to the second function. The second function uses the received tensor to construct the digital twin corresponding to the first task. In this process, the terminal, access network equipment, and the first functional side all process the collected data as needed (which can also be understood as on-network data representation or on-network data processing), and report the processed data layer by layer (the layers can include the terminals, access network equipment, the first function, and the second function, respectively) (which can also be understood as multi-level data transmission). Thus, the processed data (such as tensors or feature vectors) is smaller in size compared to the unprocessed data. Therefore, transmitting the processed data consumes fewer communication resources (such as bandwidth) compared to transmitting unprocessed data (which can also be understood as full metadata transmission). The overhead is lower; at the same time, compared with the scheme of centralized collection and centralized processing of digital twin-related data, when the network element nodes (such as terminals, access network equipment, and first functions) in the transmission path perform multi-level hierarchical representation of the collected data (i.e., multi-level on-network data representation is performed separately), the computing power requirement is distributed on multiple levels of devices, which is easier to meet, the processing latency is lower, and there is no need to deploy an additional storage center for centralized collection and processing of data, resulting in lower costs; in addition, the scheme provided in this application embodiment does not need to rely on a pre-built instruction library for data collection, and can flexibly collect data according to needs.

[0105] This application provides a data acquisition method, as shown in Figure 2, applied to an access network device, the method including:

[0106] Step 201: Receive first information sent by the first function, the first information being used to instruct the collection of data related to the first task, the first task being related to the digital twin, and the first function being used to process user plane data at least.

[0107] Step 202: Send a second message to one or more terminals, the second message being used to instruct the collection of data related to the first task;

[0108] Step 203: Receive first data reported by the one or more terminals, the first data containing one or more first vectors, the first vectors including vectors obtained by the terminals performing feature representation on the collected data;

[0109] Step 204: Perform a fusion representation on one or more of the received first data and second data to obtain a second vector, wherein the second data includes local data of the access network device related to the first task;

[0110] Step 205: Send third data to the first function, the third data containing the second vector.

[0111] In practical applications, the access network device can specifically be a base station, such as a gNB. This application embodiment does not limit the name of the access network device, as long as it performs its functions. The terminal can be called user equipment (UE), user, etc., and this application embodiment does not limit this. The first function is at least used for processing user plane data, and the first function may include a UPF; this application embodiment does not limit this.

[0112] In practical applications, at least the second function for network data analysis can be improved by constructing a digital twin to obtain more accurate and complete network data, thereby enhancing the accuracy and reliability of data analysis. This second function may include NWDAF, but this embodiment of the application does not limit its scope.

[0113] The second function can take building a digital twin as the first task and send an instruction message to the first function to collect data related to the first task (i.e., the data required to build the digital twin), so as to instruct the first function to collect data related to the first task (such as user plane data of the network).

[0114] Upon receiving the instruction information sent by the second function, the first function sends first information to one or more access network devices related to the first task, instructing the one or more access network devices to collect data related to the first task. Thus, when collecting data related to the first task, relevant data from the access network device side (which can also be understood as the domain where the access network device resides) can be collected, ensuring the diversity and richness of the collected data. The first information can also be called data collection instruction information; this application embodiment does not limit the name of the first information. The first function can send the first information to the one or more access network devices via user plane messages; that is, the first function sends the user plane messages to the one or more access network devices, and the user plane messages contain the first information. These user plane messages can be called data collection request messages.

[0115] After receiving the first information sent by the first function, in step 202, the access network device sends second information to one or more terminals related to the first task to instruct the one or more terminals to collect data related to the first task. Thus, when collecting data related to the first task, relevant data from the terminal side (which can also be understood as the domain where the terminal is located) can be collected, ensuring the diversity and richness of the collected data. The second information can also be called data collection instruction information; however, this embodiment does not limit the name of the second information.

[0116] In practical applications, the second information can be used to instruct the terminal to collect data related to the digital twin task, enabling the terminal to perform customized data collection and reporting. Simultaneously, the second information can also be used to instruct the terminal to perform feature characterization on the collected data, so that the terminal reports the vector obtained after feature characterization, rather than reporting metadata, thereby reducing the communication resources required for data transmission. That is, in one embodiment, the second information may include at least one or more of the following (or can be understood as at least one):

[0117] The third information is used to instruct the terminal to collect data related to the digital twin task;

[0118] The fourth information is used to instruct the terminal to perform feature characterization on the collected data.

[0119] In practical applications, the access network device can send the second information to each of the one or more terminals via RRC signaling. That is, in one embodiment, step 202 may specifically include: the access network device sending RRC signaling to one or more terminals, the RRC signaling containing the second information.

[0120] Specifically, the access network device can send RRC measurement configuration (RRC MeasConfig, i.e., the second information mentioned above) to the terminal to trigger the terminal to report data related to the digital twin task. RRC MeasConfig includes at least an event ID field and a reportQuantity field; the eventId field may contain a digital twin (DT) identifier (i.e., the third information mentioned above), where the DT identifier indicates the collection of digital twin-related data; the reportQuantity field may contain a single-mode aggregation (S-MA) method (i.e., the fourth information mentioned above), where S-MA indicates that the terminal's processing method for the collected data includes single-mode fusion characterization.

[0121] After receiving the second information, the terminal collects data related to the first task, performs feature characterization on the collected data to obtain one or more first vectors, and reports the one or more first vectors to the access network device.

[0122] In practical applications, the terminal can be configured with a Single-Modal Aggregation Representation Function (S-MAF). In this case, after collecting data related to the first task, each of the one or more terminals performs single-modal data representation on all the collected data based on the configured S-MAF function (i.e., feature representation for each modality (or data type)), obtaining feature vectors (i.e., the first vector) representing the data under one or more modalities. Then, the terminal aggregates the one or more feature vectors to obtain (or form) a single-modal vector set, and reports the vector set as the first data to the access network device. The first data can also be called single-modal representation information; this application embodiment does not limit the name of the first data.

[0123] For example, assuming that the data related to the first task collected by the terminal includes measurement report data, perception data, business data, semantic information, video data, single terminal behavior, etc., the terminal can use a suitable model (such as a Convolutional Neural Network (CNN) model, a Recurrent Neural Network (RNN) model, or a Transformer model, etc.) to perform single-modal data representation on the collected data, and obtain feature vectors F1, F2, ... FN corresponding to each modality, where N is the number of single modalities corresponding to the data collected by the terminal; then, the terminal can aggregate the feature vectors corresponding to each modality into a vector set (which can also be understood as a single-modal vector set) Vs = (F1, F2, ... FN), and report the vector set Vs to the access network device.

[0124] In practical applications, the terminal can encapsulate the first data (e.g., a vector set) in an RRC measurement report based on the RRC protocol, and send the first data to the access network device through the RRC measurement report. That is, the terminal sends an RRC measurement report to the access network device, and the RRC measurement report contains the first data. Accordingly, in one embodiment, step 203 may specifically include: receiving an RRC measurement report sent by each of the one or more terminals, the RRC measurement report containing the first data.

[0125] Upon receiving the first data reported by one or more terminals (i.e., data collected by the terminal side for the first task), the access network device performs fusion representation using all the first data and the second data to obtain a second vector that fuses the terminal-side data features and the access network-side data features. The second data includes local data related to the first task from the access network device, such as environmental perception measurement data (perception may include one or more of video perception, sound perception, image perception, sensor perception, etc. (one or more can also be understood as at least one)), network performance and management data, control plane (C-plane) data, user plane (U-plane) data, topology data, user behavior trajectory data, location data, log text data, etc. The second vector can also be called a multimodal fusion representation vector; however, this application embodiment does not limit the name of the second vector. In practical applications, a multimodal aggregation function (M-MAF) can be configured in the access network device to enable the access network device to perform fusion representation using all the first data and the second data.

[0126] In practical applications, the access network device can first extract the cross-correlation between all received first data to obtain one or more third vectors representing the cross-correlation between the first vectors; simultaneously, the access network device can use a suitable model to perform single-modal data representation on the local second data of the access network device to obtain one or more fourth vectors (i.e. feature vectors) corresponding to the modalities; then, the access network device can use the third and fourth vectors to perform feature fusion (which can also be understood as feature-level fusion) and decision fusion (which can also be understood as decision-level fusion) to obtain the second vector.

[0127] Specifically, in one embodiment, the implementation of step 204 may include:

[0128] Using all the first vectors contained in one or more of the first data, determine one or more third vectors, the third vectors representing the cross-correlation among all the first vectors; and perform feature characterization on the second data to obtain one or more fourth vectors;

[0129] The fifth vector is obtained by fusing the features of the one or more third vectors and the one or more fourth vectors.

[0130] The decision fusion of the one or more third vectors and the one or more fourth vectors yields the sixth vector;

[0131] The second vector is determined using the fifth vector and the sixth vector.

[0132] For example, suppose the access network device receives first data reported by three terminals, namely Vs1, Vs2, and Vs3, where Vs1 contains three first vectors F1, F2, and F3, Vs2 contains three first vectors F4, F5, and F6, and Vs3 contains three first vectors F7, F8, and F9; simultaneously, the access network device performs feature representation on the second data to obtain multiple fourth vectors Va1, Va2, Va3, ...; at this time, the step of the access network device determining the second vector may include:

[0133] Step 1: The access network device extracts the cross-correlation between F1, F2, ..., F9 to obtain multiple third vectors Vb1, Vb2, Vb3...;

[0134] Step 2: The access network device performs feature-level fusion using multiple third vectors and multiple fourth vectors to generate a local fusion representation vector VL (i.e., the fifth vector);

[0135] Step 3: The access network device uses a suitable model to make independent decisions on each of the multiple third vectors and multiple fourth vectors, generating multiple decision results D1, D2, D3, ...;

[0136] Step 4: The access network device performs decision-level fusion using multiple decision results to generate a decision fusion representation vector VD (i.e., the sixth vector);

[0137] Step 5: The access network device integrates the local fusion representation vector VL and the decision fusion representation vector VD to obtain the multimodal fusion representation vector VM (i.e., the second vector).

[0138] As can be seen from the above description, the access network device fuses and represents all the received first data and second data to obtain a second vector, which can effectively remove redundancy in the collected data, reduce the amount of data that needs to be transmitted, and thus reduce the consumption of communication resources.

[0139] After obtaining the second vector, the access network device reports the second vector to the first function, so that the first function can obtain the fusion representation of the data related to the first task from the access network device side and the terminal side, and provide corresponding data to the second function, so that the second function can use diverse and rich data to construct a digital twin related to the first task.

[0140] Accordingly, this application also provides a data acquisition method applied to a terminal, as shown in Figure 3. The method includes:

[0141] Step 301: Receive second information sent by the access network device, the second information being used to instruct the collection of data related to the first task, the first task being related to digital twin;

[0142] Step 302: Collect data related to the first task;

[0143] Step 303: Perform feature representation on the collected data to obtain one or more first vectors;

[0144] Step 304: Send first data to the access network device, the first data including the one or more first vectors.

[0145] In practical applications, the specific processing procedures on the access network device side have been detailed above and will not be repeated here.

[0146] In practical applications, the access network device can send second information to one or more terminals related to the first task to instruct the one or more terminals to collect data related to the first task. In this way, when collecting data related to the first task, relevant data from the terminal side (which can also be understood as the domain where the terminal is located) can be collected, ensuring the diversity and richness of the collected data.

[0147] In practical applications, the access network device can send the second information to each of the one or more terminals via RRC signaling. Accordingly, in one embodiment, step 301 may specifically include: receiving RRC signaling sent by the access network device, wherein the RRC signaling contains the second information.

[0148] In step 302, after receiving the second information, the terminal collects data related to the first task based on the instruction of the second information. Then, in step 303, the terminal performs single-modal data representation (i.e., feature representation for each modality (or data type)) on the collected data related to the first task to obtain feature vectors (i.e., the first vectors) representing the data under one or more modalities. Specifically, in one embodiment, the specific implementation of step 303 may include: performing feature representation on the collected data according to the data type to obtain one or more first vectors.

[0149] In practical applications, S-MAF can be configured in the terminal to enable the terminal to perform single-modal data representation on the collected data related to the first task.

[0150] When the terminal receives the second information sent by the access network device via RRC signaling, in step 304, after the terminal obtains one or more first vectors, it can encapsulate the first data (such as a vector set) in an RRC measurement report based on the RRC protocol, and send the first data to the access network device through the RRC measurement report. That is, in one embodiment, the specific implementation of step 304 may include: sending an RRC measurement report to the access network device, wherein the RRC measurement report contains the first data.

[0151] For example, the RRC measurement report format may include an eventId field, a data result field, a compression method field, and a timestamp field. The eventId field may include a DT identifier, the data result field may include a single-modal vector set Vs = (F1, F2, ... FN), and the compression method field may include the compression method corresponding to the first data.

[0152] Accordingly, this application also provides a data acquisition method applied to a first function, the first function being used to process user plane data, as shown in FIG4, the method comprising:

[0153] Step 401: Receive the fifth information sent by the second function, the fifth information being used to instruct the collection of data related to the first task, the first task being related to digital twins, and the second function being used at least for network data analysis;

[0154] Step 402: Send first information to one or more access network devices, the first information being used to instruct the collection of data related to the first task;

[0155] Step 403: Receive third data sent by the one or more access network devices, the third data including a second vector, the second vector including a vector obtained by the access network devices through fusion representation of the data related to the first task;

[0156] Step 404: Perform relation extraction and joint feature learning on the received one or more third data and fourth data to obtain a first tensor, wherein the fourth data contains data related to the first function and the first task.

[0157] Step 405: Send sixth data to the second function, the sixth data containing the first tensor.

[0158] In practical applications, at least the second function used for network data analysis (such as NWDAF) can improve the accuracy and reliability of data analysis by constructing digital twins to obtain more accurate and complete network data.

[0159] The second function may take the construction of a digital twin as its first task and send a fifth instruction to the first function to instruct the first function to collect data related to the first task (such as network user plane data). In one embodiment, the fifth instruction may include at least one or more of the following:

[0160] The task identifier of the first task, such as the task ID (which can be expressed as Twin Task ID in English), is used to indicate that the first task is related to the digital twin;

[0161] The twin range information for the first task is used to indicate which network elements (such as access network devices) contain data related to the first task for collection.

[0162] The real-time degree of the first task is used by the first function to determine the priority of processing the data streams related to the first task.

[0163] In practical applications, in step 401, the first function receives the fifth information sent by the second function and parses the fifth information. Specifically, in one embodiment, the method may further include:

[0164] Using the twin region information of the first task, determine the one or more access network devices; and / or,

[0165] The priority for sending the sixth data to the second function is determined by using the real-time level of the first task.

[0166] In practical applications, after receiving the fifth information, the first function uses the twin area information of the first task to determine that data related to the first task needs to be obtained from one or more access network devices corresponding to the twin area; then, in step 402, the first function sends the first information to the one or more access network devices to instruct the one or more access network devices to collect and report data related to the first task.

[0167] In practical applications, each of the one or more access network devices can collect data related to the first task after receiving the first information. Then, the access network device can use the M-MAF function to fuse and represent the collected data to obtain a second vector, and send third data containing the second vector to the first function. Accordingly, in step 403, the first function receives the third data sent by the one or more access network devices.

[0168] Upon receiving one or more of the third data, in step 404, the first function can perform relation extraction and joint feature learning on the received one or more third data and fourth data using methods such as multi-head attention mechanisms to obtain a first tensor adapted to the first task. The fourth data includes data locally related to the first task within the first function, such as user plane data (which may include terminal application data, session context data, etc.), network performance and management data (which may include traffic statistics, QoS-related data, etc.), or one or more of these. In practical applications, a Cross-Modal Aggregation Function (C-MAF) can be configured in the first function to enable it to perform relation extraction and joint feature learning on the received one or more third data and fourth data using methods such as multi-head attention mechanisms.

[0169] Specifically, in one embodiment, the implementation of step 404 may include:

[0170] The fourth data is characterized to obtain one or more seventh vectors;

[0171] For each of the one or more third data and fifth data, a fusion representation is performed based on a multi-head attention mechanism to obtain an eighth vector, wherein the fifth data includes the one or more seventh vectors;

[0172] Using all the obtained eighth vectors, determine one or more ninth vectors, wherein the ninth vectors characterize the cross-correlation among all the eighth vectors;

[0173] Joint feature learning is performed using one or more of the ninth vectors to obtain the learning results;

[0174] The first tensor is constructed using the learning results.

[0175] In practical applications, the first function can use a suitable model to perform single-modal data representation on the fourth data locally generated by the first function, obtaining one or more modal-corresponding seventh vectors (i.e., feature vectors). Then, the first function can use the obtained one or more seventh vectors to determine the fifth data, and use the fifth data and each third data as input data for a multi-head attention mechanism, processing them with multiple independent attention heads to obtain the eighth vector. The eighth vector can also be called an attention fusion representation vector; however, this application embodiment does not limit the name of the eighth vector.

[0176] Specifically, the process of processing each input data (such as the fifth or third data) using multiple independent attention heads may include:

[0177] For the i-th attention head (i.e., attention head i) in the multi-head attention mechanism, a query vector Qi, a key vector Ki, and a value vector Vi are generated using the weight matrix corresponding to attention head i and the input data. Then, the attention weight of attention head i is calculated using Qi, Ki, and Vi. Specifically, the attention weight of attention head i can be calculated using formula (1):

[0178] Where Attention(Qi,Ki,Vi) represents the attention value (or attention output) corresponding to attention head i; the dot product of the transposes of Qi and Ki represents the similarity matrix; dk represents the dimensions of Qi and Ki. The scaling factor is used to avoid the gradient vanishing problem caused by excessively large dot product results when the dimension is large. The softmax() function is used to convert the scaled dot product result into a probability distribution so that the sum of the attention weights corresponding to all attention heads is 1. The result calculated by the softmax() function can also be understood as the attention weight corresponding to attention head i. After obtaining the attention value corresponding to each attention head, all attention values ​​can be concatenated or weighted to obtain the eighth vector. The number of attention heads in the multi-head attention mechanism can be set according to actual needs, such as the number of modalities, etc., and this application embodiment does not limit this. The first function can determine the weight matrix corresponding to each attention head based on the task identifier of the first task included in the fifth information.

[0179] After obtaining one or more eighth vectors, the first function can process the one or more eighth vectors based on a relational modeling method to further extract relational features between modalities. Specifically, the first function can extract the cross-correlation between the eighth vectors to obtain one or more ninth vectors. The ninth vectors can also be understood as feature vectors after relational modeling. The names of the ninth vectors are not limited in this embodiment.

[0180] After obtaining one or more ninth vectors, the first function can perform joint feature learning on the one or more ninth vectors, such as using a method based on a self-attention mechanism to process the one or more ninth vectors to obtain learning results, so as to further capture the complex relationships between modalities.

[0181] After obtaining the learning results, the first function can combine the learning results into the first tensor (i.e., construct the first tensor). The first tensor contains the deep correlation information (or modal features) between the various modal data related to the first task collected by the first function.

[0182] In practical applications, after determining the first tensor, the first function uses the first tensor to determine the sixth data and sends the sixth data to the second function, so that the second function can use each received sixth data to determine the first tensor and use all the first tensors to construct a digital twin related to the first task.

[0183] Specifically, when the data size of the first tensor is large, the first function can optimize and / or compress the first tensor by performing tensor decomposition or other tensor operations to ensure that the first tensor can be transmitted correctly.

[0184] Based on this, in one embodiment, the specific implementation of step 405 may include:

[0185] The first tensor is decomposed into multiple decomposed tensors.

[0186] Send the plurality of decomposed tensors to the second function.

[0187] As can be seen from the above description, after acquiring data related to the first task, the first function extracts explicit and implicit relationships between the data and performs feature learning to remove data redundancy, obtaining a first tensor, and then reports the first tensor to the second function. Thus, compared to directly reporting all unprocessed data related to the first task (which can also be understood as reporting the full amount of data), reporting the first tensor effectively reduces communication resource consumption and enables more efficient data transmission.

[0188] Accordingly, this application also provides a data acquisition method for a second function, which is at least used for network data analysis, as shown in Figure 5. The method includes:

[0189] Step 501: Send a fifth message to one or more first functions, the fifth message being used to instruct the collection of data related to a first task, the first task being related to a digital twin, and the first function being used to process user plane data at least;

[0190] Step 502: Receive the sixth data reported by one or more first functions, the sixth data containing a first tensor, the first tensor containing a tensor obtained by the first function from relation extraction and joint feature learning of the data related to the first task;

[0191] Step 503: Construct a digital twin corresponding to the first task using all the first tensors.

[0192] In practical applications, the second function can construct a digital twin to obtain more accurate and complete network data, thereby improving the accuracy and reliability of data analysis.

[0193] In practical applications, the second function takes the construction of a digital twin as the first task; then, in step 501, the second function sends a fifth message to one or more first functions related to the first task to instruct the one or more first functions to collect data related to the first task (i.e., the data required to construct the digital twin).

[0194] In practical applications, after each of the one or more first functions receives the fifth information, it can collect data related to the first task according to the method described above for the first function side, and perform relation extraction and joint feature learning on the collected data to obtain a first tensor, and use the first tensor to determine the sixth data. Here, the operation of the first function side has been described in detail above, and will not be repeated here.

[0195] After determining the sixth data, the first function reports the sixth data to the second function; correspondingly, in step 502, the second function receives the sixth data sent by the first function.

[0196] Specifically, when the data size of the first tensor is large, the first function can optimize and / or compress the first tensor by performing tensor decomposition or other tensor operations on the first tensor to ensure that the first tensor can be transmitted correctly.

[0197] In one embodiment, when the first function performs tensor decomposition on the first tensor and transmits the decomposed tensor, the specific implementation of step 503 may include:

[0198] For each first function, receive multiple decomposed tensors sent by the first function; use the received multiple decomposed tensors to determine the sixth data.

[0199] In practical applications, the second function determines the corresponding first tensor for each received sixth data; then, the second function uses all the determined first tensors to construct a digital twin related to the first task, so as to improve the accuracy and reliability of the data analysis of the second function.

[0200] The data acquisition method provided in this application embodiment includes: a second function sending fifth information to one or more first functions, the fifth information indicating the acquisition of data related to a first task, the first task being related to a digital twin; the first function being used at least to process user plane data; and the second function being used at least for network data analysis; the first function receiving the fifth information sent by the second function; the first function sending first information to one or more access network devices, the first information indicating the acquisition of data related to the first task; the access network devices receiving the first information sent by the first function; the access network devices sending second information to one or more terminals, the second information indicating the acquisition of data related to the first task; the terminals receiving the second information sent by the access network devices; the terminals acquiring data related to the first task; the terminals performing feature characterization on the acquired data to obtain one or more first vectors; and the terminals sending first data to the access network devices, the first data packet being... The first function receives first data reported by one or more terminals, including one or more first vectors; the access network device performs fusion representation on the received one or more first data and second data to obtain a second vector, the second data including local data related to the first task of the access network device; the access network device sends third data to the first function, the third data including the second vector; the first function receives the third data sent by one or more access network devices; the first function performs relation extraction and joint feature learning on the received one or more third data and fourth data to obtain a first tensor, the fourth data including local data related to the first task of the first function; the first function sends sixth data to the second function, the sixth data including the first tensor; the second function receives the sixth data reported by one or more first functions; the second function uses all the first tensors to construct a digital twin corresponding to the first task.

[0201] The solution provided in this application embodiment involves the following steps: When a second function needs to construct a digital twin corresponding to a first task, the second function sends a data collection request to the first function associated with the first task (e.g., a User Plane Function (UPF)). The first function then sends a data collection request to the access network device associated with the first task, and the access network device sends a data collection request to the terminal associated with the first task. The terminal then collects data related to the first task, performs feature representation on the collected data to obtain a feature vector for the terminal, and reports the feature vector to the access network device. The access network device performs fusion representation on the received feature vector of the terminal and the local data related to the first task to obtain a feature vector for the access network device, and reports the feature vector to the first function. The first function performs relation extraction and joint feature learning on the received feature vector of the access network device and the local data related to the first task to obtain a tensor for constructing the digital twin, and reports the tensor to the second function. The second function then uses the received tensor to construct the digital twin corresponding to the first task. In this process, the terminal, access network equipment, and the first functional side all process the collected data as needed (which can also be understood as on-network data representation or on-network data processing), and report the processed data layer by layer (the layers can include the terminals, access network equipment, the first function, and the second function, respectively) (which can also be understood as multi-level data transmission). Thus, the processed data (such as tensors or feature vectors) is smaller in size compared to the unprocessed data. Therefore, transmitting the processed data consumes fewer communication resources (such as bandwidth) compared to transmitting unprocessed data (which can also be understood as full metadata transmission). The overhead is lower; at the same time, compared with the scheme of centralized collection and centralized processing of digital twin-related data, when the network element nodes (such as terminals, access network equipment, and first functions) in the transmission path perform multi-level hierarchical representation of the collected data (i.e., multi-level on-network data representation is performed separately), the computing power requirement is distributed on multiple levels of devices, which is easier to meet, the processing latency is lower, and there is no need to deploy an additional storage center for centralized collection and processing of data, resulting in lower costs; in addition, the scheme provided in this application embodiment does not need to rely on a pre-built instruction library for data collection, and can flexibly collect data according to needs.

[0202] The following section provides a more detailed description of this application with reference to application examples.

[0203] This application proposes a multi-level, high-efficiency data acquisition framework for NWDAF to the UE, Radio Access Network (RAN), and other domains, as shown in Figure 6. It adopts a multi-level, high-order fusion representation and transmission method, which can extract and represent data on demand at network element nodes, thereby achieving redundant and efficient data transmission. It eliminates the need for centralized data collection and processing, greatly reducing the consumption of data transmission communication resources, shortening data processing latency, and improving the efficiency of digital twin modeling.

[0204] To implement the above-mentioned transmission scheme for data representation processing during network transmission, functional enhancements need to be designed for relevant network elements, specifically including:

[0205] 1) A new S-MAF is added to the UE (i.e., the terminal mentioned above) side to realize the aggregation and representation of single-modal data such as measurement report data, perception data, service data, semantic information, video data, and single UE behavior, and form a single-modal vector set Vs = (F1, F2, F3...); where F1, F2... represent single-modal feature vectors (i.e. the first vector mentioned above);

[0206] 2) Add M-MAF on the gNB (i.e. the access network equipment mentioned above) side to collect multiple single-modal vector sets Vs of the underlying UE, extract the cross-correlation between different features, and then perform feature-level + decision-level fusion representation with the local data of gNB to form a multimodal fusion representation vector VM = F(VD, VL), where F represents the fusion function, VL represents the local fusion representation vector, and VD represents the decision fusion representation vector;

[0207] 3) Enhance computing power on the UPF (i.e., the first function mentioned above) side and achieve on-network programmability. At the same time, add C-MAF to collect multimodal fusion representation data from multiple different data sources (such as gNB, UE) at the bottom layer. Then, based on multi-head attention mechanism and other means, perform cross-modal relation extraction and joint feature learning on the collected multimodal fusion representation data to reduce data redundancy and form a high-order tensor TC = (M1, M2, M3...) adapted to the new twin task. Here, M1, M2... represent the modal features obtained through multi-head attention mechanism and joint feature learning.

[0208] 4) Building upon its intelligent decision-making and data analysis functions, NWDAF (i.e., the second function mentioned above) adds a Network Digital Twin Function (NDTF) to provide digital twin task processing services to other network elements or third-party consumers. In this case, digital twin task types can be defined in NWDAF, and the data type and its domain (i.e., twin region) can be defined according to the task type.

[0209] With the functional enhancement of relevant network elements, this application proposes an efficient data acquisition method for multi-level transmission in digital twins, as shown in Figure 7, which includes the following process:

[0210] Step 701: When NWDAF constructs a digital twin, it sends a request message to UPF (which can also be understood as sending) to collect the data required for constructing the digital twin (i.e., the fifth information mentioned above); wherein, the request message contains at least Twin Task ID, Twin range, and Real-time degree; then proceed to step 702.

[0211] Step 702: The UPF decomposes the request message, determines the data collection requirements, and sends a request message to the gNB to collect the data required for building the digital twin (i.e., the first information mentioned above); then proceeds to step 703.

[0212] Specifically, the UPF can determine that the request message is related to the twin task based on the Twin Task ID. After obtaining the data related to the twin task, the UPF first extracts and represents the explicit and implicit associations of the data streams corresponding to multiple network nodes, and then reports the extracted and represented data to the NWDAF without transmitting and forwarding the entire data stream (i.e., without reporting the full data). At the same time, the UPF can parse the Twin range field to determine which data—the twin task-related data of the network element objects—need to be collected. Additionally, the UPF can parse the Real-time degree field to determine the priority of processing the data stream.

[0213] Step 703: The gNB parses the request message sent by the UPF, determines the data collection requirement, and sends information to the UE instructing it to collect data (i.e., the second information mentioned above); then proceeds to step 704.

[0214] Step 704: The UE performs single-modal data aggregation and representation on multi-source data such as perception data, service data, semantic information, video data, and single UE behavior based on S-MAF;

[0215] Specifically, the UE can use a suitable model (such as CNN, RNN, Transformer, etc.) to extract the representation of a single modality, obtain the feature vectors F1, F2, ... after the data is represented in each modality, and aggregate them into a vector set Vs = (F1, F2, F3, ...); then, the UE can send the vector set Vs to the gNB.

[0216] In practical applications, to meet customized UE-side single-modal data acquisition and reporting requirements, steps 703 to 704 can be executed based on the improved RRC protocol. Specifically, the gNB can configure customized UE-side data acquisition requirements to the UE through RRC MeasConfig; after the UE parses the configuration information, it can collect configuration-related single-modal data, perform corresponding data processing through M-MAF, and report the processed data to the gNB. As shown in Figure 8, this can specifically include the following steps:

[0217] Step 801: The gNB determines the RRC configuration information so that the UE can report data via event-triggered reporting.

[0218] The RRC configuration information includes RRC MeasConfig, which can be represented as (eventId: DT, reportQuantity: S-MA). The DT identifier (i.e., the third information mentioned above) is used to instruct the UE to collect digital twin-related data, and the collected data has high priority and real-time requirements. The S-MA (i.e., the fourth information mentioned above) is used to instruct the UE to process the data through single-modal fusion representation and report the processed data, rather than directly reporting the unprocessed metadata.

[0219] Step 802: The gNB sends RRC configuration information to the UE (which can also be understood as data collection through RRC MeasConfig configuration sending);

[0220] Step 803: The UE parses the received RRC configuration information, determines the data collection and processing method, and performs data collection and processing according to the determined method to obtain the processed collection data (i.e., the vector set Vs mentioned above);

[0221] Specifically, the UE can determine, by parsing the eventId field, that when the eventId field value is "DT", to use a near real-time processing and reporting method that does not require threshold triggering for data processing and reporting, instead of using the traditional reporting method that only triggers reporting when a preset event threshold is reached; at the same time, the UE can determine, by parsing the reportQuantity field, to use S-MAF for corresponding data processing.

[0222] Step 804: The UE encapsulates the processed collected data into a measurement report based on the RRC protocol and transmits the measurement report to the gNB (which can also be understood as transmitting data encapsulated through the RRC protocol).

[0223] Specifically, the measurement report can be represented as [eventId:DT, DataResults: Single-modal vector set Vs, compression method, timestamp].

[0224] By executing steps 801 to 804, the gNB can collect the single-modal aggregated representation information Vs from the UE side, thus completing the digital twin-related data collection for the domain where the UE is located. Then step 705 can be executed.

[0225] Step 705: gNB extracts the cross-correlation between multiple single-modal features collected from the UE side, and performs feature-level + decision-level fusion representation with local data (i.e., gNB local data related to twin mission) to form a multimodal fusion representation vector VM (i.e. the second vector mentioned above), and reports the multimodal fusion representation vector VM to UPF; then execute step 706.

[0226] Specifically, gNB determining the modality fusion representation vector VM may include the following steps:

[0227] Step 1: Extract the cross-correlation between the single-modal feature vectors Vs1, Vs2, ... reported by the UE to obtain one or more feature vectors Vb1, Vb2, ... associated with the cross-correlation; at the same time, perform single-modal feature representation on the local data of gNB to obtain feature vectors VO1, VO2, ...

[0228] Step 2: Perform feature-level fusion using Vb1, Vb2, ... and VO1, VO2, ... to generate a local fusion representation vector VL;

[0229] Step 3: Use Vb1, Vb2, ... and VO1, VO2, ... to perform independent decision-making on single-modal data and generate decision results D1, D2, ...;

[0230] Step 4: Perform decision-level fusion using D1, D2, ... to generate the decision fusion representation vector VD;

[0231] Step 5: Combine the local fusion representation vector VL and the decision fusion representation vector VD to obtain the multimodal fusion representation vector VM.

[0232] Step 706: After UPF collects the multimodal fusion representation vector VM from multiple gNBs (which can also be understood as data sources), it performs cross-modal relation extraction and joint feature learning based on head attention mechanism and other means to form a high-order tensor Tc (i.e. the first tensor mentioned above) adapted to the Siamese task; then step 707 is executed.

[0233] Specifically, the UPF process for determining the high-order tensor Tc suitable for the Siamese task may include the following steps:

[0234] Step 1: UPF constructs a multi-head attention mechanism, and based on the multi-head attention mechanism, determines each multimodal fusion representation vector VM and the attention fusion representation vector VA corresponding to the single-modal fusion representation vector Vs-upf corresponding to the UPF local data;

[0235] UPF can determine the weight settings related to the multi-head attention mechanism based on the Twin Task ID.

[0236] Step 2: UPF performs cross-modal relation extraction;

[0237] Specifically, UPF uses cross-correlation or other relationship modeling methods to process all attention fusion representation vectors (VAs) to further extract relationship features between modalities; then, it performs joint learning on the relationship-modeled features to further capture the complex relationships between modalities.

[0238] Step 3: UPF generates a higher-order tensor Tc, which may include:

[0239] Tensor Construction: UPF combines the multimodal fusion representation VM after joint feature learning into a high-order tensor TC, which contains the deep correlation information of each modality;

[0240] Tensor output: Output high-order tensor Tc = (M1, M2, M3, ...), where M1, M2, ... represent modal features learned through multi-head attention mechanism and joint feature learning.

[0241] Step 707: NWDAF obtains the high-order tensor Tc reported by the UPF side and constructs a digital twin.

[0242] The data acquisition method proposed in this application adopts a multi-level data transmission framework, which does not rely on the construction of a centralized instruction library and can achieve efficient data transmission. At the same time, it adopts a technology of extracting and representing data on demand at network element nodes, eliminating the need for centralized data collection and processing, which greatly reduces the consumption of data transmission communication resources, shortens data processing latency, and improves the efficiency of digital twin modeling. In addition, a scheme is designed to configure the UE to collect and report single-modal fusion representation data on demand through RRC signaling, which can meet the needs of large collection volume and frequent collection in the scenario of collecting digital twin data.

[0243] To implement the method on the access network device side of this application embodiment, this application embodiment also provides a data acquisition device, which is installed on the access network device, as shown in FIG9. The device includes:

[0244] The first receiving unit 901 is configured to receive first information sent by the first function, the first information being used to indicate the collection of data related to the first task, the first task being related to a digital twin, and the first function being used to process user plane data; and to receive first data reported by one or more terminals, the first data containing one or more first vectors, the first vectors including vectors obtained by the terminals performing feature representation on the collected data.

[0245] The first processing unit 902 is used to fuse and characterize one or more received first data and second data to obtain a second vector, wherein the second data includes local data of the access network device related to the first task.

[0246] The first sending unit 903 is configured to send second information to one or more terminals, the second information being used to instruct the collection of data related to the first task; and to send third data to the first function, the third data including the second vector.

[0247] In one embodiment, the first processing unit 902 is specifically used for:

[0248] Using all the first vectors contained in one or more first data sets, determine one or more third vectors, whereby the third vectors characterize the cross-correlation among all the first vectors; and perform feature characterization on the second data sets to obtain one or more fourth vectors.

[0249] The fifth vector is obtained by fusing the features of one or more third vectors and one or more fourth vectors.

[0250] The decision fusion of the one or more third vectors and one or more fourth vectors yields the sixth vector;

[0251] The second vector is determined using the fifth and sixth vectors.

[0252] In one embodiment, the first transmitting unit 903 is specifically used for:

[0253] Send RRC signaling to one or more terminals, the RRC signaling containing the second information.

[0254] In one embodiment, the first receiving unit 901 is specifically used for:

[0255] Receive an RRC measurement report sent by each of the one or more terminals, the RRC measurement report containing the first data.

[0256] In practical applications, the first receiving unit 901 and the first sending unit 903 can be implemented by the communication interface in the data acquisition device, and the first processing unit 902 can be implemented by the processor in the data acquisition device.

[0257] To implement the terminal-side method of this application embodiment, this application embodiment also provides an information configuration device, which is installed on the terminal, as shown in FIG10. The device includes:

[0258] The second receiving unit 1001 is used to receive second information sent by the access network device. The second information is used to indicate the collection of data related to the first task, which is related to digital twin.

[0259] The second processing unit 1002 is used to collect data related to the first task; and to perform feature characterization on the collected data to obtain one or more first vectors.

[0260] The second sending unit 1003 is used to send first data to the access network device, the first data including the one or more first vectors.

[0261] In one embodiment, the second processing unit 1002 is specifically used for:

[0262] Based on the data type, the collected data is characterized to obtain one or more first vectors.

[0263] In one embodiment, the second receiving unit 1001 is specifically used for:

[0264] The access network device receives RRC signaling, which includes the second information.

[0265] In one embodiment, the second transmitting unit 1003 is specifically used for:

[0266] Send an RRC measurement report to the access network device, the RRC measurement report containing the first data.

[0267] In practical applications, the second receiving unit 1001 and the second sending unit 1003 can be implemented by the communication interface in the data acquisition device, and the second processing unit 1002 can be implemented by the processor in the data acquisition device.

[0268] To implement the method of the first functional side of the embodiments of this application, the embodiments of this application also provide an information configuration device, which is disposed on the first function, the first function being used at least to process user plane data, as shown in FIG11, the device includes:

[0269] The third receiving unit 1101 is configured to receive fifth information sent by the second function, the fifth information being used to indicate the collection of data related to the first task, the first task being related to digital twins, and the second function being used for at least network data analysis; and to receive third data sent by one or more access network devices, the third data containing a second vector, the second vector including a vector obtained by the access network devices through fusion representation of the data related to the first task.

[0270] The third processing unit 1102 is used to perform relation extraction and joint feature learning on one or more third data and fourth data received to obtain a first tensor, wherein the fourth data includes data related to the first function and the first task.

[0271] The third sending unit 1103 is configured to send first information to one or more access network devices, the first information being used to instruct the collection of data related to the first task; and to send sixth data to the second function, the sixth data containing the first tensor.

[0272] In one embodiment, the third processing unit 1102 is specifically used for:

[0273] The fourth data is characterized to obtain one or more seventh vectors;

[0274] For each of the one or more third data and fifth data, a fusion representation is performed based on a multi-head attention mechanism to obtain an eighth vector, wherein the fifth data includes the one or more seventh vectors;

[0275] Using all the obtained eighth vectors, determine one or more ninth vectors, wherein the ninth vectors characterize the cross-correlation among all the eighth vectors;

[0276] Joint feature learning is performed using one or more of the ninth vectors to obtain the learning results;

[0277] The first tensor is constructed using the learning results.

[0278] In one embodiment, the third transmitting unit 1103 is specifically used for:

[0279] The first tensor is decomposed into multiple decomposed tensors.

[0280] Send the plurality of decomposed tensors to the second function.

[0281] In one embodiment, the fifth information includes one or more of the following: twin region information of the first task; real-time level of the first task; the third processing unit 1102 is further configured to:

[0282] Using the twin region information of the first task, determine the one or more access network devices; and / or,

[0283] The priority for sending the sixth data to the second function is determined by using the real-time level of the first task.

[0284] In practical applications, the third receiving unit 1101 can be implemented by the communication interface in the data acquisition device, the third processing unit 1102 can be implemented by the processor in the data acquisition device, and the third sending unit 1103 can be implemented by the processor in the data acquisition device in combination with the communication interface.

[0285] To implement the method of the second functional side of the embodiments of this application, the embodiments of this application also provide an information configuration device, which is configured on the second function, the second function being used for at least network data analysis, as shown in FIG12, the device includes:

[0286] The fourth sending unit 1201 is used to send fifth information to one or more first functions, the fifth information being used to indicate the collection of data related to a first task, the first task being related to a digital twin, and the first function being used to process user plane data at least.

[0287] The fourth receiving unit 1202 is used to receive the sixth data reported by the one or more first functions. The sixth data includes a first tensor, which includes a tensor obtained by the first function performing relation extraction and joint feature learning on the data related to the first task.

[0288] The fourth processing unit 1203 is used to construct a digital twin corresponding to the first task using all the first tensors.

[0289] In one embodiment, the fourth receiving unit 1202 is specifically used for:

[0290] For each first function, receive multiple decomposed tensors sent by the first function; use the received multiple decomposed tensors to determine the sixth data.

[0291] In practical applications, the fourth transmitting unit 1201 can be implemented by the communication interface in the data acquisition device, the fourth processing unit 1203 can be implemented by the processor in the data acquisition device, and the fourth receiving unit 1202 can be implemented by the processor in the data acquisition device in combination with the communication interface.

[0292] It should be noted that the data acquisition device provided in the above embodiments is only illustrated by the division of the above-described program units. In practical applications, the above processing can be assigned to different program units as needed, that is, the internal structure of the device can be divided into different program units to complete all or part of the processing described above. In addition, the data acquisition device and the data acquisition method embodiments provided in the above embodiments belong to the same concept, and their specific implementation process can be found in the method embodiments, which will not be repeated here.

[0293] Based on the hardware implementation of the above program modules, and in order to implement the method on the access network device side of this application embodiment, this application embodiment also provides an access network device, as shown in FIG13, the access network device 1300 includes:

[0294] The first communication interface 1301 is capable of exchanging information with other devices;

[0295] The first processor 1302 is connected to the first communication interface 1301 to enable information interaction with other devices and to execute the methods provided by one or more technical solutions on the access network device side when running computer programs.

[0296] The computer program is stored in the first memory 1303.

[0297] Specifically, the first communication interface 1301 is used for:

[0298] The system receives first information sent by a first function, the first information indicating the collection of data related to a first task, the first task being related to a digital twin, and the first function being at least used to process user plane data; sends second information to one or more terminals, the second information indicating the collection of data related to the first task; receives first data reported by the one or more terminals, the first data containing one or more first vectors, the first vectors including vectors obtained by the terminals performing feature representation on the collected data; and sends third data to the first function, the third data containing the second vector.

[0299] The first processor 1302 is used for:

[0300] The received one or more first data and second data are fused and represented to obtain a second vector, wherein the second data includes local data of the access network device related to the first task.

[0301] In one embodiment, the first processor 1302 is specifically used for:

[0302] Using all the first vectors contained in one or more first data sets, determine one or more third vectors, whereby the third vectors characterize the cross-correlation among all the first vectors; and perform feature characterization on the second data sets to obtain one or more fourth vectors.

[0303] The fifth vector is obtained by fusing the features of one or more third vectors and one or more fourth vectors.

[0304] The decision fusion of the one or more third vectors and one or more fourth vectors yields the sixth vector;

[0305] The second vector is determined using the fifth and sixth vectors.

[0306] In one embodiment, the first communication interface 1301 is specifically used for:

[0307] Send RRC signaling to one or more terminals, the RRC signaling containing the second information.

[0308] In one embodiment, the first communication interface 1301 is specifically used for:

[0309] Receive an RRC measurement report sent by each of the one or more terminals, the RRC measurement report containing the first data.

[0310] It should be noted that the specific processing procedures of the first processor 1302 and the first communication interface 1301 can be understood by referring to the above method.

[0311] Of course, in practical applications, the various components in the access network device 1300 are coupled together through the bus system 1304. It can be understood that the bus system 1304 is used to realize the connection and communication between these components. In addition to the data bus, the bus system 1304 also includes a power bus, a control bus, and a status signal bus. However, for clarity, all buses are labeled as bus system 1304 in Figure 13.

[0312] The first memory 1303 in this embodiment is used to store various types of data to support the operation of the access network device 1300. Examples of such data include any computer program used to operate on the access network device 1300.

[0313] The methods disclosed in the above embodiments of this application can be applied to the first processor 1302, or implemented by the first processor 1302. The first processor 1302 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the integrated logic circuit of the hardware or by instructions in the form of software in the first processor 1302. The first processor 1302 may be a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The first processor 1302 can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor may be a microprocessor or any conventional processor, etc. The steps of the methods disclosed in the embodiments of this application can be directly manifested as being executed by a hardware decoding processor, or being executed by a combination of hardware and software modules in the decoding processor. The software modules may be located in a storage medium, which is located in the first memory 1303. The first processor 1302 reads the information in the first memory 1303 and completes the steps of the aforementioned method in combination with its hardware.

[0314] In an exemplary embodiment, the access network device 1300 may be implemented by one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), general-purpose processors, controllers, microcontrollers (MCUs), microprocessors, or other electronic components to perform the aforementioned method.

[0315] Based on the hardware implementation of the above program modules, and in order to implement the terminal-side method of this application embodiment, this application embodiment also provides a terminal, as shown in FIG14, the terminal 1400 including:

[0316] The second communication interface 1401 is capable of exchanging information with other devices;

[0317] The second processor 1402 is connected to the second communication interface 1401 to enable information interaction with other devices and to execute the methods provided by one or more of the above-mentioned terminal-side technical solutions when running computer programs.

[0318] The computer program is stored in the second memory 1403.

[0319] Specifically, the second communication interface 1401 is used for:

[0320] The system receives second information sent by an access network device, the second information being used to instruct the collection of data related to a first task, the first task being related to a digital twin; and sends first data to the access network device, the first data containing one or more first vectors.

[0321] The second processor 1402 is used for:

[0322] Collect data related to the first task; and perform feature characterization on the collected data to obtain one or more first vectors.

[0323] In one embodiment, the second processor 1402 is specifically used for:

[0324] Based on the data type, the collected data is characterized to obtain one or more first vectors.

[0325] In one embodiment, the second communication interface 1401 is specifically used for:

[0326] The access network device receives RRC signaling, which includes the second information.

[0327] In one embodiment, the second communication interface 1401 is specifically used for:

[0328] Send an RRC measurement report to the access network device, the RRC measurement report containing the first data.

[0329] It should be noted that the specific processing procedures of the second processor 1402 and the second communication interface 1401 can be understood by referring to the above method.

[0330] Of course, in practical applications, the various components in terminal 1400 are coupled together through bus system 1404. It can be understood that bus system 1404 is used to realize the connection and communication between these components. In addition to the data bus, bus system 1404 also includes a power bus, a control bus, and a status signal bus. However, for clarity, all buses are labeled as bus system 1404 in Figure 14.

[0331] The second memory 1403 in this embodiment is used to store various types of data to support the operation of the terminal 1400. Examples of such data include any computer program used to operate on the terminal 1400.

[0332] The methods disclosed in the embodiments of this application can be applied to, or implemented by, the second processor 1402. The second processor 1402 may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method can be completed by the integrated logic circuitry of the hardware or by instructions in the form of software within the second processor 1402. The second processor 1402 may be a general-purpose processor, a DSP, or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The second processor 1402 can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor may be a microprocessor or any conventional processor, etc. The steps of the methods disclosed in the embodiments of this application can be directly manifested as execution by a hardware decoding processor, or execution by a combination of hardware and software modules in the decoding processor. The software modules may be located in a storage medium, specifically a second memory 1403. The second processor 1402 reads information from the second memory 1403 and, in conjunction with its hardware, completes the steps of the aforementioned method.

[0333] In an exemplary embodiment, terminal 1400 may be implemented by one or more ASICs, DSPs, PLDs, CPLDs, FPGAs, general-purpose processors, controllers, MCUs, microprocessors, or other electronic components to perform the aforementioned method.

[0334] Based on the hardware implementation of the above program modules, and in order to implement the method of the first functional side of the embodiments of this application, the embodiments of this application also provide a first function, which is at least used to process user plane data, as shown in FIG15. The first function 1500 includes:

[0335] The third communication interface 1501 is capable of exchanging information with other devices;

[0336] The third processor 1502 is connected to the third communication interface 1501 to enable information interaction with other devices and to execute the methods provided by one or more technical solutions of the first functional side when running a computer program.

[0337] The computer program is stored in the third memory 1503.

[0338] Specifically, the third communication interface 1501 is used for:

[0339] The system receives a fifth message from a second function, the fifth message indicating the collection of data related to a first task, the first task being related to a digital twin, and the second function being at least used for network data analysis; sends a first message to one or more access network devices, the first message indicating the collection of data related to the first task; receives third data from the one or more access network devices, the third data containing a second vector, the second vector including a vector obtained by the access network devices through fusion representation of the data related to the first task; and sends sixth data to the second function, the sixth data containing a first tensor.

[0340] The third processor 1502 is used for:

[0341] Relation extraction and joint feature learning are performed on one or more third data and fourth data received to obtain a first tensor, wherein the fourth data contains data related to the first function locally and the first task.

[0342] In one embodiment, the third processor 1502 is specifically used for:

[0343] The fourth data is characterized to obtain one or more seventh vectors;

[0344] For each of the one or more third data and fifth data, a fusion representation is performed based on a multi-head attention mechanism to obtain an eighth vector, wherein the fifth data includes the one or more seventh vectors;

[0345] Using all the obtained eighth vectors, determine one or more ninth vectors, wherein the ninth vectors characterize the cross-correlation among all the eighth vectors;

[0346] Joint feature learning is performed using one or more of the ninth vectors to obtain the learning results;

[0347] The first tensor is constructed using the learning results.

[0348] In one embodiment, the third processor 1502 is specifically used for:

[0349] The first tensor is decomposed into multiple decomposed tensors.

[0350] The decomposed tensors are sent to the second function via the third communication interface 1501.

[0351] In one embodiment, the fifth information includes one or more of the following: twin region information of the first task; real-time level of the first task; the third processor 1502 is further configured to:

[0352] Using the twin region information of the first task, determine the one or more access network devices; and / or,

[0353] The priority for sending the sixth data to the second function is determined by using the real-time level of the first task.

[0354] It should be noted that the specific processing procedures of the third processor 1502 and the third communication interface 1501 can be understood by referring to the above method.

[0355] Of course, in practical applications, the various components in the first function 1500 are coupled together through the bus system 1504. It can be understood that the bus system 1504 is used to realize the connection and communication between these components. In addition to the data bus, the bus system 1504 also includes a power bus, a control bus, and a status signal bus. However, for clarity, all buses are labeled as bus system 1504 in Figure 15.

[0356] The third memory 1503 in this embodiment is used to store various types of data to support the operation of the first function 1500. Examples of such data include any computer program used to operate on the first function 1500.

[0357] The methods disclosed in the embodiments of this application can be applied to, or implemented by, the third processor 1502. The third processor 1502 may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method can be completed by the integrated logic circuitry of the hardware or by instructions in the software form of the third processor 1502. The third processor 1502 may be a general-purpose processor, a DSP, or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The third processor 1502 can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor may be a microprocessor or any conventional processor, etc. The steps of the methods disclosed in the embodiments of this application can be directly manifested as execution by a hardware decoding processor, or execution by a combination of hardware and software modules in the decoding processor. The software modules may be located in a storage medium, specifically a third memory 1503. The third processor 1502 reads information from the third memory 1503 and, in conjunction with its hardware, completes the steps of the aforementioned method.

[0358] In an exemplary embodiment, the first function 1500 may be implemented by one or more ASICs, DSPs, PLDs, CPLDs, FPGAs, general-purpose processors, controllers, MCUs, microprocessors, or other electronic components to perform the aforementioned method.

[0359] Based on the hardware implementation of the above program modules, and in order to implement the method of the second functional side of the embodiments of this application, the embodiments of this application also provide a second function, which is at least used for network data analysis, as shown in FIG16. The second function 1600 includes:

[0360] The fourth communication interface 1601 enables information exchange with other devices;

[0361] The fourth processor 1602 is connected to the fourth communication interface 1601 to enable information interaction with other devices and to execute the methods provided by one or more of the above-mentioned second functional side technical solutions when running computer programs.

[0362] The fourth memory 1603, on which the computer program is stored.

[0363] Specifically, the fourth communication interface 1601 is used for:

[0364] Send a fifth message to one or more first functions, the fifth message being used to instruct the collection of data related to a first task, the first task being related to a digital twin, the first function being used to process user plane data at least; and receive a sixth data reported by the one or more first functions, the sixth data containing a first tensor, the first tensor containing a tensor obtained by the first function from relation extraction and joint feature learning of the data related to the first task.

[0365] The fourth processor 1602 is used for:

[0366] Construct a digital twin corresponding to the first task using all the first tensors.

[0367] In one embodiment, the fourth processor 1602 is specifically used for:

[0368] For each first function, multiple decomposed tensors sent by the first function are received through the fourth communication interface 1601; the sixth data is determined using the received multiple decomposed tensors.

[0369] It should be noted that the specific processing procedures of the fourth processor 1602 and the fourth communication interface 1601 can be understood by referring to the above method.

[0370] Of course, in practical applications, the various components in the second function 1600 are coupled together through the bus system 1604. It can be understood that the bus system 1604 is used to realize the connection and communication between these components. In addition to the data bus, the bus system 1604 also includes a power bus, a control bus, and a status signal bus. However, for clarity, all buses are labeled as bus system 1604 in Figure 16.

[0371] The fourth memory 1603 in this embodiment is used to store various types of data to support the operation of the second function 1600. Examples of such data include any computer program used to operate on the second function 1600.

[0372] The methods disclosed in the embodiments of this application can be applied to, or implemented by, the fourth processor 1602. The fourth processor 1602 may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method can be completed by the integrated logic circuitry of the hardware or by instructions in the software form of the fourth processor 1602. The fourth processor 1602 may be a general-purpose processor, a DSP, or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The fourth processor 1602 can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor may be a microprocessor or any conventional processor, etc. The steps of the methods disclosed in the embodiments of this application can be directly manifested as execution by a hardware decoding processor, or execution by a combination of hardware and software modules in the decoding processor. The software modules may be located in a storage medium, specifically a fourth memory 1603. The fourth processor 1602 reads information from the fourth memory 1603 and, in conjunction with its hardware, completes the steps of the aforementioned method.

[0373] In an exemplary embodiment, the second function 1600 may be implemented by one or more ASICs, DSPs, PLDs, CPLDs, FPGAs, general-purpose processors, controllers, MCUs, microprocessors, or other electronic components to perform the aforementioned method.

[0374] It is understood that the memories (first memory 1303, second memory 1403, third memory 1503, and fourth memory 1603) in the embodiments of this application can be volatile memory or non-volatile memory, or both. Non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), ferromagnetic random access memory (FRAM), flash memory, magnetic surface memory, optical disc, or compact disc read-only memory (CD-ROM); magnetic surface memory can be disk storage or magnetic tape storage. Volatile memory can be random access memory (RAM), which is used as an external cache.By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Synchronous Static Random Access Memory (SSRAM), Dynamic Random Access Memory (DRAM), Synchronous Dynamic Random Access Memory (SDRAM), Double Data Rate Synchronous Dynamic Random Access Memory (DDRSDRAM), Enhanced Synchronous Dynamic Random Access Memory (ESDRAM), SyncLink Dynamic Random Access Memory (SLDRAM), and Direct Rambus Random Access Memory (DRRAM). The memories described in the embodiments of this application are intended to include, but are not limited to, these and any other suitable types of memory.

[0375] In an exemplary embodiment, this application also provides a storage medium, namely a computer storage medium, specifically a computer-readable storage medium. For example, it may include a first memory 1303 storing a computer program, which can be executed by a first processor 1302 of an access network device 1300 to complete the steps described in the access network device-side method. Another example is a second memory 1403 storing a computer program, which can be executed by a second processor 1402 of a terminal 1400 to complete the steps described in the terminal-side method. Yet another example is a third memory 1503 storing a computer program, which can be executed by a third processor 1502 of a first function 1500 to complete the steps described in the first function-side method. Yet another example is a fourth memory 1603 storing a computer program, which can be executed by a fourth processor 1602 of a second function 1600 to complete the steps described in the second function-side method. The computer-readable storage medium may be a memory such as FRAM, ROM, PROM, EPROM, EEPROM, Flash Memory, magnetic surface memory, optical disc, or CD-ROM.

[0376] In an exemplary embodiment, this application also provides a computer program product, including a computer program that can be executed by a first processor 1302 of an access network device 1300 to complete the steps of the aforementioned access network device-side method; or, the computer program can be executed by a second processor 1402 of a terminal 1400 to complete the steps of the aforementioned terminal-side method; or, the computer program can be executed by a third processor 1502 of a first function 1500 to complete the steps of the aforementioned first function-side method; or, the computer program can be executed by a fourth processor 1602 of a second function 1600 to complete the steps of the aforementioned second function-side method.

[0377] To implement the method of the embodiments of this application, the embodiments of this application also provide a data acquisition system, as shown in FIG17. The system includes: a second function 1701, one or more first functions 1702, one or more access network devices 1703, and one or more terminals 1704.

[0378] It should be noted that the specific processing procedures of the second function 1701, the first function 1702, the access network device 1703, and the terminal 1704 have been detailed above and will not be repeated here.

[0379] It should be noted that terms such as "first" and "second" are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0380] Furthermore, the technical solutions described in the embodiments of this application can be combined arbitrarily without conflict.

[0381] The above description is merely a preferred embodiment of this application and is not intended to limit the scope of protection of this application.

Claims

1. A data acquisition method applied to access network equipment, wherein, The method includes: Receive first information sent by a first function, the first information being used to instruct the collection of data related to a first task, the first task being related to a digital twin, and the first function being used to process user plane data at least; Send a second message to one or more terminals, the second message being used to instruct the collection of data related to the first task; Receive first data reported by one or more terminals, the first data containing one or more first vectors, the first vectors including vectors obtained by the terminals performing feature representation on the collected data; The received one or more of the first data and the second data are fused and represented to obtain a second vector, wherein the second data includes local data of the access network device related to the first task; Send third data to the first function, the third data containing the second vector.

2. The method according to claim 1, wherein, The step of fusing and representing one or more of the received first data and second data to obtain a second vector includes: Using all the first vectors contained in one or more of the first data, determine one or more third vectors, the third vectors representing the cross-correlation among all the first vectors; and perform feature characterization on the second data to obtain one or more fourth vectors; The fifth vector is obtained by fusing the features of the one or more third vectors and the one or more fourth vectors. The decision fusion of the one or more third vectors and the one or more fourth vectors yields the sixth vector; The second vector is determined using the fifth vector and the sixth vector.

3. The method according to claim 1, wherein, Sending the second information to one or more terminals includes: Send Radio Resource Control (RRC) signaling to one or more terminals, wherein the RRC signaling contains the second information.

4. The method according to claim 3, wherein, The second information includes at least one or more of the following: The third information is used to instruct the terminal to collect data related to the digital twin task; The fourth information is used to instruct the terminal to perform feature characterization on the collected data.

5. The method according to claim 1, characterized in that, The receiving of the first data reported by the one or more terminals includes: Receive an RRC measurement report sent by each of the one or more terminals, the RRC measurement report containing the first data.

6. A data acquisition method applied to a terminal, wherein, The method includes: Receive second information sent by the access network device, the second information being used to instruct the collection of data related to the first task, the first task being related to digital twin; Collect data related to the first task; The collected data is characterized to obtain one or more first vectors; Send first data to the access network device, the first data including the one or more first vectors.

7. The method according to claim 6, wherein, The process of performing feature characterization on the collected data to obtain one or more first vectors includes: Based on the data type, the collected data is characterized to obtain one or more first vectors.

8. The method according to claim 6, wherein, The second information received from the access network device includes: The device receives RRC (Radio Resource Control) signaling sent by the access network device, the RRC signaling containing the second information.

9. The method according to claim 8, wherein, The second information includes at least one or more of the following: The third information is used to instruct the terminal to collect data related to the digital twin task; The fourth information is used to instruct the terminal to perform feature characterization on the collected data.

10. The method according to claim 6, wherein, Sending the first data to the access network device includes: Send an RRC measurement report to the access network device, the RRC measurement report containing the first data.

11. A data acquisition method, applied to a first function, wherein, The first function is used to process user plane data at least, and the method includes: The system receives a fifth message sent by the second function, the fifth message being used to instruct the collection of data related to the first task, the first task being related to digital twins, and the second function being used at least for network data analysis. Send a first message to one or more access network devices, the first message being used to instruct the collection of data related to the first task; Receive third data sent by the one or more access network devices, the third data including a second vector, the second vector including a vector obtained by the access network device by fusing and representing the data related to the first task; Relation extraction and joint feature learning are performed on one or more third data and fourth data received to obtain a first tensor, wherein the fourth data includes data related to the first function and the first task. Send sixth data to the second function, the sixth data containing the first tensor.

12. The method according to claim 11, wherein, Relation extraction and joint feature learning are performed on one or more received third and fourth data to obtain a first tensor, including: The fourth data is characterized to obtain one or more seventh vectors; For each of the one or more third data and fifth data, a fusion representation is performed based on a multi-head attention mechanism to obtain an eighth vector, wherein the fifth data includes the one or more seventh vectors; Using all the obtained eighth vectors, determine one or more ninth vectors, wherein the ninth vectors characterize the cross-correlation among all the eighth vectors; Joint feature learning is performed using one or more of the ninth vectors to obtain the learning results; The first tensor is constructed using the learning results.

13. The method according to claim 11, wherein, Sending the sixth data to the second function includes: The first tensor is decomposed into multiple decomposed tensors. Send the multiple decomposed tensors to the second function.

14. The method according to any one of claims 11 to 13, wherein, The fifth piece of information includes one or more of the following: twin region information of the first task; real-time level of the first task; the method further includes: Using the twin region information of the first task, determine the one or more access network devices; and / or, The priority for sending the sixth data to the second function is determined by using the real-time level of the first task.

15. A data acquisition method, applied to a second function, wherein, The second function is used at least for network data analysis, and the method includes: Send a fifth message to one or more first functions, the fifth message being used to instruct the collection of data related to a first task, the first task being related to a digital twin, and the first function being used to process user plane data at least; Receive sixth data reported by one or more first functions, the sixth data containing a first tensor, the first tensor containing a tensor obtained by the first function from relation extraction and joint feature learning of the data related to the first task; Using all of the first tensors, construct the digital twin corresponding to the first task.

16. The method according to claim 15, wherein, The receiving of the sixth data reported by the one or more first functions includes: For each of the first functions, receive multiple decomposed tensors sent by the first function; use the received multiple decomposed tensors to determine the sixth data.

17. A data acquisition device, installed in access network equipment, wherein, include: The first receiving unit is configured to receive first information sent by a first function, the first information being used to indicate the collection of data related to a first task, the first task being related to a digital twin, and the first function being used to process user plane data at least; and to receive first data reported by one or more terminals, the first data containing one or more first vectors, the first vectors including vectors obtained by the terminals performing feature representation on the collected data. The first processing unit is configured to fuse and characterize one or more of the received first data and second data to obtain a second vector, wherein the second data includes local data of the access network device related to the first task. A first sending unit is configured to send second information to one or more terminals, the second information being used to instruct the collection of data related to the first task; and to send third data to the first function, the third data including the second vector.

18. A data acquisition device, wherein, include: The second receiving unit is used to receive second information sent by the access network device. The second information is used to indicate the collection of data related to the first task, which is related to the digital twin. The second processing unit is used to collect data related to the first task; And perform feature characterization on the collected data to obtain one or more first vectors; The second sending unit is configured to send first data to the access network device, the first data including the one or more first vectors.

19. A data acquisition device, configured with a first function, wherein the first function is at least used for processing user plane data, wherein... include: The third receiving unit is configured to receive fifth information sent by the second function, the fifth information being used to indicate the collection of data related to the first task, the first task being related to digital twins, and the second function being used at least for network data analysis; and to receive third data sent by one or more access network devices, the third data containing a second vector, the second vector including a vector obtained by the access network devices through fusion representation of the data related to the first task. The third processing unit is used to perform relation extraction and joint feature learning on one or more received third data and fourth data to obtain a first tensor, wherein the fourth data includes data related to the first function and the first task. The third sending unit is configured to send first information to one or more access network devices, the first information being used to instruct the collection of data related to the first task; and to send sixth data to the second function, the sixth data containing the first tensor.

20. A data acquisition device, wherein, include: The fourth sending unit is used to send fifth information to one or more first functions, the fifth information being used to indicate the collection of data related to a first task, the first task being related to a digital twin, and the first function being used to process user plane data at least. The fourth receiving unit is configured to receive the sixth data reported by the one or more first functions, the sixth data including a first tensor, the first tensor including a tensor obtained by the first function from relation extraction and joint feature learning of data related to the first task. The fourth processing unit is used to construct a digital twin corresponding to the first task using all of the first tensors.

21. An access network device, wherein, include: A first processor and a first communication interface; wherein... The first communication interface is used to receive first information sent by a first function, the first information being used to indicate the collection of data related to a first task, the first task being related to a digital twin, and the first function being used to process user plane data at least; and to send second information to one or more terminals, the second information being used to indicate the collection of data related to the first task. Receive first data reported by one or more terminals, the first data containing one or more first vectors, the first vectors including vectors obtained by the terminals performing feature representation on the collected data; send third data to the first function, the third data containing a second vector; The first processor is configured to fuse and characterize one or more of the received first data and second data to obtain a second vector, wherein the second data includes local data of the access network device related to the first task.

22. A terminal, wherein, include: A second processor and a second communication interface; wherein... The second communication interface is used to receive second information sent by the access network device, the second information being used to instruct the collection of data related to a first task, the first task being related to a digital twin; and to send first data to the access network device, the first data containing one or more first vectors; The second processor is used to collect data related to the first task; and to perform feature characterization on the collected data to obtain one or more first vectors.

23. A first function, the first function being used at least to process user plane data, including: A third processor and a third communication interface; wherein... The third processor is configured to receive fifth information sent by the second function, the fifth information indicating the collection of data related to a first task, the first task being related to a digital twin, and the second function being at least used for network data analysis; send first information to one or more access network devices, the first information indicating the collection of data related to the first task; receive third data sent by the one or more access network devices, the third data containing a second vector, the second vector including a vector obtained by the access network devices through fusion representation of the data related to the first task; and send sixth data to the second function, the sixth data containing a first tensor. The third processor is used to perform relation extraction and joint feature learning on one or more received third data and fourth data to obtain a first tensor, wherein the fourth data includes data related to the first function locally and the first task.

24. A second function, the second function being used at least for network data analysis, comprising: The fourth processor and the fourth communication interface; wherein, The fourth communication interface is used to send fifth information to one or more first functions, the fifth information being used to instruct the collection of data related to a first task, the first task being related to a digital twin, and the first function being used to process user plane data at least; and to receive sixth data reported by the one or more first functions, the sixth data containing a first tensor, the first tensor containing a tensor obtained by the first function from relation extraction and joint feature learning of the data related to the first task. The fourth processor is used to construct a digital twin corresponding to the first task using all the first tensors.

25. An access network device, wherein, include: A first processor and a first memory for storing computer programs capable of running on the processor. Wherein, when the first processor is used to run the computer program, it performs the steps of the method according to any one of claims 1 to 5.

26. A terminal, wherein, include: A second processor and a second memory for storing computer programs that can run on the processor. Wherein, when the second processor is used to run the computer program, it performs the steps of the method according to any one of claims 6 to 10.

27. A primary function, wherein, include: A third processor and a third memory for storing computer programs that can run on the processor. When the third processor runs the computer program, it performs the steps of the method according to any one of claims 11 to 14.

28. A second function, wherein, include: A fourth processor and a fourth memory for storing computer programs that can run on the processor. When the fourth processor runs the computer program, it performs the steps of the method according to any one of claims 15 to 16.

29. A storage medium having a computer program stored thereon, wherein, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 5, or the steps of the method according to any one of claims 6 to 10, or the steps of the method according to any one of claims 11 to 14, or the steps of the method according to any one of claims 15 to 16.

30. A computer program product comprising a computer program, wherein, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 5, or the steps of the method according to any one of claims 6 to 10, or the steps of the method according to any one of claims 11 to 14, or the steps of the method according to any one of claims 15 to 16.

Images