Method for executing AI-ML data collection, user equipment and storage medium

Abstract

Provided in an embodiment of the present invention are a method for performing AI-ML data collection, user equipment and a storage medium, the method for performing AI-ML data collection comprising: receiving, by a UE, a data collection request initiated by a UE server, the UE being connected with an RAN node of a wireless network; obtaining a data collection configuration from the RAN node, wherein the data collection configuration configures AI-ML model related parameters for the UE; performing data collection according to the data collection configuration to collect AI-ML model related data; and sending the collected AI-ML model related data of which the destination is the UE server through a transmission channel. According to the method and the device, AI-ML data collection can be better executed.

Description

Technical Field

[0001] The present invention relates to wireless communications, and more particularly to data collection without radio access network (RAN) awareness. Background Art

[0002] Artificial intelligence (AI) and machine learning (ML) have been widely used in wireless networks to improve performance, user experience and reduce complexity / overhead. In traditional networks of 3GPP 5G new radio (NR), AI / ML techniques can be leveraged to address the challenges caused by the foreseen increase in complexity of air interface deployments, which applies to both the network and user equipment (UE). Data collection is a key step in AI / ML life cycle management (LCM), which provides the basis for creating effective and accurate models.

[0003] In wireless AI technology, offline training is the most viable option and the foundation of wireless AI. For UE-side models, training can be performed on the UE side or at a neutral site. However, it is impractical to perform UE-side model training within the UE due to the lack of a suitable training environment. This includes but is not limited to data availability, storage capacity, computing power, and compilation capabilities. For effective offline model training, data collection solutions should be designed to "reach" the UE side or a neutral site, such as an over the top (OTT) server. Moreover, a large amount of non-standardized / proprietary data should be collected, which can be customized according to various UE internal / external conditions and design choices (e.g., UE resource limitations, radio environment, feature engineering). In addition, the collected data should be sufficient for good model generalization.

[0004] Improvements and enhancements to data collection are needed without RAN awareness. Summary of the invention

[0005] An embodiment of the present invention provides a method for performing AI-ML data collection, comprising: receiving, by a UE, a data collection request initiated by a UE server, wherein the UE is connected to a RAN node of a wireless network; obtaining a data collection configuration from the RAN node, wherein the data collection configuration configures AI-ML model-related parameters for the UE; performing data collection according to the data collection configuration to collect AI-ML model-related data; and sending the collected AI-ML model-related data to the UE server through a transmission channel. By utilizing the present invention, AI-ML data collection can be better performed.

[0006] An embodiment of the present invention provides a method for performing AI-ML data collection, including: initiating, by a UE server, a data collection request with a destination of one or more UEs, wherein the data collection request starts data collection for the one or more UEs for an AI-ML model used by the one or more UEs; receiving one or more data collection responses; and receiving AI-ML model-related data and assistance information through a data channel connected to a wireless network.

[0007] An embodiment of the present invention provides a user equipment, including: a transceiver, used to transmit and receive radio frequency signals in a wireless network; a detection module, used to receive a data collection request initiated by a UE server, wherein the UE is connected to a RAN node of the wireless network; a configuration module, used to obtain a data collection configuration from the RAN node, wherein the data collection configuration configures AI-ML model-related parameters for the UE; a collection module, used to perform data collection according to the data collection configuration to collect AI-ML model-related data; and a transmission module, used to send the collected AI-ML model-related data with the destination being the UE server through a transmission channel.

[0008] An embodiment of the present invention provides a storage medium storing a program, which, when executed, enables a device to perform the steps of the method for performing AI-ML data collection in the present application.

[0009] By utilizing the present invention, AI-ML data collection can be better performed. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] The accompanying drawings illustrate embodiments of the present invention, wherein like numerals indicate like components.

[0011] Figure 1 is a system diagram of an exemplary wireless network supporting data collection according to an embodiment of the present invention.

[0012] Figure 2 This is an example process of collecting data to a UE server without RAN awareness according to an embodiment of the present invention.

[0013] Figure 3 This is an example process of triggering data collection from a UE server to a UE without involving a network node according to an embodiment of the present invention.

[0014] Figure 4 This is an example process of data collection triggering from a UE server to a CN according to an embodiment of the present invention.

[0015] Figure 5 This is an example process of triggering data collection from a UE server to a CN entity and from a UE server to a UE without RAN awareness according to an embodiment of the present invention.

[0016] Figure 6 This is an example process of a UE performing AI-ML data collection according to an embodiment of the present invention.

[0017] Figure 7 This is an example process of a UE server performing AI-ML data collection according to an embodiment of the present invention. DETAILED DESCRIPTION

[0018] Reference will now be made in detail to some embodiments of the present invention, examples of which are illustrated in the accompanying drawings.

[0019] Several aspects of telecommunication systems will now be presented with reference to various apparatus and methods. These apparatus and methods will be described in the following detailed description and illustrated in the accompanying drawings by various blocks, components, circuits, processes, algorithms, etc. (collectively referred to as "elements"). These elements may be implemented using electronic hardware, computer software, or any combination thereof. Whether these elements are implemented as hardware or software depends on the specific application and the design constraints imposed on the overall system.

[0020] Figure 11 is a system diagram of an exemplary wireless network supporting data collection according to an embodiment of the present invention. The wireless communication network 100 includes one or more fixed infrastructure units forming a network distributed over a geographical area. These basic units may also be referred to as access points, access terminals, base stations, node Bs, evolved node Bs (eNBs), next generation node Bs (gNBs), or other terms used in the art. For example, a base station serves multiple mobile stations within a service area (e.g., a cell or a cell sector). In some systems, one or more base stations are coupled to a controller to form an access network, wherein the access network is coupled to one or more core networks. gNB 102, gNB 107, and gNB 108 are base stations in a wireless network, and their service areas may or may not overlap. gNB 102 is connected to gNB 107 via an Xn interface 121. gNB 102 is connected to gNB 108 via an Xn interface 122. gNB 107 is connected to gNB 108 via an Xn interface 123. The core network (CN) entity 103 is connected to gNBs 102 and 107 via NG interfaces 125 and 126, respectively. The network entity CN 109 is connected to gNB 108 via NG connection 127. The exemplary CN 103 and CN 109 are connected to the AI ​​server 105 via the Internet 106. CN 103 and CN 109 include core components such as user plane function (UPF) and access and mobility management function (AMF). In one embodiment, the AI ​​server 105 is a UE server. According to some examples, the UE server can be inside a mobile network operator (MNO). According to other examples, the UE server can be outside the MNO and can also be referred to as an OTT server. In one example, the UE server is an OTT server on the UE side. In another example, the UE server is an OTT server on the network side or an OTT server on a neutral site. Please note that some of the following examples are described in the context of OTT servers, but they can also be applied to other types of UE servers.

[0021] Figure 1A general data collection framework for UE, RAN node / gNB, CN, and AI server (such as OTT server) is further described. CN 103 is the backbone of the wireless network and communicates with the UE server via the Internet 106. In one embodiment, CN 103 includes network nodes / entities / functions such as data collection application function (DCAF), CN, operations, administration, and maintenance (OAM), etc.

[0022] In one example 180, the UE performs data collection for the AI-ML model. Data collection is the process of collecting data for the purpose of AI / ML model training, data analysis and inference through a network node, a management entity, a UE-side server or a neutral server. Data collection allows data to reach the UE-side or a neutral site (i.e., a UE server) and collects enough data for model generalization. In step 181, the UE obtains a data collection configuration / trigger. In step 182, the UE collects AI-ML related data. In step 184, the UE establishes a data transmission channel for AI-ML related data transmission. In step 185, the UE transmits AI-ML related data through the established channel. In one embodiment, in step 183, the UE obtains assistance information and transmits the assistance information and AI-ML related data through the established channel.

[0023] Figure 1 A simplified block diagram of a RAN node / base station, UE server, and mobile device / UE supporting data collection is further shown. The gNB / RAN node has an antenna 156 that transmits and receives radio signals. An RF transceiver circuit 153 coupled to the antenna receives RF signals from the antenna 156, converts the RF signals to baseband signals, and sends the baseband signals to the processor 152. The RF transceiver 153 also converts the baseband signals received from the processor 152 to RF signals and sends them to the antenna 156. The processor 152 processes the received baseband signals and calls different functional modules to perform functional features in the gNB 107. The memory 151 stores program instructions and data 155 to control the operation of the base station / gNB. The base station / gNB also includes a set of control modules 157 for performing functional tasks of communicating with the mobile station. These control modules can be implemented by circuits, software, firmware, or a combination of the above.

[0024] Figure 1Also included is a simplified block diagram of a UE (e.g., UE 101). A UE may also be referred to as a mobile station, mobile terminal, mobile phone, smartphone, wearable device, IoT device, tablet, laptop, or other terms used in the art. The UE performs functions to perform data collection for AI-ML model training and includes a detection module, a configuration module, a collection module, and a transmission module. The UE interacts with the gNB over an air interface. The UE has an antenna 166 for sending and receiving radio signals. An RF transceiver circuit 163 coupled to the antenna receives an RF signal from the antenna 166, converts the RF signal to a baseband signal, and sends the baseband signal to the processor 162. The RF transceiver 163 also converts the baseband signal received from the processor 162 to an RF signal and sends it to the antenna 166. The processor 162 processes the received baseband signal and calls different functional modules to perform functional features in the UE 101. The memory 161 stores program instructions and data 165 to control the operation of the UE 101. Antenna 166 sends uplink transmissions and receives downlink transmissions from antenna 156 of the base station / gNB.

[0025] The UE also includes a set of control modules for performing functional tasks. These control modules can be implemented by circuits, software, firmware, or a combination thereof. The detection module 191 receives a data collection request initiated by a UE server, where the UE is connected to a RAN node of a wireless network. The configuration module 192 obtains a data collection configuration from the RAN node, where the data collection configuration configures AI-ML model-related parameters for the UE. The collection module 193 performs data collection according to the data collection configuration to collect AI-ML model-related data. The transmission module 194 sends the collected AI-ML model-related data to the UE server through a transmission channel.

[0026] Figure 1 Also included is a simplified block diagram of a UE server (e.g., UE server 105). The UE server has a network interface module 173 that transmits and receives signals / messages over the network. A processor 172 processes received messages and calls different functional modules to perform functions in the UE server. A memory 171 stores program instructions and data 175 to control the operation of the UE server. The UE server also includes a set of control modules 177 that perform functional tasks of communicating with a mobile station. These control modules can be implemented by circuits, software, firmware, or a combination thereof.

[0027] Figure 2is an example process of data collection to a UE server without RAN awareness according to an embodiment of the present invention. One or more UEs, such as UE 201a, 201b, and 201c, are connected to a RAN node / gNB 202. The gNB 202 is connected to a core network 203. The core network 203 includes network functions and / or entities, such as DCAF, OAM, and other network functions / entities. The core network 203 is connected to a UE server 205. In one embodiment, data flows from the UE 201a / 201b / 201c to the UE server 205. In one embodiment, data is transmitted from the UE to a network entity / function in the CN 203, which forwards the data to the UE server 205. In one embodiment, the data transmission channel is a user plane (UP) channel. In another embodiment, the data transmission channel from the UE to the network entity / function in the CN 203 and from the network entity / function in the CN 203 to the UE server 205 is through a control plane (CP) channel. In one embodiment 210, the collected AI-ML related data is non-standardized / standardized / proprietary data for various models, customized for internal and / or external conditions and design choices of the UE. In one embodiment, the design choices include UE resource limitations, radio environment and / or feature engineering. In one embodiment, the UE collects data related to the AI-ML model. The collected data includes one or more of the following elements: model input, model true value, data quality, and other information (e.g., AI-ML model related data for generalization of the AI-ML model). In one embodiment 220, the UE communicates with the RAN node / gNB via layer 1 (layer-1, L1) uplink control information (uplink control information, UCI), L2 MAC control element (control element, CE), radio resource control (radio resource control, RRC) message or a new radio bearer for AI to obtain data collection configuration and optional assistance information. In one embodiment 230, the assistance information is also transmitted to the UE server. The assistance information includes one or more of the following elements: use cases for AI-ML model related data, functions of AI-ML model related data, scenario information, location information, and RAN configuration information for AI-ML model related data. In one embodiment 250, data transmission related to AI-ML is authenticated. Authentication can be performed by a core network entity / function, OAM, or UE server. In one embodiment, a network node / entity / function (e.g., DCAF, CN, OAM, etc.) performs an authentication check and determines whether to allow data collection and transmission to the UE server.In one embodiment, the network node / entity / function (eg, DCAF, CN, OAM, etc.) also determines the type of data allowed to be collected and transmitted to the UE server.

[0028] Figure 3 It is an example process of data collection trigger from UE server to UE without involving network node according to an embodiment of the present invention. In one embodiment, the data collection trigger / request is from UE server to UE through application layer. UE 301 is connected to RAN node 302 and CN 303, and CN 303 is connected to UE server 304. In step 311, UE server 304 directly sends data collection request to UE 301. In one embodiment, the data collection request is sent through application layer message. In one embodiment, in step 321, UE 301 further sends data collection request to RAN node 302. In one embodiment, the data collection request is indicated from application layer to RRC layer, through which the data collection request is sent from UE to network. In one embodiment, the data collection request to RAN node can be L1 / L2 signaling. In one embodiment, in step 322, RAN node 302 provides data collection configuration / assistance information according to UE request. In one embodiment 320, data collection configuration and / or assistance information are transmitted by RAN node 302 through unicast or multicast. In one embodiment, RAN node provides data collection configuration / assistance information for each UE. In one embodiment, the RAN node provides data collection configuration / assistance information for a group of UEs. In step 331, a data transmission channel between the UE and the UE server is established. In one embodiment 330, the data transmission channel is a UP channel or a CP channel. In step 341, the UE 301 performs measurements according to the data collection configuration from the RAN node 302. In step 351, the UE 301 sends a data collection response to the UE server 304. In one embodiment, the UE 301 attaches the assistance information received from the RAN node 302 to the collected data. In step 361, the UE 301 transmits the collected data (attached assistance information) to the UE server 304 through the data transmission channel.

[0029] Figure 4is an example process of data collection triggering from a UE server to a CN according to an embodiment of the present invention. UE 401 is connected to a RAN node 402 and one or more CN entities 403, and the CN entity 403 is connected to a UE server 404. In one embodiment, the CN entity is a network node / entity / function (e.g., DCAF, CN, OAM, etc.). In step 411, the UE server 404 sends a data collection request to the CN entity 403. In step 431, the CN entity 403 further sends the data collection request to the UE 401. In step 412, the CN entity 403 responds to the data collection request and sends a data collection response to the UE server 404. In step 421, a data transmission channel between the CN entity 403 and the UE server 404 is established. In one embodiment, the data transmission channel is a UP channel.

[0030] In step 432, the CN entity 403 sends a data collection assistance information request to the RAN node 402. The RAN node 402 provides assistance information to the CN entity 403 in step 446. In one embodiment, the CN entity 403 performs an authentication check and determines whether data collection and data transmission to the UE server are allowed. In one embodiment, the CN entity 403 also determines the type of data allowed to be collected and transmitted to the UE server. In one embodiment, the RAN node 402 provides assistance information based on a service agreement between the RAN and the UE server. In step 433, the RAN node 402 sends a data collection configuration and a trigger to the UE 401. In step 441, the UE 401 performs data collection according to the data collection configuration. In step 445, the UE 401 sends a configuration completion message to the RAN node 402. In step 451, the UE 401 sends a data collection response to the CN entity 403. In step 461, the UE 401 transmits the collected data to the CN entity 403. In one embodiment, UE 401 interacts with CN entity 403 via non-access-stratum (NAS) signaling (CP channel). In one embodiment, UE 401 interacts with CN entity 403 via application layer (UP channel). In one embodiment, in step 462, CN entity 403 appends assistance information to the collected data received from UE (such as UE 401). In step 464, CN entity 403 transmits the collected data (with the assistance information appended) to UE server 404 through a data transmission channel after an authentication check.

[0031] Figure 51 is an example flow of data collection trigger from UE server to CN entity and from UE server to UE without RAN awareness according to an embodiment of the present invention. UE 501 is connected to RAN node 502 and CN entity 503, and CN entity 503 is connected to UE server 504. In step 511, a data collection trigger / request from UE server 504 is sent to CN entity 503. In step 512, CN entity 503 sends a data collection response to UE server 504. In one embodiment, in step 510, a data collection trigger / request is sent from UE server 505 to UE 501 through an application layer. In one embodiment, in step 521, a data transmission channel is established between UE 501 and UE server 505. In one embodiment, the data transmission channel is a UP channel. In step 531, CN entity 503 sends a data collection and assistance information request to RAN node 502. In step 546, in response, RAN node 502 provides assistance information to CN entity 503 for authentication check. In one embodiment, CN entity 503 performs an authentication check and determines whether data collection and data transmission to UE server are allowed. In one embodiment, the CN entity 503 also determines the types of data that are allowed to be collected and transmitted to the UE server 504 .

[0032] In step 532, the RAN node 502 sends the data collection configuration and trigger to the UE 501. In step 541, the UE 501 performs data collection according to the data collection configuration. In step 545, the UE 501 sends a configuration completion message to the RAN node 502. In one embodiment, in step 551, the CN 503 sends a data collection request and assistance information to the UE after an authentication check. In step 552, the UE 501 sends a data collection response to the CN entity 503. In one embodiment, the UE 501 attaches the assistance information to the collected data. In step 561, the UE 501 sends the collected data together with the assistance information to the UE server 505 through a data transmission channel.

[0033] Figure 6 This is an example process of a UE performing AI-ML data collection according to an embodiment of the present invention. In step 601, the UE receives a data collection request initiated by a UE server, where the UE is connected to a RAN node of a wireless network. In step 602, the UE obtains a data collection configuration from the RAN node, where the data collection configuration configures AI-ML model-related parameters for the UE. In step 603, the UE performs data collection according to the data collection configuration to collect AI-ML model-related data. In step 604, the UE sends the collected AI-ML model-related data to the UE server through a transmission channel.

[0034] Figure 7This is an example process of a UE server performing AI-ML data collection according to an embodiment of the present invention. In step 701, the UE server initiates a data collection request with a destination of one or more UEs, wherein the data collection request initiates data collection of one or more UEs for an AI-ML model used by one or more UEs. In step 702, the UE server receives one or more data collection responses. In step 703, the UE server receives AI-ML model related data and assistance information through a data channel connected to a wireless network.

[0035] Although the present invention has been disclosed above with respect to the preferred embodiments, it is not intended to limit the present invention. Various changes, modifications and combinations may be made to the features in the various embodiments without departing from the protection scope of the present invention defined by the claims.

Claims

1. A method for performing AI-ML data collection, comprising: Receiving, by a user equipment UE, a data collection request initiated by a UE server, wherein the UE is connected to a radio access network RAN ​​node of a wireless network; Obtaining a data collection configuration from the RAN node, wherein the data collection configuration configures artificial intelligence-machine learning AI-ML model related parameters for the UE; Performing data collection according to the data collection configuration to collect AI-ML model related data; as well as The collected AI-ML model related data is sent to the UE server through the transmission channel.

2. The method for performing AI-ML data collection as described in claim 1, wherein the UE receives the data collection request directly from the UE server through an application layer message, and wherein the transmission channel is established between the UE and the UE server.

3. The method of performing AI-ML data collection as claimed in claim 2, further comprising: Sending a UE data collection request to the RAN node; Data collection configuration and assistance information is received from the RAN node.

4. The method for performing AI-ML data collection as described in claim 3, wherein the assistance information includes one or more of the following elements: a use case for the AI-ML model related data, a function of the AI-ML model related data, scenario information, location information, and RAN configuration information of the AI-ML model related data.

5. The method for performing AI-ML data collection as claimed in claim 3, wherein the assistance information is received from the RAN node via unicast or multicast.

6. The method for performing AI-ML data collection as described in claim 3, wherein the UE sends the assistance information together with the collected AI-ML model related data to the UE server.

7. The method for performing AI-ML data collection as described in claim 2, wherein the transmission channel established between the UE and the UE server is a user plane UP channel or a control plane CP channel.

8. The method for performing AI-ML data collection as claimed in claim 1, wherein the UE receives the data collection request from a network entity of the wireless network.

9. The method for performing AI-ML data collection as described in claim 8, wherein the UE receives a data collection configuration and a data collection trigger from the RAN node, and wherein the transmission channel includes a data path established between the UE and the network entity and a channel between the network entity and the UE server.

10. The method for performing AI-ML data collection as claimed in claim 8, wherein the data path between the UE and the network entity is a CP channel through a non-access stratum (NAS) layer or a UP channel through an application layer.

11. The method for performing AI-ML data collection as claimed in claim 1, wherein the UE receives the data collection request and assistance information from a network entity of the wireless network.

12. The method for performing AI-ML data collection as claimed in claim 11, wherein the UE receives the data collection configuration and the data collection trigger from the RAN node, and wherein the transmission channel established between the UE and the UE server is a UP channel or a CP channel.

13. The method for performing AI-ML data collection as described in claim 11, wherein the assistance information includes one or more of the following elements: use cases for the AI-ML model related data, functions of the AI-ML model related data, scenario information, location information, and RAN configuration information of the AI-ML model related data.

14. The method of performing AI-ML data collection as claimed in claim 1, wherein the UE communicates with the RAN node via layer 1 uplink control information, layer 2 MAC control element, radio resource control message or new radio bearer for AI.

15. The method for performing AI-ML data collection as claimed in claim 1, wherein the data collection configuration is received from the RAN node via unicast or multicast.

16. A method of performing AI-ML data collection, comprising: Initiating, by the UE server, a data collection request destined for one or more UEs, wherein the data collection request initiates data collection by the one or more UEs for an AI-ML model used by the one or more UEs; receiving one or more data collection responses; and Receive AI-ML model related data and assistance information through a data channel connected to the wireless network.

17. The method for performing AI-ML data collection as described in claim 16, wherein the data collection request is sent directly to the one or more UEs or to a network entity of the wireless network via an application layer message.

18. The method for performing AI-ML data collection as claimed in claim 16, wherein the assistance information is attached to the AI-ML model related data by the corresponding UE or the network entity of the wireless network.

19. The method for performing AI-ML data collection as described in claim 16, wherein the data channel is a UP channel directly between the corresponding UE, a CP channel directly between the corresponding UE, a UP channel between the network entity and the data path to the corresponding UE, or a CP channel between the network entity and the data path to the corresponding UE.

20. A user equipment, comprising: transceivers, used to transmit and receive radio frequency signals in wireless networks; a detection module, configured to receive a data collection request initiated by a UE server, wherein the UE is connected to a RAN node of the wireless network; A configuration module, configured to obtain a data collection configuration from the RAN node, wherein the data collection configuration configures AI-ML model related parameters for the UE; A collection module, used to perform data collection according to the data collection configuration to collect AI-ML model related data; as well as The transmission module is used to send the collected AI-ML model related data to the UE server through the transmission channel.

21. A storage medium storing a program, which, when executed, enables a user device to perform the steps of the method for performing AI-ML data collection according to any one of claims 1 to 15.

Images