A system and method for predicting the perceived quality of user devices using an artificial intelligence model.

Abstract

A system and method are presented for predicting the perceived quality of a user's equipment (UE) using an artificial intelligence (AI) model. A first network node of a radio access network (RAN) may receive first support information from a second network node of the RAN for use with first perceived quality (QoE) information to perform a first function of a neural network model. The first network node of the RAN may then perform the first function using the first QoE information and the first support information.

Description

Technical Field

[0001] The present disclosure generally relates to wireless communication including, but not limited to, systems and methods for predicting the perceived quality of experience (QoE) of a user equipment (UE) using an artificial intelligence (AI) model.

Background Art

[0002] Background The 3rd Generation Partnership Project (3GPP®), a standardization body, is currently promoting the specification of a new radio interface called 5G New Radio (5G NR), as well as a next-generation packet core network (NG-CN or NGC). 5G NR has three main components: a 5G access network (5G-AN), a 5G core network (5GC), and a user equipment (UE). To facilitate the enabling of different data services and requirements, the elements of the 5GC, also called network functions, are simplified, some of which are software-based and some are hardware-based, whereby they can be adapted as needed.

Summary of the Invention

Means for Solving the Problems

[0003] Summary The exemplary embodiments disclosed herein are directed to solving problems related to one or more of the problems presented in the prior art and providing further features that will become readily apparent by reference to the following detailed description in conjunction with the accompanying drawings. In accordance with various embodiments, exemplary systems, methods, devices, and computer program products are disclosed herein. However, it is understood that these embodiments are presented by way of example and not limitation, and that various modifications to the disclosed embodiments can be made while staying within the scope of this disclosure, as will be apparent to those skilled in the art upon reading this disclosure.

[0004] At least one aspect relates to the following system, method, apparatus, or computer-readable medium: A first network node of a radio access network (RAN) (e.g., a gNB, or a distributed unit (DU), or a centralized unit (CU)) may receive first support information from a second network node of the RAN (e.g., a gNB, or a distributed unit (DU), or a centralized unit (CU)) for use together with first quality-of-emotion (QoE) information (e.g., from one or more UEs) to perform a first function of a neural network model (e.g., model inference or model training, or other model processes / functions / steps). The first network node of the RAN may perform the first function using the first QoE information and the first support information. Input information collected from neighboring gNBs for an artificial intelligence (AI) function may influence the Xn Application Protocol (XnAP).

[0005] In some embodiments, a first network node may receive first quality-of-experience (QoE) information obtained from measurements in a wireless communication device (e.g., user equipment (UE)). The first network node may receive second QoE information obtained from measurements (e.g., monitoring, detection) in the wireless communication device and / or another wireless communication device for use in performing a second function of a neural network model (e.g., model training or model inference, or other model / AI process / function / step). Input QoE information in the AI ​​function can be collected from the UE side, thereby preventing potential influence of the Uu interface on QoE measurements. The first or second QoE information may include an indication of at least one of the following: a QoE reporting container, at least one protocol data unit (PDU) session identifier (ID), at least one QoS flow ID, at least one data radio bearer (DRB) ID, a slice list, or a list of information that may be included in the QoE information for use with the AI ​​function. Currently, QoE measurement information is not used for AI functions.

[0006] In some embodiments, a first network node may receive first user equipment (UE) support information provided by a wireless communication device or another wireless communication device for use in performing a second function of the neural network model (e.g., model inference or model training, or other model / AI process / function / step). The first network node may receive second UE support information provided by the wireless communication device for use in performing a first function of the neural network model. Input information may be collected from the UE side as support information for use in AI model training or inference, separate from QoE information, and may include information covering the impact on the Uu. The first or second UE support information may include at least one indication of service type, application, slice information, radio bearer information, cell information, beam information, binding information or group identifier (ID), UE location information, UE history information, radio link quality-related information, UE measurements regarding reference signal received power (RSRP), reference signal received quality (RSRQ), or signal-to-interference noise ratio (SINR) for the serving cell and at least one neighboring cell, drive test minimization (MDT) measurements, or UE performance information. A list of information may be included in the UE support for the use of AI functionality. Currently, there is no support information collected from the UE for use in predicting the UE's QoE.

[0007] In some embodiments, a first or third network node (e.g., a core network (CN) or an operations, management, and maintenance (OAM)) may send a message to the second network node requesting second support information for use in performing a second function of the neural network model (e.g., model training). The message may include at least one of the following: an indication of what information is requested from the second network node to the first network node; a request to the second network node to provide feedback on the neural network model; or a request to the second network node to provide data collected to evaluate the performance of a trained version of the neural network model. The collected data may include the following items: UE identifier, QoE measurement results collected at NG-RAN node 2, RAN visible QoE results collected at NG-RAN node 2, or UE mobility information. The first or third network node may receive the second support information from the second network node. The model training module may be deployed in the OAM or the core network (CN). In the case of core networks, the NG Application Protocol (NGAP) may have an impact (for example, if a CN requests support information from a gNB, the gNB may report the support information to the CN).

[0008] In some embodiments, the first or second support information may include at least one user equipment (UE) identifier (ID) of a UE associated with the second network node, binding information or group ID of multiple UEs associated with the second network node, predicted or historical trajectories of one or more of the UEs, mobility information or UE history information (UHI) of one or more of the UEs, at least one previous QoE measurement result of one or more of the UEs, measured or predicted QoE information of one or more of the UEs (e.g., if gNB2 also has model inference capabilities), transmission delay, cell list, or resource status indication. The first or second support information may be a list of items in the support information between network nodes, particularly when model training is deployed on OAM or CN.

[0009] In some embodiments, the first network node may send a request message to the second network node requesting first support information. The request message may include at least one of the following: an indication of what information the second network node is requesting from the first network node; a request to the second network node to provide feedback on the neural network model; or a request to the second network node to provide data collected to evaluate the performance of a trained version of the neural network model.

[0010] In some embodiments, the first network node may transmit information predicted or inferred according to a second function (e.g., model inference) to at least one other network node (e.g., a neighboring gNB) via the F1AP interface, the information including at least one identifier of at least one user equipment (UE) that has provided measurement results or has not provided any measurement results, predicted QoE information for at least one UE, predicted trajectory information for at least one UE, predicted or updated grouping of at least one UE, correlation coefficients showing the correlation between one UE and another UE, time information in the validity period, action to be taken, mobility information or UE history information (UHI) of at least one UE, at least one previous QoE measurement result of at least one of the at least one UE, a modified QoE measurement configuration as seen in the RAN, updated grouping information for at least one UE, a proposed or predicted QoE configuration as seen in the RAN, or at least one indication to deactivate or suspend reporting of the QoE configuration as seen in the RAN. The output of the model inference module can be forwarded to neighboring nodes as a basis for further action. In such cases, XnAP may have an impact.

[0011] In some embodiments, the first function may include model inference using a trained version of the neural network model. The second function may include model training of the neural network model.

[0012] In some embodiments, the first network node may include a first base station, a base station central unit (CU), or a base station distributed unit (DU). The second network node may include a second base station, a base station DU, or a base station CU. The third network node may include a core network (CN) node, or an operations, management, and maintenance (OAM) node.

[0013] In some embodiments, a first or second network node may receive a QoE configuration from an Operations, Management, and Maintenance (OAM) node or an Access and Mobility Management function. The first or second network node may transmit the QoE configuration to a wireless communication device. The QoE configuration may include at least one of the following: an indicator for using a first or second function of a neural network model, or binding information. The binding information may include at least one of the following: group information of multiple user equipment (UEs), service types of multiple UEs, applications or application types of multiple UEs, protocol data unit (PDU) session identifiers (IDs) of multiple UEs, quality of service (QoS) flow IDs of multiple UEs, radio bearer information of multiple UEs, location information of multiple UEs, or QoE user consent of multiple UEs.

[0014] In some embodiments, a first network node may send a request signaling to a second network node, the request signaling including at least one of the following: a flag used to request the second network node to provide feedback for neural network model inference; an indication of what information is being requested from the second network node to the first network node; or a flag used to request the second network node to provide collected real data for the first network node to evaluate the feedback for neural network model inference. Feedback can be a critical feature in AI, thereby enabling the node performing model inference to evaluate the performance of the AI ​​model.

[0015] In some embodiments, a first network node may receive feedback from a second network node. The feedback may be calculated based on predictive information provided by the first network node and actual data collected at the second network node. The feedback may include an indication of at least one of the following: the accuracy of neural network model inference, the confidence in the predictions of the neural network model (e.g., neural network model inference) for actual measured or collected data, a generalized value for evaluating the performance of neural network model inference, or a correlation of QoE between at least two user devices (UEs).

[0016] In some embodiments, the first network node may receive data collected by the second network node from the second network node. The data received from the second network node may cover the influence of XnAP on the feedback procedure.

[0017] In some embodiments, a second network node of a radio access network (RAN) (e.g., a gNB, or a distributed unit (DU), or a centralized unit (CU)) may transmit first support information to a first network node of the RAN (e.g., a gNB, or a distributed unit (DU), or a centralized unit (CU)) for use together with first quality-of-effect (QoE) information to perform a first function of a neural network model (e.g., model inference). The first QoE information can be obtained from measurements in a wireless communication device (e.g., a user equipment (UE)) and can be received by the first network node. [Brief explanation of the drawing]

[0018] Hereinafter, examples of various embodiments of the present solution will be described in detail with reference to the following figures or drawings. The drawings are provided for illustrative purposes only and merely depict exemplary embodiments of the present solution to facilitate the reader's understanding of the present solution. Therefore, the drawings should not be regarded as limiting the scope, extent, or applicability of the present solution. It should be noted that these drawings are not necessarily drawn to scale for clarity and ease of illustration.

[0019] [Figure 1] An example of a cellular communication network in which the techniques disclosed herein can be implemented according to an embodiment of the present disclosure is illustrated.

[0020] [Figure 2] A block diagram of an example of a base station and a user equipment device according to some embodiments of the present disclosure is illustrated.

[0021] [Figure 3] An example of a functional framework for a radio access network (RAN) incorporating artificial intelligence according to some embodiments of the present disclosure is illustrated.

[0022] [Figure 4] A sequence diagram for predicting the perceived quality of a user equipment (UE) using an artificial intelligence (AI) model according to some embodiments of the present disclosure is illustrated.

[0023] [Figure 5] A sequence diagram for predicting the perceived quality of a user equipment (UE) using an artificial intelligence (AI) model according to some embodiments of the present disclosure is illustrated.

[0024] [Figure 6] A sequence diagram for predicting the perceived quality of a user equipment (UE) using an artificial intelligence (AI) model according to some embodiments of the present disclosure is illustrated.

[0025] [Figure 7] The present disclosure illustrates a sequence diagram for predicting the perceived quality of a user's equipment (UE) using an artificial intelligence (AI) model, according to some embodiments of this disclosure.

[0026] [Figure 8] The present disclosure illustrates a sequence diagram for predicting the perceived quality of a user's equipment (UE) using an artificial intelligence (AI) model, according to some embodiments of this disclosure.

[0027] [Figure 9] The present disclosure illustrates a sequence diagram for predicting the perceived quality of a user's equipment (UE) using an artificial intelligence (AI) model, according to some embodiments of this disclosure.

[0028] [Figure 10] This disclosure illustrates a flowchart for predicting the perceived quality of a user's equipment (UE) using an artificial intelligence (AI) model, according to one embodiment of this disclosure. [Modes for carrying out the invention]

[0029] Detailed explanation 1. Mobile communication technologies and environment Figure 1 illustrates an example of a wireless communication network and / or system 100 in which the techniques disclosed herein may be implemented, according to one embodiment of the present disclosure. In the following discussion, the wireless communication network 100 may be any wireless network, such as a cellular network or a narrowband Internet of Things (NB-IoT) network, and will be referred to herein as “Network 100”. An example of such Network 100 is a cluster of base stations 102 (hereinafter referred to as BS102, also referred to as wireless communication nodes) and user equipment devices 104 (hereinafter referred to as UE104, also referred to as wireless communication devices) that can communicate with each other via a communication link 110 (e.g., a wireless communication channel), as well as clusters of cells 126, 130, 132, 134, 136, 138, and 140 that overlap with a geographical area 101. In Figure 1, BS102 and UE104 are contained within the respective geographical boundaries of cell 126. Each of the other cells 130, 132, 134, 136, 138, and 140 may include at least one base station operating within its allocated bandwidth to provide adequate radio coverage to its intended users.

[0030] For example, BS102 may operate within an allocated channel transmission bandwidth to provide adequate coverage to UE104. BS102 and UE104 may communicate via downlink radio frames 118 and uplink radio frames 124, respectively. Each radio frame 118 / 124 may be further divided into subframes 120 / 127, which may contain data symbols 122 / 128. In this disclosure, BS102 and UE104 are described herein as non-limiting examples of “communication nodes” capable of implementing the methods disclosed herein. Such communication nodes may be capable of performing wireless and / or wired communications according to various embodiments of this solution.

[0031] Figure 2 illustrates a block diagram of an example of a wireless communication system 200 for transmitting and receiving wireless communication signals (e.g., OFDM / OFDMA signals) according to several embodiments of the present solution. The system 200 may include components and elements configured to support known or conventional operating features that do not need to be described in detail herein. In one exemplary embodiment, the system 200 can be used to communicate (e.g., transmit and receive) data symbols in a wireless communication environment such as the wireless communication environment 100 in Figure 1, as described above.

[0032] System 200 generally includes a base station 202 (hereinafter, "BS202") and a user equipment device 204 (hereinafter, "UE204"). BS202 includes a BS (base station) transceiver module 210, a BS antenna 212, a BS processor module 214, a BS memory module 216, and a network communication module 218, each module being coupled and interconnected to one another as needed via a data communication bus 220. UE204 includes a UE (user equipment) transceiver module 230, a UE antenna 232, a UE memory module 234, and a UE processor module 236, each module being coupled and interconnected to one another as needed via a data communication bus 240. BS202 communicates with UE204 via a communication channel 250, which can be any wireless channel or other medium suitable for data transmission as described herein.

[0033] As will be understood by those skilled in the art, System 200 may further include any number of modules other than those shown in Figure 2. As will be apparent to those skilled in the art, various exemplary blocks, modules, circuits, and processing logic described in relation to the embodiments disclosed herein may be implemented in hardware, computer-readable software, firmware, or any practical combination thereof. To clearly illustrate this compatibility and compatibility of hardware, firmware, and software, various exemplary components, blocks, modules, circuits, and steps are generally described in relation to their functions. Whether such functions are implemented as hardware, firmware, or software may depend on the specific application and the design constraints imposed on the system as a whole. Those familiar with the concepts described herein may implement such functions in a manner suitable for a specific application, but such implementation decisions should not be construed as limiting the scope of this disclosure.

[0034] According to some embodiments, the UE transceiver 230 may be referred herein as an “uplink” transceiver 230, comprising a radio frequency (RF) transmitter and an RF receiver, each having a circuit coupled to the antenna 232. Alternatively, a duplex switch (not shown) may couple the uplink transmitter or receiver to the uplink antenna in a time-duplexed manner. Similarly, according to some embodiments, the BS transceiver 210 may be referred herein as a “downlink” transceiver 210, comprising an RF transmitter and an RF receiver, each having a circuit coupled to the antenna 212. Alternatively, a downlink duplex switch may couple the downlink transmitter or receiver to the downlink antenna 212 in a time-duplexed manner. The operation of the two transceiver modules 210 and 230 may be time-coordinated so that the downlink transmitter is coupled to the downlink antenna 212 and at the same time the uplink receiver circuit is coupled to the uplink antenna 232 for receiving transmissions over the wireless transmission link 250. Conversely, the operation of the two transceivers 210 and 230 may be time-coordinated so that the uplink transmitter is coupled to the uplink antenna 232 and at the same time the downlink receiver is coupled to the downlink antenna 212 to receive transmissions over the wireless transmission link 250. In some embodiments, strict time synchronization exists with a minimum guard time between changes in duplex direction.

[0035] The UE transceiver 230 and base station transceiver 210 are configured to communicate over a wireless data communication link 250 and to cooperate with a appropriately configured RF antenna structure 212 / 232 capable of supporting specific wireless communication protocols and modulation schemes. In some exemplary embodiments, the UE transceiver 210 and base station transceiver 210 are configured to support industry standards such as Long-Term Evolution (LTE) and emerging 5G standards. However, it should be understood that this disclosure is not necessarily limited to application to specific standards and associated protocols. Rather, the UE transceiver 230 and base station transceiver 210 may be configured to support alternative or additional wireless data communication protocols, including future standards or variations thereof.

[0036] According to various embodiments, BS202 may be, for example, an evolved node B (eNB), a service-providing eNB, a target eNB, a femtostation, or a picostation. In one embodiment, UE204 may be embodied in various types of user devices, such as mobile phones, smartphones, personal digital assistants (PDAs), tablets, laptop computers, and wearable computing devices. Processor modules 214 and 236 may be implemented or realized with general-purpose processors, content-addressable memory, digital signal processors, application-specific integrated circuits, field-programmable gate arrays, any suitable programmable logic devices, discrete gates or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. Thus, the processor may be realized as a microprocessor, controller, microcontroller, state machine, etc. The processor may also be implemented as a combination of computing devices, for example, a combination of a digital signal processor and a microprocessor, multiple microprocessors, one or more microprocessors working with a digital signal processor core, or any other such configuration.

[0037] Furthermore, steps of methods or algorithms described in relation to embodiments disclosed herein may be directly embodied in hardware, firmware, software modules executed by processor modules 214 and 236, respectively, or any practical combination thereof. Memory modules 216 and 234 may be implemented as RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disks, removable disks, CD-ROMs, or any other form of storage medium known in the art. In this regard, memory modules 216 and 234 may be coupled to processor modules 210 and 230, respectively, so that processor modules 210 and 230 can read information from and write information to memory modules 216 and 234, respectively. Memory modules 216 and 234 may also be incorporated into their respective processor modules 210 and 230. In some embodiments, memory modules 216 and 234 may each include cache memory for storing temporary variables or other intermediate information during the execution of instructions to be executed by processor modules 210 and 230, respectively. Furthermore, memory modules 216 and 234 may each include non-volatile memory for storing instructions to be executed by processor modules 210 and 230, respectively.

[0038] The network communication module 218 generally corresponds to the hardware, software, firmware, processing logic, and / or other components of the base station 202 that enable bidirectional communication between the base station transceiver 210 and other network components and communication nodes configured to communicate with the base station 202. For example, the network communication module 218 may be configured to support Internet or WiMAX traffic. In a typical deployment, but not limited to, the network communication module 218 provides an 802.3 Ethernet® interface so that the base station transceiver 210 can communicate with a conventional Ethernet®-based computer network. In this way, the network communication module 218 may include a physical interface for connecting to a computer network (e.g., a mobile communications switch (MSC)). The terms “configured for,” “configured,” and their inflections as used herein in relation to a specified operation or function refer to a device, component, circuit, structure, machine, signal, etc., that is physically built, programmed, formatted, and / or arranged to perform a specified operation or function.

[0039] The Open System Interconnection (OSI) model (hereinafter referred to as the “Open System Interconnection Model”) is a conceptual and logical layout that defines network communications used by systems (e.g., wireless communication devices, wireless communication nodes) that are open to interconnection and communication with other systems. The model is divided into seven subcomponents or layers, each of which corresponds to a conceptual set of services provided to its upper and lower layers. The OSI model also defines a logical network and effectively describes computer packet forwarding by using different layer protocols. The OSI model is sometimes referred to as the 7-layer OSI model or 7-layer model. In some embodiments, the first layer may be the physical layer. In some embodiments, the second layer may be the medium access control (MAC) layer. In some embodiments, the third layer may be the radio link control (RLC) layer. In some embodiments, the fourth layer may be the packet data convergence protocol (PDCP) layer. In some embodiments, the fifth layer may be the radio resource control (RRC) layer. In some embodiments, the sixth layer may be the Non-Accessable Layer (NAS) layer or the Internet Protocol (IP) layer, and the seventh layer may be any other layer.

[0040] To enable those skilled in the art to fabricate and use the present solution, various exemplary embodiments of the present solution are described below with reference to the accompanying drawings. As will be apparent to those skilled in the art, after reading this disclosure, various changes or modifications can be made to the examples described herein without departing from the scope of the present solution. Thus, the present solution is not limited to the exemplary embodiments and uses described and illustrated herein. Furthermore, the particular order or hierarchy of steps in the methods disclosed herein is merely illustrative. Based on design preferences, the particular order or hierarchy of steps in the disclosed methods or processes can be rearranged while remaining within the scope of the present solution. Thus, those skilled in the art will understand that the methods and techniques disclosed herein present various steps or operations in a sample order, and the present solution is not limited to the specific order or hierarchy presented unless otherwise specified.

[0041] 2. System and method for predicting the perceived quality of experience (QoE) of user devices (UE) using artificial intelligence (AI) models. Artificial intelligence (AI) capabilities can be used for data prediction based on collected real-world data, and for training / inference with models. However, AI / machine learning (ML) training / inference for perceived quality of experience (QoE) is not supported. This disclosure uses AI / ML / model functions to predict future QoE outcomes for a UE (e.g., including at least one UE and other UEs) based on collected QoE outcomes for at least one UE.

[0042] Figure 3 illustrates an example of a functional framework for radio access network (RAN) intelligence according to some embodiments of this disclosure. AI functionality can be deployed on RAN nodes for RAN intelligence.

[0043] Data collection: A function that can provide input data for model training and inference functions.

[0044] Model Training: Capabilities that enable AI / ML model training, validation, and testing, which can generate model performance metrics as part of the model testing procedure.

[0045] Model inference: A function that can provide AI / ML model inference output (e.g., prediction or decision).

[0046] QoE: Can support QoE measurement acquisition (QMC) functionality. New radio (NR) QMC functionality can be activated by the OAM via a separate QMC framework. For signaling-based QoE, the OAM can send a QMC configuration for a specific UE to the core network (CN), which may then send the QMC configuration to the RAN node via UE-related signaling. For management-based QoE, the OAM may send the QMC configuration to the RAN node, which may select UEs that meet the conditions for QoE measurement and send the configuration to the UE.

[0047] In a standalone architecture, for QoE reporting, the UE APP / application layer may collect QoE metrics and send the collected data to the UE Access Layer (AS) layer via AT commands. The UE AS layer may send the QoE report to a RAN node, which may then forward the received QoE report to a Measurement Collection Entity (MCE). The MCE can be an entity that collects QoE measurement reports and performs analysis for optimization.

[0048] RAN visible QoE measurement collection can be configured by a gNB, which can be configured when there are ongoing QoE measurements at the application layer. RAN visible QoE measurement results can be sent to the gNB as explicit information elements (IEs) readable by the gNB. Buffer levels and playback delays for DASH streaming and VR services can be supported as RAN visible QoE metrics. In addition to RAN visible QoE metrics, PDU session IDs can be sent to the RAN node along with the RAN visible QoE measurement results.

[0049] In a RAN overload situation, a RAN node may send a pause indication to the UE to notify the UE to pause QoE reporting. After the RAN overload situation is resolved, the RAN node may send a resume indication to the UE to request that reporting be resumed. During a mobility event (e.g., a transition between RAN nodes), the pause state of the source RAN node can be passed to the target node to indicate that QoE reporting was paused at the UE. RAN visible QoE reporting may not be affected by RAN overload, meaning that RAN visible QoE reporting will not be paused when reporting from a QoE reporting container is paused.

[0050] Implementation Example 1: Predicting QoE results for a set of UEs, both model training and inference performed on the RAN side. Figure 4 illustrates a sequence diagram for predicting the perceived quality of a user device (UE) using an artificial intelligence (AI) model, according to some embodiments of this disclosure. In this example implementation, both the model training and model inference modules can be deployed on a RAN node.

[0051] Step 0: QoE measurement data collection can be activated. Details are provided in Implementation Example 3.

[0052] Step 1: The UE may send QoE measurement results to the gNB1, which may include at least one of the following items: QoE reporting container, at least one protocol data unit (PDU) session identifier (ID), at least one QoS flow ID, at least one data radio bearer (DRB) ID, slice list, or RAN-visible QoE measurement results (e.g., buffer level, playback delay for media boot, or generalized RAN-visible QoE value). In existing implementations, QoE measurement information is not used for AI functions.

[0053] Step 1a: In addition to the QoE measurement results reported to the gNB, the UE may also report several other pieces of information that can assist the RAN node for AI model training / inference. Supporting information from the UE may include, at least one of the following: service type, application, slice information, radio bearer information, cell information, beam information, binding information or group identifier (ID) (e.g., UEs sharing the same group ID are interpreted as belonging to the same group), UE location information, UE history information, radio link quality-related information, UE measurements related to reference signal received power (RSRP), reference signal received quality (RSRQ), or signal-to-interference noise ratio (SINR) for the serving cell and at least one neighboring cell, drive test minimization (MDT) measurements, or UE performance information. A list of this information may be included in the UE assistance for use with AI functionality. Currently, there is no supporting information collected from the UE for use in predicting the UE's QoE.

[0054] Step 2: gNB1 may send a request message to gNB2 to request support information for model training. The information in the request message may include at least one of the following items, namely an indication of what information is requested to be provided from NG-RAN node 2 to NG-RAN node 1.

[0055] Step 3: gNB2 may transmit support information via XnAP for gNB1 to perform model training. Support information from other gNBs may include at least one of the following items: at least one user equipment (UE) identifier (ID) of a UE associated with the second network node, binding information or group ID of multiple UEs associated with the second network node, predicted or historical trajectory of one or more of the UEs, mobility information or UE history information (UHI) of one or more of the UEs, at least one previous QoE measurement result of one or more of the UEs, measured or predicted QoE information of one or more of the UEs (e.g., buffer level, playback delay for media boot, throughput, frame rate, round-trip time, or generalized RAN visible QoE value), transmission delay, cell list, or resource status. The first or second support information may be a list of items for support information between network nodes, in particular when model training is deployed in OAM or CN. The supporting information may cover one or more UEs, and if there are multiple UEs, the information may represent the average, minimum, or maximum among the multiple UEs.

[0056] Step 4: gNB1 may train a model based on QoE measurements and supporting information received from the UE and other neighboring gNBs in order to predict the QoE results of the UE. The UE may include UEs that provide QoE measurements and other UEs that do not provide QoE measurements.

[0057] Steps 5 and 5a: The UE may send the QoE measurement results and UE support information to the gNB1, as in steps 1 and 1a.

[0058] Step 6: gNB1 may send a request message to gNB2 to request support information for model inference. The information in the request message may include at least one of the following items: an indication of what information is requested from NG-RAN node 2 to NG-RAN node 1, a request to gNB2 to provide feedback on the neural network model, or a request to gNB2 to provide data collected to evaluate the performance of a trained version of the neural network model.

[0059] Step 7: gNB2 may send support information to gNB1 for model inference, and the information via XnAP may include at least one of the following items: at least one user device (UE) identifier (ID) of a UE associated with gNB2, binding information or group ID of multiple UEs associated with gNB2, predicted or historical trajectory of one or more UEs, mobility information or UE history information (UHI) of one or more UEs, at least one previous QoE measurement result of one or more UEs, or predicted QoE information of one or more UEs (for example, if gNB2 also has model inference capabilities).

[0060] Step 8: gNB1 may perform model inference to predict the QoE results for a group of UEs or the RAN-visible QoE results. Apart from the QoE or RAN-visible QoE results, the prediction information may also include at least one of the following items: several predicted trajectories for UEs, UE mobility information, or updated group information for UEs. gNB1 may also take some action based on the results of the model inference (e.g., making a decision regarding handover preparation and updating the UE binding information). In the case of model inference, the model inference function can predict the QoE results for UEs regardless of whether the UEs provided QoE measures or other measures for model training / inference.

[0061] Step 9 (Optional): gNB1 may transmit predictive information to a neighboring gNB (e.g., gNB2) via XnAP. The predictive information may include at least the following items: at least one identifier of at least one user device (UE) that has provided measurement results or has not provided any measurement results; predicted QoE information for at least one UE (e.g., buffer level, playback delay for media boot, throughput, frame rate, round-trip time, or generalized RAN visible QoE value); predicted trajectory information for at least one UE; predicted or updated grouping of at least one UE; correlation coefficients showing the correlation between one UE and another UE; time information regarding validity (e.g., start / end time, duration); or one of the actions to be taken based on model inference.

[0062] Step 10: gNB2 can send feedback to gNB1, which is specifically explained in Implementation Example 3.

[0063] Implementation Example 2: Predicting QoE results for a set of UEs, model training deployed on OAM / CN, and model inference deployed on the RAN side. Figure 5 illustrates a sequence diagram for predicting the perceived quality of a user device (UE) using an artificial intelligence (AI) model, according to some embodiments of this disclosure. In this example implementation, the model training module can be deployed on the OAM or core network (CN), and the model inference module is deployed on the RAN node.

[0064] Step 0: QoE measurement data collection can be activated. Details are provided in Implementation Example 3.

[0065] Step 1: The UE may send QoE measurement results to the gNB1, which may include at least one of the following items: QoE reporting container, at least one protocol data unit (PDU) session identifier (ID), at least one QoS flow ID, at least one data radio bearer (DRB) ID, slice list, or RAN-visible QoE measurement results (e.g., buffer level, playback delay for media boot, or generalized RAN-visible QoE value). Currently, QoE measurement information is not used for AI functionality.

[0066] Step 1a: In addition to the QoE measurement results reported to the gNB, the UE may also report several other pieces of information that can assist the RAN node for AI model training / inference. Supporting information from the UE may include, at least one of the following: service type, application, slice information, radio bearer information, cell information, beam information, binding information or group identifier (ID) (e.g., UEs sharing the same group ID are interpreted as being in the same group), UE location information, UE history information, radio link quality-related information, UE measurements related to reference signal received power (RSRP), reference signal received quality (RSRQ), or signal-to-interference noise ratio (SINR) for the serving cell and at least one neighboring cell, drive test minimization (MDT) measurements, or UE performance information. A list of this information may be included in the UE assistance for the use of AI functionality. Currently, there is no supporting information collected from the UE for use in predicting the UE's QoE. UEs may include UEs that provide QoE measurement results and other UEs that do not.

[0067] Step 2: OAM / CN may send a request message to gNB1 to request support information for model training. The information in the request message may include at least one of the following items: an indication of what information is requested to be provided from NG-RAN node 2 to NG-RAN node 1; an indication requesting gNB1 to provide feedback on the AI / ML model; or an indication requesting gNB1 to provide data collected to evaluate the performance of the trained model.

[0068] Step 3: gNB1 may send support information to the OAM or CN via NGAP for the OAM / CN to train the model. Support information from other gNBs may include at least one of the following items: at least one user equipment (UE) identifier (ID), binding information or group ID of multiple UEs, predicted or historical trajectory of one or more UEs, mobility information or UE history information (UHI) of one or more UEs, at least one previous QoE measurement result of one or more UEs (e.g., UE ID), or measured or predicted QoE information of one or more UEs (e.g., buffer level, playback delay for media boot, throughput, frame rate, round-trip time, or generalized RAN visible QoE value). The support information may apply to one or more UEs, and if there are multiple UEs, the information may be the average, minimum, or maximum among the multiple UEs.

[0069] Step 4: The OAM / CN may train a model based on QoE measurements and supporting information received from the UE and other neighboring gNBs in order to predict the QoE results of the UE. The UE may include UEs that provide QoE measurements and other UEs that do not provide QoE measurements.

[0070] Step 5: After model training, OAM / CN may send the AI ​​model to gNB1, which can then be deployed / updated within gNB1.

[0071] Step 5a: The UE may send the QoE measurement results and supporting information to the gNB1, as in Step 1 and Step 1a.

[0072] Step 6: gNB1 may send a request message to gNB2 to ask for support information for model inference.

[0073] Step 7: gNB2 may send support information to gNB1 for model inference. The information via XnAP may include at least one of the following items: UE ID, UE group information, UE predicted trajectory information, UE mobility information / UHI, UE previous QoE measurement results, or several UE predicted QoE (e.g., if gNB2 also has model inference capabilities, buffer level, playback delay for media startup, throughput, frame rate, round-trip time, or generalized RAN visible QoE value).

[0074] Step 8: gNB1 may perform model inference to predict the QoE results for a group of UEs or the RAN-visible QoE results. Apart from the QoE or RAN-visible QoE results, the prediction information may include at least one of the following items: predicted trajectory information for several UEs, mobility information for UEs, or updated group information for UEs. gNB1 may also take some action based on the results of the model inference (e.g., making a decision regarding handover preparation and updating the binding information for UEs). In the case of model inference, the model inference function can predict the QoE results for UEs regardless of whether the UEs provided QoE measures or other measures for model training / inference.

[0075] Step 9 (Optional): gNB1 may transmit predictive information to a neighboring gNB (e.g., gNB2) via XnAP. The predictive information may include at least one identifier of at least one user device (UE) that has provided measurement results or has not provided any measurement results; predicted QoE information for at least one UE (e.g., buffer level, playback delay for media boot, throughput, frame rate, round-trip time, or generalized RAN visible QoE value); predicted trajectory information for at least one UE; predicted or updated grouping of at least one UE; correlation coefficients showing the correlation between one UE and another UE; time information for validity (e.g., start / end time, duration); or one of the actions to be taken based on model inference.

[0076] Step 10: gNB2 may send feedback to gNB1, which is specifically described in Implementation Example 4.

[0077] Implementation Example 3: QoE Measurement Configuration for AI / ML Data Acquisition Figure 6 illustrates a sequence diagram for predicting the perceived quality of a user's equipment (UE) using an artificial intelligence (AI) model, according to some embodiments of this disclosure.

[0078] Step 1: The OAM / AMF may send the QoE measurement configuration to the gNB1. For management-based QoE, the configuration can be sent from the OAM to the gNB. For signaling-based QoE, the configuration can be sent from the OAM to the gNB via the Access and Mobility Management Function (AMF). For QoE measurement acquisition and AI functionality, supporting information may be sent to the RAN node along with the QoE configuration, which may include at least one of the following items: indicators or binding information for using the first or second function of the neural network model. Binding information may include at least one of the following: group information of multiple user devices (UEs) (e.g., an ID that identifies a group of UEs that may have correlated perceived quality); service types of multiple UEs (e.g., UEs running the same or similar service types can be marked as a group); applications or application types of multiple UEs (e.g., UEs running the same or similar applications can be marked as a group); protocol data unit (PDU) session identifiers (IDs) of multiple UEs; quality of service (QoS) flow IDs of multiple UEs; radio bearer information of multiple UEs (e.g., UEs using the same radio bearer can be marked as a group); location information of multiple UEs (e.g., UEs sharing similar location information can be marked as a group); or QoE user consent of multiple UEs.

[0079] Step 2: gNB1 may send a QoE measurement configuration to the UE to trigger the collection and reporting of application layer measurements in the UE. For managed QoE, when the RAN node makes a UE selection, gNB1 may select UEs that satisfy the binding information in the QoE measurement configuration for the purpose of predicting the UE's QoE. The UE may include UEs that provide QoE measurement results and other UEs that do not provide QoE measurement results.

[0080] Step 3: The UE may perform QoE measurements at the application layer and report them to gNB1.

[0081] Implementation Example 4: Feedback Procedure Figure 7 illustrates a sequence diagram for predicting the perceived quality of a user device (UE) using an artificial intelligence (AI) model, according to some embodiments of this disclosure. In some embodiments, a feedback procedure can be performed after model inference. In this example implementation, model inference may be deployed on NG-RAN node 1, but this does not preclude NG-RAN node 2 from deploying its own model training and inference capabilities.

[0082] Option 1

[0083] Step 1: NG-RAN node 1 may send a request message to NG-RAN node 2. The request message may include at least one of the following items: a flag used to request NG-RAN node 2 to provide feedback for model inference, or an indication of what information is requested to be provided from NG-RAN node 2 to NG-RAN node 1.

[0084] Step 2: NG-RAN node 2 may respond to NG-RAN node 1 with a calculated feedback based on the predictive information provided by NG-RAN node 1 and the actual data collected at NG-RAN node 2. The feedback from NG-RAN node 2 may include at least one of the following information: the accuracy of the model inference, the confidence of the prediction against the actual measured and collected data, a generalized value (e.g., good or bad) for evaluating the performance of the inference model, and the correlation of the QoE of different UEs.

[0085] Step 3: NG-RAN node 1 may evaluate the AI / ML model based on the feedback received from NG-RAN node 2.

[0086] Option 2

[0087] Step 1: NG-RAN node 1 may send a request message to NG-RAN node 2. The request message may include the following items, namely flags used by NG-RAN node 1 to request NG-RAN node 2 to provide the actual data collected for evaluation of the model inference feedback.

[0088] Step 2: NG-RAN node 2 may provide the collected data to NG-RAN node 1. The collected data may include the following items: UE identifier, QoE measurement results collected at NG-RAN node 2, RAN visible QoE results collected at NG-RAN node 2, or UE mobility information.

[0089] Step 3: NG-RAN node 1 may compute feedback for evaluating and updating the model based on the actual data received from NG-RAN node 2. The feedback may include at least one of the following: the accuracy of the model inference, the confidence of the predictions against the actual measured and collected data, a generalized value (e.g., good or bad) for evaluating the performance of the inference model, or the correlation of the QoE of different UEs.

[0090] Implementation Example 5: CU-DU Partitioning - Model Inference Expanded with CU Figure 8 illustrates a sequence diagram for predicting the perceived quality of a user's equipment (UE) using an artificial intelligence (AI) model, according to some embodiments of this disclosure.

[0091] Step 1: The UE may report the QoE measurement results to the gNB-CU, which may include at least one of the following items: QoE reporting container, PDU session ID, QoS flow ID, DRB ID, slice list, or RAN visible QoE measurement results (e.g., buffer level, playback delay for media boot).

[0092] Step 1a: The UE may also transmit some other supporting information to the gNB-CU to assist in QoE prediction within the gNB-CU. The information may include at least one of the following items: service type, application, slice information, radio bearer information, cell information, beam information, binding information (e.g., group ID), UE location information, UE history information, radio link quality-related information, UE measurements related to RSRP, RSRQ, SINR of the serving cell and adjacent cells, or MDT measurements.

[0093] Step 2: The gNB-CU may send a request message to the gNB-DU to request support information provided by the gNB-DU to assist in model training. The information in the request message may include at least one of the following items: information that the gNB-CU wants to obtain from the gNB-DU (e.g., data collected to assist in model training), an indication that requests the gNB-DU to provide feedback on the AI / ML model, or an indication that requests the gNB-DU to provide data collected to evaluate the trained model.

[0094] Step 3: The gNB-DU may send support information to the gNB-CU via the F1AP. The support information may include at least one of the following items: transmission delay, cell list, or resource status.

[0095] Step 4: gNB-CU may collect information from UE and gNB-DU and perform model training.

[0096] Step 5, 5a: UE may provide QoE measurement results and other measurements as supporting information, as in Step 1 and Step 1a.

[0097] Step 6: The gNB-CU may send a request message to the gNB-DU to request supporting information provided by the gNB-DU to assist in model inference. The information in the request message may include at least one of the following items: information that the gNB-CU wants to obtain from the gNB-DU (e.g., collected data to assist in model inference), an indication that requests the gNB-DU to provide feedback on the AI / ML model, or an indication that requests the gNB-DU to provide collected data to evaluate the trained model.

[0098] Step 7: The gNB-DU may send support information to the gNB-CU via the F1AP. The support information may include at least one of the following items: transmission delay, cell list, or resource usage.

[0099] Step 8: The gNB-CU may collect information from the UE and gNB-DU and perform model inference to predict the QoE or RAN-visible QoE results for a group of UEs. Apart from the QoE or RAN-visible QoE results, the prediction information may include at least one of the following items: predicted trajectory information for several UEs, mobility information / UHI for the UEs, or previous / predicted QoE measurements for the UEs. The gNB-CU may take some action based on the results of the model inference (e.g., modify the RAN-visible QoE configuration). In the case of model inference, the model inference function can predict the QoE results for the UEs regardless of whether the UEs provided QoE measurements or other measurements for model training / inference.

[0100] Step 9: The gNB-CU may send predicted information and actions to the gNB-DU via the F1AP to assist in network optimization. The information sent via the F1AP may include at least one of the following items: UE ID, predicted QoE results for the corresponding UE (e.g., buffer level, playback delay for media boot, throughput, frame rate, round-trip time, or generalized RAN visible QoE value), predicted trajectory information for the UE, mobility information / UHI for the UE, updated group information for the UE, previous QoE measurement results for the UE (e.g., buffer level, playback delay for media boot, throughput, frame rate, round-trip time, or generalized RAN visible QoE value), or modified RAN visible QoE configuration (e.g., reporting cycle, RAN visible QoE metric, threshold trigger, event trigger).

[0101] Step 10: The gNB-DU may provide feedback to the gNB-CU as requested by the gNB-CU in the previous step. The feedback message may include at least one of the following items: the accuracy of the model inference, the variance of the predictions against the actual measured and collected data, a generalized value for evaluating the performance of the inference model (e.g., good or bad), or actual collected data after the gNB-DU has received the prediction data from the gNB-CU.

[0102] Implementation Example 6: CU-DU Partition - Model Inference Expanded with DU Figure 9 illustrates a sequence diagram for predicting the perceived quality of a user device (UE) using an artificial intelligence (AI) model, according to some embodiments of this disclosure.

[0103] Step 1: The UE may report QoE measurement results to the gNB-CU, which may include at least one of the following items: QoE reporting container, PDU session ID, QoS flow ID, DRB ID, slice list, or RAN visible QoE measurement results (e.g., buffer level, playback delay for media boot, jitter, throughput).

[0104] Step 1a: The UE may also transmit some other supporting information to the gNB-CU to assist in QoE prediction within the gNB-CU. The information may include at least one of the following items: service type, application, slice information, radio bearer information, cell information, beam information, binding information (e.g., group ID), UE location information, UE history information, radio link quality-related information, UE measurements related to RSRP, RSRQ, SINR of the serving cell and adjacent cells, or MDT measurements.

[0105] Step 2: The gNB-CU may send a request message to the gNB-DU to request support information provided by the gNB-DU to assist in model training. The information in the request message may include at least one of the following items: information that the gNB-CU wants to obtain from the gNB-DU (e.g., data collected to assist in model training), an indication that requests the gNB-DU to provide feedback on the AI / ML model, or an indication that requests the gNB-DU to provide data collected to evaluate the trained model.

[0106] Step 3: The gNB-DU may provide support information to the gNB-DU via the F1AP. The support information may include at least one of the following items: transmission delay, cell list, or resource usage.

[0107] Step 4: gNB-CU may collect information from UE and gNB-DU and perform model training.

[0108] Step 4a: After model training, gNB-CU may send the AI ​​model to gNB-DU, which can then be deployed / updated on gNB-DU / gNB1.

[0109] Step 5, 5a: UE may provide QoE measurement results and other measurements as supporting information, as in Step 1 and Step 1a.

[0110] Step 6: The gNB-CU may transfer the QoE measurement results to the gNB-DU via the F1AP message.

[0111] Step 6a: The gNB-CU may forward the UE support information received in Step 5a to the gNB-DU via the F1AP.

[0112] Step 7: The gNB-DU may send a request message to the gNB-CU to request supporting information provided by the gNB-CU to assist in model inference. The information in the request message may include at least one of the following items: information that the gNB-DU wants to obtain from the gNB-CU (e.g., collected data to assist in model inference), an indication that requests the gNB-CU to provide feedback on the AI / ML model, or an indication that requests the gNB-CU to provide collected data to evaluate the trained model.

[0113] Step 8: The gNB-CU can send support information to the gNB-DU via F1AP. The support information may include at least one of the following items: UE ID, UE group information, UE predicted trajectory information, UE mobility information / UHI, UE previous QoE measurement results (e.g., buffer level, playback delay for media boot, throughput, frame rate, round trip time, or generalized RAN visible QoE value), or UE predicted QoE (e.g., if the gNB-CU also has model inference capabilities, buffer level, playback delay for media boot, throughput, frame rate, round trip time, or generalized RAN visible QoE value).

[0114] Step 9: The gNB-DU may collect information from the UE and gNB-CU and perform model inference to predict the QoE results or RAN-visible QoE results for a group of UEs. Apart from the QoE or RAN-visible QoE results, the prediction information may include at least one of the following items: predicted trajectory information for several UEs, mobility information / UHI for UEs, or previous / predicted QoE measurements for UEs. The gNB-DU may take any action based on the results of the model inference (e.g., deactivating or pausing RAN-visible QoE reporting via F1AP). In the case of model inference, the model inference function can predict the QoE results for UEs regardless of whether the UEs provided QoE measurements or other measurements for model training / inference.

[0115] Step 10: The gNB-DU may send the predicted information and actions to the gNB-CU via the F1AP. The information sent via the F1AP may include at least one of the following items: UE ID, predicted QoE results for the corresponding UE (e.g., buffer level, playback delay for media boot, throughput, frame rate, round-trip time, or generalized RAN visible QoE value), predicted trajectory information for the UE, mobility information / UHI for the UE, updated group information for the UE, proposed RAN visible QoE configuration (e.g., reporting cycle, RAN visible QoE metric, threshold trigger, event trigger), or an indication to deactivate / pause reporting of RAN visible QoE via the F1AP.

[0116] Step 11: In a particular embodiment, the gNB-DU may provide feedback to the gNB-CU as requested by the gNB-CU in the previous step. The feedback message may include at least one of the following items: the accuracy of the model inference, the variance of the predictions against the actual measured and collected data, a generalized value for evaluating the performance of the inference model (e.g., good or bad), or the actual collected data after the gNB-DU has received the prediction data from the gNB-CU.

[0117] It should be understood that one or more features from the above implementation examples are not exclusive to any particular implementation example and / or embodiment, and can be combined in any way (e.g., in any priority and / or order, simultaneously or otherwise).

[0118] Figure 10 illustrates a flowchart of Method 1000 for predicting the perceived quality of experience (QoE) of a user device (UE) using an artificial intelligence (AI) model. Method 1000 may be implemented using any one or more of the components and devices detailed herein in relation to Figures 1 and 2. In general, Method 1000 may be implemented by network nodes in some embodiments. Depending on the embodiment, Method 1000 may perform additional, fewer, or different operations. At least one aspect of the operation relates to a system, method, apparatus, or computer-readable medium.

[0119] A first network node of a Radio Access Network (RAN) (e.g., a gNB, or a distributed unit (DU), or a centralized unit (CU)) may receive first support information from a second network node of the RAN (e.g., a gNB, or a distributed unit (DU), or a centralized unit (CU)) for use with first quality of experience (QoE) information to perform a first function (e.g., model inference or model training) of a neural network model (e.g., model function / step / process). The first network node of the RAN may then perform the first function using the first QoE information and the first support information. Input information collected from neighboring gNBs for artificial intelligence (AI) functions may influence the Xn Application Protocol (XnAP).

[0120] In some embodiments, a first network node may receive first quality-of-effect (QoE) information obtained from measurements in a wireless communication device (e.g., a user device (UE)). The first network node may also receive second QoE information obtained from measurements in the wireless communication device or another wireless communication device for use in performing a second function of a neural network model (e.g., model training or model inference). Input QoE information for AI functions can be collected from the UE side, thereby preventing potential influence of the Uu interface on QoE measurements. The first or second QoE information may include an indication of at least one of the following: a QoE reporting container, at least one protocol data unit (PDU) session identifier (ID), at least one QoS flow ID, at least one data radio bearer (DRB) ID, a slice list, or QoE measurement results visible in the RAN. A list of information that may be included in the QoE information for use with AI functions. Currently, QoE measurement information is not used for AI functions.

[0121] In some embodiments, a first network node may receive first user equipment (UE) support information provided by one or another wireless communication device for use in performing a second function (e.g., model inference or model training) of a neural network model (e.g., model function / step / process). The first network node may also receive second UE support information provided by the wireless communication device for use in performing a first function of the neural network model. Input information collected from the UE side as support information for using AI model training or inference, apart from QoE information that can cover Uu effects. The first or second UE support information may include at least one indication of service type, application, slice information, radio bearer information, cell information, beam information, binding information or group identifier (ID), UE location information, UE history information, radio link quality-related information, UE measurements regarding reference signal received power (RSRP), reference signal received quality (RSRQ), or signal-to-interference noise ratio (SINR) for the serving cell and at least one neighboring cell, drive test minimization (MDT) measurements, or UE performance information. A list of information may be included in the UE support for the use of AI functionality. Currently, there is no support information collected from the UE for use in predicting the UE's QoE.

[0122] In some embodiments, a first or third network node (e.g., a core network (CN) or an operations, management, and maintenance (OAM)) may send a message to the second network node requesting second support information for use in performing a second function of the neural network model (e.g., model training). The message may include at least one of the following: an indication of what information is requested from the second network node to the first network node; a request to the second network node to provide feedback on the neural network model; or a request to the second network node to provide data collected to evaluate the performance of a trained version of the neural network model. The collected data may include the following items: UE identifier, QoE measurement results collected at NG-RAN node 2, RAN visible QoE results collected at NG-RAN node 2, or UE mobility information. The first or third network node may receive the second support information from the second network node. The model training module may be deployed in the OAM or the core network (CN). In the case of core networks, the NG Application Protocol (NGAP) may have an impact (for example, if a CN requests support information from a gNB, the gNB may report the support information to the CN).

[0123] In some embodiments, the first or second support information may include at least one user equipment (UE) identifier (ID) of a UE associated with the second network node, binding information or group ID of multiple UEs associated with the second network node, predicted or historical trajectories of one or more of the UEs, mobility information or UE history information (UHI) of one or more of the UEs, at least one previous QoE measurement result of one or more of the UEs, measured or predicted QoE information of one or more of the UEs (e.g., if gNB2 also has model inference capabilities), transmission delay, cell list, or resource status indication. The first or second support information may be a list of items in the support information between network nodes, particularly when model training is deployed on OAM or CN.

[0124] In some embodiments, a first network node may send a request message to a second network node requesting first support information. The request message may include at least one of the following: an indication of what information is requested from the second network node to the first network node; a request to the second network node to provide feedback on the neural network model; or a request to the second network node to provide data collected to evaluate the performance of a trained version of the neural network model.

[0125] In some embodiments, the first network node may transmit information predicted or inferred according to a second function (e.g., model inference) to at least one other network node (e.g., a neighboring gNB) via the F1AP interface, the information including at least one identifier of at least one user equipment (UE) that has provided measurement results or has not provided any measurement results, predicted QoE information for at least one UE, predicted trajectory information for at least one UE, predicted or updated grouping of at least one UE, correlation coefficients showing the correlation between one UE and another UE, time information in the validity period, action to be taken, mobility information or UE history information (UHI) of at least one UE, at least one previous QoE measurement result of at least one of the at least one UE, a modified QoE measurement configuration as seen in the RAN, updated grouping information for at least one UE, a proposed or predicted QoE configuration as seen in the RAN, or at least one indication to deactivate or suspend reporting of the QoE configuration as seen in the RAN. The output of the model inference module can be forwarded to neighboring nodes as a basis for further action.

[0126] In some embodiments, the first function may include model inference using a trained version of the neural network model. The second function may include model training of the neural network model.

[0127] In some embodiments, the first network node may include a first base station, a base station central unit (CU), or a base station distributed unit (DU). The second network node may include a second base station, a base station DU, or a base station CU. The third network node may include a core network (CN) node, or an operations, management, and maintenance (OAM) node / function.

[0128] In some embodiments, a first or second network node may receive a QoE configuration from an Operations, Management, and Maintenance (OAM) node or an Access and Mobility Management function. The first or second network node may transmit the QoE configuration to a wireless communication device. The QoE configuration may include at least one of the following: an indicator for using a first or second function of a neural network model, or binding information. The binding information may include at least one of the following: group information of multiple user equipment (UEs), service types of multiple UEs, applications or application types of multiple UEs, protocol data unit (PDU) session identifiers (IDs) of multiple UEs, quality of service (QoS) flow IDs of multiple UEs, radio bearer information of multiple UEs, location information of multiple UEs, or QoE user consent of multiple UEs.

[0129] In some embodiments, a first network node may send a request signaling to a second network node, the request signaling including at least one of the following: a flag used to request the second network node to provide feedback for neural network model inference; an indication of what information is being requested from the second network node to the first network node; or a flag used to request the second network node to provide collected real data for the first network node to evaluate the feedback for neural network model inference. Feedback can be a critical feature in AI, thereby enabling the node performing model inference to evaluate the performance of the AI ​​model.

[0130] In some embodiments, a first network node may receive feedback from a second network node. The feedback may be calculated based on predictive information provided by the first network node and actual data collected at the second network node. The feedback may include an indication of at least one of the following: the accuracy of neural network model inference, the confidence in the predictions of the neural network model (e.g., neural network model inference) for actual measured or collected data, a generalized value for evaluating the performance of neural network model inference, or a correlation of QoE between at least two user devices (UEs).

[0131] In some embodiments, the first network node may receive data collected by the second network node from the second network node. Receiving data from the second network node may cover the impact of XnAP for the feedback procedure.

[0132] In some embodiments, a second network node of a radio access network (RAN) (e.g., a gNB, or a distributed unit (DU), or a centralized unit (CU)) may transmit first support information to a first network node of the RAN (e.g., a gNB, or a distributed unit (DU), or a centralized unit (CU)) for use together with first quality-of-effect (QoE) information to perform a first function of a neural network model (e.g., model inference). The first QoE information can be obtained from measurements in a wireless communication device (e.g., a user equipment (UE)) and can be received by the first network node.

[0133] While various embodiments of this solution have been described above, it should be understood that they are presented only as examples and not as limitations. Similarly, various figures may depict exemplary architectures or configurations, which are provided to enable those skilled in the art to understand the exemplary features and functions of this solution. However, such those skilled in the art will understand that this solution is not limited to the exemplary architectures or configurations described above and can be implemented using various alternative architectures and configurations. Furthermore, as will be understood by those skilled in the art, one or more features of one embodiment may be combined with one or more features of another embodiment described herein. Thus, the breadth and scope of this disclosure should not be limited by any of the exemplary embodiments described above.

[0134] Furthermore, it should be understood that any reference to elements in this specification using designations such as "first," "second," etc., does not generally limit the number or order of those elements. Rather, these designations may be used in this specification as a convenient means of distinguishing two or more elements or instances of elements. Thus, references to a first element and a second element do not mean that only two elements can be used, nor that the first element must precede the second element in any way.

[0135] Furthermore, those skilled in the art will understand that information and signals can be represented using any of the various different technologies and techniques. For example, the data, instructions, commands, information, signals, bits, and symbols that may be mentioned in the above description can be represented by voltage, electric current, electromagnetic waves, magnetic fields or particles, optical fields or particles, or any combination thereof.

[0136] As those skilled in the art will further see, any of the various exemplary logic blocks, modules, processors, means, circuits, methods, and functions described in relation to the embodiments disclosed herein can be implemented by various forms of programs or design code incorporating electronic hardware (e.g., digital implementation, analog implementation, or a combination of both), firmware, instructions (for convenience, referred to herein as “software” or “software modules”), or any combination of these techniques. To clearly illustrate this compatibility of hardware, firmware, and software, various exemplary components, blocks, modules, circuits, and steps are described above in general with respect to their functions. Whether such functions are implemented as hardware, firmware, or software, or as a combination of these techniques, depends on the specific application and the design constraints imposed on the overall system. Those skilled in the art can implement the described functions in various ways for specific applications, but such implementation decisions do not deviate from the scope of this disclosure.

[0137] Furthermore, those skilled in the art will understand that the various exemplary logic blocks, modules, devices, components, and circuits described herein may be implemented, or can be implemented, within an integrated circuit (IC) which may include a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic device, or any combination thereof. The logic blocks, modules, and circuits may further include antennas and / or transceivers for communicating with various components within a network or device. The general-purpose processor may be a microprocessor, but alternatively, the processor may be any conventional processor, controller, or state machine. The processor may also be implemented as a combination of computing devices, e.g., a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors combined with a DSP core, or any other suitable configuration for performing the functions described herein.

[0138] When implemented in software, the functionality can be stored as one or more instructions or code on a computer-readable medium. Thus, the steps of the methods or algorithms disclosed herein can be implemented as software stored on a computer-readable medium. The computer-readable medium includes both computer storage media and communication media, which include any media that can enable the transfer of computer programs or code from one location to another. The storage media can be any available medium that can be accessed by a computer. Such computer-readable media may, but are not limited to, include, for example, RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store desired program code in the form of instructions or data structures and can be accessed by a computer.

[0139] As used herein, the term “module” refers to software, firmware, hardware, and any combination thereof for performing the relevant functions described herein. Furthermore, although various modules are described as individual modules for the purposes of discussion, as will be apparent to those skilled in the art, two or more modules may be combined to form a single module that performs the relevant functions according to the embodiments of this solution.

[0140] Furthermore, in embodiments of this solution, memory or other storage, as well as communication components, may be used. For clarity, it will be understood that the above description illustrates embodiments of this solution with reference to different functional units and processors. However, it will be clear that any appropriate distribution of functions between different functional units, processing logic elements, or regions may be used without impairing the solution. For example, functions exemplified as being performed by separate processing logic elements or controllers may be performed by the same processing logic element or controller. Thus, references to specific functional units are not intended to indicate a strict logical or physical structure or organization, but merely to refer to appropriate means for providing the described functions.

[0141] Various modifications to the embodiments described herein will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other embodiments without departing from the scope of this disclosure. Therefore, this disclosure is not limited to the embodiments shown herein, but should be given the broadest scope that coincides with the novel features and principles disclosed herein, as set forth in the following claims.

Claims

1. It is a method, A first network node of a wireless access network (RAN) receives first support information from a second network node of the RAN to be used together with first quality-of-effect (QoE) information to perform a first function of a neural network model, The first network node performs the first function using the first QoE information and the first support information. Methods that include...

2. The first network node receives first quality of experience (QoE) information obtained from measurements in a wireless communication device, or The first network node receives second QoE information obtained from measurements in the wireless communication device or another wireless communication device for use in performing a second function of the neural network model. The method according to claim 1, comprising at least one of the following.

3. The first QoE information or the second QoE information is, QoE reporting container, At least one Protocol Data Unit (PDU) session identifier (ID), At least one QoS flow ID, At least one data radio bearer (DRB) ID, Slice list, or QE measurement results visible in the aforementioned RAN The method according to claim 2, comprising at least one indication from among the following.

4. The first network node receives first user equipment (UE) support information provided by the wireless communication device or another wireless communication device for use in performing a second function of the neural network model, or The first network node receives second UE-assisted information provided by the wireless communication device for use in performing the first function of the neural network model. The method according to claim 1, comprising at least one of the following.

5. The first UE support information or the second UE support information is, Service type, application, Slice information, Wireless bearer information, Cell information, Beam information, Binding information or group identifier (ID), UE location information, UE history information, Wireless link quality related information, UE measurements related to reference signal received power (RSRP), reference signal received quality (RSRQ), or the signal-to-interference and noise ratio (SINR) of the serving cell and at least one neighboring cell. Drive test minimum (MDT) measurement, or UE performance information The method according to claim 4, comprising at least one indication of the following.

6. The first network node or the third network node sends a message to the second network node requesting second support information to be used to perform the second function of the neural network model, wherein the message is: Indication of which information is requested from the second network node to the first network node, A request to the second network node to provide feedback regarding the neural network model, or A request to the second network node to provide data collected to evaluate the performance of the trained version of the neural network model. Having at least one of the following, The first or third network node receives the second support information from the second network node. The method according to claim 1, comprising at least one of the following.

7. The first support information or the second support information is At least one user equipment (UE) identifier (ID) of a UE associated with the second network node, Binding information or group ID of multiple UEs associated with the second network node, The predicted trajectory or historical trajectory of one or more of the aforementioned UEs, Mobility information or UE history information (UHI) of one or more of the aforementioned UEs, At least one previous QoE measurement result for one or more of the aforementioned UEs, Measured or predicted QoE information of one or more of the aforementioned UEs, Transmission delay, Cell list, or Resource status The method according to claim 6, comprising at least one indication of the following.

8. The first network node includes sending a request message to the second network node to request the first support information, the request message being: Indication of which information is requested from the second network node to the first network node, A request to the second network node to provide feedback regarding the neural network model, or A request to the second network node to provide data collected to evaluate the performance of the trained version of the neural network model. The method according to claim 1, comprising at least one of the following.

9. The first network node includes transmitting information predicted or inferred in accordance with the second function to at least one other network node via the F1AP interface, wherein the information is At least one identifier of at least one user device (UE) that has provided measurement results or has not provided any measurement results, QoE information predicted with respect to at least one UE, Trajectory information predicted for at least one of the aforementioned UEs, The predicted or updated grouping of at least one UE, The correlation coefficient, which shows the correlation between one UE and another UE. Time information regarding the validity period, Measures to be taken, The mobility information or UHI (U-Hi) of at least one UE, At least one previous QoE measurement result of one or more of the above at least one UE, The modified QoE measurement configuration shown in the aforementioned RAN, The updated group information of at least one UE, The proposed or predicted QoE configurations visible in the aforementioned RAN, or An indication to deactivate or temporarily suspend the reporting of the QoE configuration visible in the aforementioned RAN. The method according to claim 2, comprising at least one of the following.

10. The first function includes model inference using a trained version of the neural network model, or The second function includes model training of the neural network model. The method according to claim 1, 2, 4, 6, or 9, wherein at least one of the above.

11. The first network node includes a first base station, a central unit (CU) of the base station, or a distributed unit (DU) of the base station. The second network node includes a second base station, the DU of the base station, or the CU of the base station, or The third network node includes a node of the core network (CN) or an operations, management, and maintenance (OAM) node. The method according to any one of claims 1 to 10, wherein at least one of the above.

12. The first network node or the second network node receives the QoE configuration from an Operations, Management, and Maintenance (OAM) node or an Access and Mobility Management function, The first network node or the second network node transmits the QoE configuration to the wireless communication device. The method according to claim 1, including the method described in claim 1.

13. The aforementioned QoE configuration is, An indicator for using the first or second function of the neural network model, or Binding information, Group information of multiple user devices (UEs), The service types of the aforementioned multiple UEs, The aforementioned multiple UE applications or application types, The protocol data unit (PDU) session identifiers (IDs) of the plurality of UEs, The Quality of Service (QoS) flow IDs of the aforementioned multiple UEs, The wireless bearer information of the aforementioned multiple UEs, The location information of the plurality of UEs, or Multiple UE QoE user consents, Binding information including at least one of the following The method according to claim 12, comprising at least one of the indications.

14. The first network node includes sending a request signaling to the second network node, the request signaling being: A flag used to request the second network node to provide feedback for neural network model inference, Indication of which information is requested from the second network node to the first network node, or A flag used by the first network node to request the second network node to provide collected real data for evaluating the feedback for the neural network model inference. The method according to claim 1, comprising at least one of the following.

15. The first network node receives feedback from the second network node, calculated according to the prediction information provided by the first network node and the actual data collected at the second network node, wherein the feedback is: Accuracy of neural network model inference, The confidence level of predictions made by neural network models on actual measured or collected data. A generalized value for evaluating the performance of the neural network model inference, or Correlation of QoE between at least two user devices (UEs) The method according to claim 1 or 14, comprising at least one indication from among the following.

16. The method according to claim 1 or 14, wherein the first network node receives data collected by the second network node from the second network node.

17. It is a method, The second network node of a wireless access network (RAN) transmits to the first network node of the RAN first support information for use together with first quality-of-experience (QoE) information to perform a first function of a neural network model, A method wherein the first QoE information is obtained from measurements in a wireless communication device and received by the first network node.

18. A non-temporary computer-readable medium for storing instructions, wherein the instructions, when executed by at least one processor, cause the at least one processor to perform the method according to any one of claims 1 to 17.

19. An apparatus comprising at least one processor configured to perform the method described in any one of claims 1 to 17.

Images