Communication devices and methods

Abstract

Embodiments of this disclosure provide a solution for ground truth data collection for model monitoring. In this solution, a first communication device receives a model trigger condition instruction or model description related to an artificial intelligence / machine learning (AI / ML) model from a second communication device. The model trigger condition instruction indicates a trigger condition for triggering the AI / ML model applied in direct or assisted positioning of the terminal device. The model description includes attribute information for ground truth collection. Based on the model trigger condition instruction or model description, the first communication device determines a target ground truth collection mode for the AI / ML model and collects ground truth data to monitor the AI / ML model based on the target ground truth collection mode.

Description

Technical Field

[0001] Exemplary embodiments of the present disclosure generally relate to the field of communication technologies, and more particularly, to devices and methods for ground truth data collection for model monitoring.

Background Art

[0002] In the telecommunications industry, artificial intelligence / machine learning (AI / ML) models have been adopted in telecommunications systems to improve the performance of the telecommunications systems. For example, supporting various positioning mechanisms to provide a highly reliable and accurate position of a terminal device has always been one of the important features of a telecommunications system. To improve the overall performance of a telecommunications system, it has been agreed to investigate the potential of AI / ML models at the air interface. An AI / ML-based positioning mechanism for improving positioning accuracy is one of the use cases for applying AI / ML at the air interface.

Summary of the Invention

Means for Solving the Problems

[0003] Generally, embodiments of the present disclosure provide a method, a device, and a computer storage medium for ground truth data collection for model monitoring.

[0004] In a first embodiment, a first communication device is provided. The first communication device includes a processor, which is configured to receive from a second communication device a model trigger condition instruction or model description relating to an artificial intelligence / machine learning (AI / ML) model, wherein the model trigger condition instruction indicates a trigger condition for triggering an AI / ML model applied in direct positioning or assisted positioning of a terminal device, and the model description includes attribute information for ground truth collection, and to determine a target ground truth collection mode for the AI / ML model based on the model trigger condition instruction or model description, and to collect ground truth data for monitoring the AI / ML model based on the target ground truth collection mode.

[0005] In a second embodiment, a second communication device is provided. The second communication device includes a processor configured to transmit to the first communication device a model trigger condition instruction or model description relating to an artificial intelligence / machine learning (AI / ML) model, wherein the model trigger condition instruction indicates a trigger condition for triggering an AI / ML model applied in direct or assisted positioning of a terminal device, and the model description includes attribute information for ground truth collection; the first communication device is configured to monitor the AI / ML model, and a target ground truth collection mode for collecting ground truth data of the AI / ML model is determined based on the model trigger condition instruction or model description.

[0006] In a third aspect, a communication method is provided. This method includes receiving a model trigger condition instruction or model description related to an artificial intelligence / machine learning (AI / ML) model from a second communication device, wherein the model trigger condition instruction indicates a trigger condition for triggering an AI / ML model applied in direct or assisted positioning of a terminal device, and the model description includes attribute information for ground truth collection; determining a target ground truth collection mode for the AI / ML model based on the model trigger condition instruction or model description; and collecting ground truth data for monitoring the AI / ML model based on the target ground truth collection mode.

[0007] In a fourth aspect, a communication method is provided, which includes transmitting a model trigger condition instruction or model description related to an artificial intelligence / machine learning (AI / ML) model to a first communication device by a second communication device, wherein the model trigger condition instruction indicates a trigger condition for triggering an AI / ML model applied in direct or assisted positioning of a terminal device, and the model description includes attribute information for ground truth collection, the first communication device is configured to monitor the AI / ML model, and a target ground truth collection mode for collecting ground truth data of the AI / ML model is determined based on the model trigger condition instruction or model description.

[0008] In the fifth aspect, a computer-readable medium storing instructions is provided, and when the instructions are executed by at least one processor, the instructions cause at least one processor to perform the method according to the third or fourth aspect.

[0009] Other features of this disclosure should be easily understood by reading the following explanation.

[0010] The above and other purposes, features, and advantages of this disclosure will become more apparent through a more detailed description of some exemplary embodiments of this disclosure in the accompanying drawings. [Brief explanation of the drawing]

[0011] [Figure 1A] This disclosure describes an exemplary communication environment in which several exemplary embodiments of this disclosure may be implemented.

[0012] [Figure 1B] This shows the signaling flow for the positioning procedure of a terminal device.

[0013] [Figure 2] This disclosure illustrates signaling flows for model monitoring according to several embodiments of this disclosure.

[0014] [Figure 3A] This disclosure provides exemplary workflows for AI / ML-based positioning inference and validation according to several embodiments of this disclosure.

[0015] [Figure 3B] Examples of triangulation-based positioning applied to ground truth data collection, according to several embodiments of this disclosure, are shown.

[0016] [Figure 4A] The following are exemplary formats for model trigger condition indications according to some embodiments of this disclosure. [Figure 4B] The following are exemplary formats for model trigger condition indications according to some embodiments of this disclosure.

[0017] [Figure 5A] This disclosure presents exemplary models for monitoring opportunity patterns according to several embodiments of this disclosure. [Figure 5B] This disclosure presents exemplary models for monitoring opportunity patterns according to several embodiments of this disclosure.

[0018] [Figure 6] Shows the signaling flow of an exemplary positioning procedure using an AI / ML model on the network device side, according to some embodiments of the present disclosure.

[0019] [Figure 7] Shows the signaling flow of an exemplary positioning procedure using an AI / ML model on the LMF (Location Management Function) side, according to some further embodiments of the present disclosure.

[0020] [Figure 8] Shows a flowchart of a method implemented on a first communication device, according to some exemplary embodiments of the present disclosure.

[0021] [Figure 9] Shows a flowchart of a method implemented on a second communication device, according to some exemplary embodiments of the present disclosure.

[0022] [Figure 10] It is a simplified block diagram of a device suitable for implementing embodiments of the present disclosure.

Mode for Carrying Out the Invention

[0023] Throughout the drawings, the same or similar reference numerals represent the same or similar elements.

[0024] [[ID=,38]] Here, the principles of the present disclosure will be described with reference to some exemplary embodiments. These embodiments are described only for the purpose of illustration and are not intended to suggest any limitation on the scope of the present disclosure. It should be understood that they are to assist those skilled in the art in understanding and implementing the present disclosure. The disclosure described herein can be implemented in various ways other than those described below.

[0025] In the following description and claims, unless otherwise defined, all technical and scientific terms used herein have the same meanings as those generally understood by those skilled in the art to which this disclosure belongs.

[0026] As used herein, the term “terminal device” refers to any device having wireless or wired communication capabilities. Examples of terminal devices include user equipment (UE), personal computers, desktops, mobile phones, cell phones, smartphones, personal digital assistants (PDA), portable computers, tablets, wearable devices, Internet of Things (IoT) devices, ultra-reliable and low-latency communications (URLLC) devices, Internet of Everything (IoE) devices, machine-type communication (MTC) devices, in-vehicle devices for V2X communication (where X represents pedestrians, vehicles, or infrastructure / networks), integrated access and backhaul (IAB) devices, spacecraft or aircraft in non-terrestrial networks (NTN) including high-altitude platforms (HAP) encompassing satellites and unmanned aircraft systems (UAS), augmented reality (AR), and mixed reality (MR). Examples include, but are not limited to, extended reality (XR) devices that include different types of reality such as reality and virtual reality (VR), unmanned aerial vehicles (UAVs) that do not require a human pilot, commonly known as drones, devices on high-speed trains (HSTs), imaging devices such as digital cameras, sensors, game consoles, music storage and playback devices, or internet devices that enable wireless / wired internet access and browsing.The “Terminal Device” may further have “Multicast / Broadcast” capabilities to support public safety and mission-critical applications, V2X applications, transparent IPv4 / IPv6 multicast distribution, IPTV, smart TV, wireless services, software distribution over wireless, group communications, and IoT applications. It may also incorporate one or more subscriber identity modules (SIMs), known as multi-SIM (SIM: Subscriber Identity Module). The term “Terminal Device” may be used interchangeably with UE, mobile station, subscriber station, mobile terminal, user terminal, or wireless device.

[0027] The term "network device" refers to a device that can provide or host a cell or coverage from which terminal devices can communicate. Examples of network devices include, but are not limited to, Node B (NodeB or NB), evolved Node B (eNodeB or eNB), next-generation Node B (gNB), Transmission Reception Point (TRP), Remote Radio Unit (RRU), Radio Head (RH), Remote Radio Head (RRH), low-power nodes such as IAB nodes, femtonodes, and piconodes, and Reconfigurable Intelligent Surface (RIS).

[0028] Terminal devices or network devices may have artificial intelligence (AI) or machine learning capabilities. AI or machine learning capabilities generally include models trained from large amounts of data collected for a specific function, and can be used to predict certain information.

[0029] Terminal or network devices can operate in several frequency ranges, such as FR1 (e.g., 450 MHz to 6000 MHz), FR2 (e.g., 24.25 GHz to 52.6 GHz), frequency bands greater than 100 GHz, and terahertz (THz). Furthermore, terminal or network devices can operate in licensed / unlicensed / shared spectrum. Terminal devices may have multiple connections with network devices under Multi-Radio Dual Connectivity (MR-DC) application scenarios. Terminal or network devices can operate in full duplex mode, flexible duplex mode, and cross-division duplex mode.

[0030] Embodiments of this disclosure may be implemented using test equipment, such as signal generators, signal analyzers, spectrum analyzers, network analyzers, test terminal devices, test network devices, and channel emulators. In some embodiments, the terminal device may be connected to a first network device and a second network device. One of the first and second network devices may be a master node and the other may be a secondary node. The first and second network devices may use different radio access technologies (RATs). In some embodiments, the first network device may be a first RAT device and the second network device may be a second RAT device. In some embodiments, the first RAT device may be an eNB and the second RAT device may be a gNB. Information related to different RATs may be transmitted from at least one of the first or second network devices to the terminal device. In some embodiments, the first information may be transmitted from the first network device to the terminal device, and the second information may be transmitted from the second network device directly or via the first network device to the terminal device. In some embodiments, information relating to the configuration of a terminal device set by the second network device may be transmitted from the second network device via the first network device. Information relating to the reconfiguration of a terminal device set by the second network device may be transmitted from the second network device directly to the terminal device or via the first network device.

[0031] Where used herein, the singular forms “a,” “an,” and “it” are intended to include the plural unless the context makes otherwise clear. The term “including” and its variations shall be read as an open term meaning “including, but not limited to.” The term “based on” shall be read as “based at least in part.” The terms “one embodiment” and “one embodiment” shall be read as “at least one embodiment.” The term “another embodiment” shall be read as “at least one other embodiment.” Terms such as “first,” “second,” etc., may refer to different or the same subject. Other explicit and implicit definitions may be included below.

[0032] In some examples, values, procedures, or devices are referred to as “best,” “worst,” “highest,” “minimum,” “maximum,” etc. Such descriptions are intended to show that a choice may be made from among many functional alternatives, and it should be understood that such a choice does not necessarily have to be better, smaller, higher, or otherwise preferable than other choices.

[0033] As used herein, the terms “resource,” “transmission resource,” “uplink resource,” or “downlink resource” may refer to any resource for performing communication, such as a time-domain resource, a frequency-domain resource, a spatial-domain resource, a code-domain resource, or any other resource that enables communication. Hereinafter, unless expressly stated, resources in both the frequency-domain and time-domain will be used as examples of transmission resources to illustrate some exemplary embodiments of this disclosure. It should be noted that the exemplary embodiments of this disclosure are equally applicable to other resources in other domains.

[0034] As used herein, the term “model” refers to the association between an input and an output learned from training data, and therefore, after training, a corresponding output can be produced for a given input. Model generation can be based on machine learning techniques, also known as artificial intelligence (AI) techniques. Generally, machine learning models can be constructed that receive input information and make predictions based on that input information. For example, a classification model can predict the class of input information from a given set of classes. As used herein, “model” may also be referred to as a “machine learning model,” “learning model,” “machine learning network,” or “learning network,” and these terms are used interchangeably herein.

[0035] Generally, machine learning can involve three stages: the training stage, the validation stage, and the application stage (also called the inference stage). In the training stage, a given machine learning model can be iteratively trained (or optimized) using a large amount of training data until the model can derive consistent inferences from the training data similar to those that human intelligence can perform. During training, the set of parameter values ​​for the model is iteratively updated until the training goal is reached. Throughout the training process, the machine learning model can be considered capable of learning associations between inputs and outputs (also called input-output mappings) from the training data. In the validation stage, validation inputs are applied to the trained machine learning model to test whether the model can provide the correct output and to determine the model's performance. Generally, the validation stage may be considered a step in the training process, or it may be sometimes omitted. In the application stage, the resulting machine learning model can be used to process real-world model inputs based on the set of parameter values ​​obtained from the training process and to determine the corresponding model output.

[0036] Figure 1A shows a schematic diagram of an exemplary communication environment 100 that can implement some exemplary embodiments of the present disclosure. In the communication environment 100, multiple communication devices, including terminal devices 110, network devices 120, and core network functions such as a location management function (LMF) 130 and an access and mobility management function (AMF) 140, can communicate with each other. In some embodiments, there may be one or more location service (LCS) entities 150 that can communicate with the AMF 140 to request location services.

[0037] Communication in communication environment 100 may conform to any appropriate standard, including, but not limited to, Global System for Mobile Communications (GSM), Long Term Evolution (LTE), LTE-Evolution, LTE-Advanced (LTE-A), New Radio (NR), Wideband Code Division Multiple Access (WCDMA®), Code Division Multiple Access (CDMA), GSM EDGE Radio Access Network (GERAN), and Machine Type Communication (MTC). Embodiments of this disclosure may be implemented in accordance with any generation of communication protocols currently known or to be developed in the future. Examples of communication protocols include, but are not limited to, first generation (1G), second generation (2G), 2.5G, 2.75G, third generation (3G), fourth generation (4G), 4.5G, fifth generation (5G) communication protocols, 5.5G, 5G-Advanced networks, or sixth generation (6G) networks.

[0038] In the communication environment 100, the network device 120 may be a base station providing services to the terminal device 110. The service area of ​​the network device 120 may be called a cell (not shown). In the communication environment 100, the network device 120 and the terminal device 110 can communicate data and control information with each other.

[0039] It should be understood that the number of devices and their connections as shown in Figure 1A are for illustrative purposes only and do not imply any limitation. The communication environment 100 may include any appropriate number of devices configured to carry out the exemplary embodiments of this disclosure. It should be understood that one or more additional devices may be located within a cell, and one or more additional cells may be deployed in the communication environment 100, although these are not shown. It should be noted that although shown as a network device, network device 120 may be a device other than a network device. Although shown as a terminal device, terminal device 110 may be a device other than a terminal device, such as a positioning reference unit (PRU).

[0040] A positioning operation is specified for identifying the terminal device 110. Figure 1B shows the signaling flow 102 for the positioning procedure of the terminal device 110.

[0041] During operation, positioning of the terminal device 110 can be triggered for various reasons. For example, several LCS entities 150 within the core network (e.g., Gateway Mobile Location Centre (GMLC)) may request some location service (e.g., positioning) for the terminal device 110 from the service provider AMF 140 (1a). Alternatively, the service provider AMF 140 for the terminal device 110 may determine the need for some location service (e.g., to locate the terminal device 110 for an emergency call) (1b). Alternatively, the terminal device 110 may request some location service (e.g., positioning or distribution of support data) from the service provider AMF 140 at the Non-access Stratum (NAS) level (1c).

[0042] AMF140 forwards the location service request to LMF130 (2). LMF130 invokes a location procedure using the service provider and, if applicable, a nearby New Generation (NG) RAN network node, such as network device 120, to obtain positioning measurements or support data, for example (3a). In addition to or instead of step 3a, LMF130 invokes a location procedure on terminal device 110 to obtain a location estimate or positioning measurements, or to transfer location support data to terminal device 110, for example (3b).

[0043] The LMF130 provides the AMF140 with a location service response (4), which includes any necessary results, such as success or failure indications, and, if requested and obtained, a location estimation of the terminal device 110.

[0044] If step 1a is performed, the AMF 140 returns a location service response to the 5GC entity in step 1a (5a), including any necessary results, such as a location estimation of the terminal device 110. If step 1b is performed, the AMF 140 uses the location service response received in step 4 (5b) to support the service that triggered step 1b (for example, it may provide a location estimation associated with an emergency call to the GMLC). If step 1c is performed, the AMF 140 returns a location service response to the terminal device 110 (5C), including any necessary results, such as a location estimation of the terminal device 110.

[0045] Currently, it has been proposed to provide AI / ML-based positioning for terminal devices. The AI / ML model can be trained to perform direct or assisted positioning of the terminal device. In some embodiments, such an AI / ML model can be deployed in the terminal device 110, the network device 120, and / or the LMF 130, or both. As shown, AI / ML model 112 can be deployed in the terminal device 110 for direct or assisted positioning of the terminal device 110. Alternatively or additionally, AI / ML model 122 can be deployed in the network device 120 for direct or assisted positioning of the terminal device 110. Alternatively or additionally, AI / ML model 132 can be deployed in the LMF 130 for direct or assisted positioning of the terminal device 110. Alternatively or additionally, the AI / ML model can be deployed in all of the terminal device 110, the network device 120, and the LMF 130 for direct or assisted positioning of the terminal device 110.

[0046] As used herein, the term "AI / ML model" may be interchangeable with the term "model." The term "AI / ML model training" may refer to the process of training an AI / ML model, for example, by learning input / output relationships and retrieving certain features from the input / output for inference. As used herein, the term "model monitoring" may refer to the procedure of monitoring the inference performance of an AI / ML model.

[0047] An AI / ML model configured for direct positioning of a terminal device may also be called a direct AI / ML positioning model. The output of such an AI / ML model inference is the position of the terminal device. Inputs to a direct AI / ML positioning model may include channel observations such as channel observations, channel impulse response (CIR), power delay profile (PDP), reference signal received power (RSRP), reference signal time difference (RSTD), and / or other types of channel observations, including fingerprints.

[0048] An AI / ML model configured for assisted positioning of a terminal device may also be called an AI / ML assisted positioning model. The output of such an AI / ML model inference may include measurement and / or enhancement-related information that can be used to assist in determining the location of a terminal device. For example, measurement and / or enhancement-related information may include, but is not limited to, non-line of sight (NLOS) / line of sight (LOS) identification of channels, timing and / or angle of measurements (e.g., time of arrival (ToA), time difference of arrival (TDoA)), path phase, likelihood of measurement, soft information / high resolution of RSTD, etc. Inputs to an AI / ML assisted positioning model may be the same as inputs to a direct AI / ML positioning model, or may include less or different information in addition to the inputs to a direct AI / ML positioning model.

[0049] There are several use cases for improving the accuracy of AI / ML-based positioning. The first use case is UE-based positioning using a UE-side model, where the model can be a direct AI / ML or AI / ML-assisted positioning model. The second use case is UE-assisted / LMF-based positioning using a UE-side model, where the model can be an AI / ML-assisted positioning model. The third use case is UE-assisted / LMF-based positioning using an LMF-side model, where the model can be a direct AI / ML positioning model. The fourth use case is NG-RAN node-assisted positioning using a gNB-side model (model in network devices), where the model can be an AI / ML-assisted positioning model. The fifth use case is NG-RAN node-assisted positioning using an LMF-side model, where the model can be a direct AI / ML positioning model. It will be understood that there are various other use cases for AI / ML-based positioning.

[0050] In the embodiment shown in Figure 1A, the AI / ML models 112, 122, and 132 may be the same or different, and may be the same or different types of direct AI / ML positioning models and AI / ML assisted positioning models.

[0051] In some embodiments, one or more of the AI / ML models 112, 122, and 132 may be trained offline using data collected from the field. Offline training may refer to the AI / ML training process in which a model is trained based on a collected dataset and the trained model is later used or provided for inference. It should be noted that this definition is for guidance only. There may be cases that do not strictly fit this definition but can still be classified as offline training by generally accepted conventions.

[0052] In some embodiments, one or more of the AI / ML models 112, 122, and 132 may be trained online using data collected from the field. Online training may refer to an AI / ML training process in which the model used for inference is trained (usually sequentially) in (quasi) real time as new training samples arrive. Note that the concepts of (quasi) real time and non-real time are context-dependent and relate to the inference time scale. Also note that this definition is for guidance only. There may be cases that do not strictly fit this definition but can still be classified as online training by generally accepted conventions. In some embodiments, fine-tuning / retraining may be performed via online or offline training.

[0053] In some embodiments, one or more of the AI / ML models 112, 122, and 132 may be trained and then applied locally. In some embodiments, one or more of the AI / ML models 112, 122, and 132 may be trained remotely and then transferred to a corresponding device for use. For example, AI / ML model 112 may be trained on a network device 120 or LMF 130 and then transferred to a terminal device 110 for use. In some embodiments, there may be a separate entity within the communication environment 100, such as a model transfer entity 160 configured to train the trained AI / ML models 112, 122, and / or 132 and transfer them to a corresponding device for use. The model transfer entity 160 may include a network device in the RAN or CN, or it may be a third-party entity.

[0054] The lifecycle of an AI / ML model includes at least a model training phase, a model reference phase, a model fine-tuning phase, and a model monitoring phase. In the model training phase, the AI / ML model is trained using a training dataset. The training dataset generally includes the inputs to the AI / ML model and ground truth data related to the corresponding inputs. In the model reference phase, the trained AI / ML model is transferred to corresponding devices (e.g., terminal device 110, network device 120, and / or LMF 130) to process actual inputs. For example, in a positioning scenario, the trained AI / ML model is provided to terminal device 110 to determine (depending on the type of AI / ML model) location or measurement values ​​and / or enhancement-related information for positioning. During the model reference phase, the performance of the AI / ML model may be monitored.

[0055] Depending on the circumstances, if the AI / ML model is degraded, the output of AI / ML-based positioning and AI / ML-assisted positioning may deviate significantly from ground truth. For example, if the environment of a terminal device changes, the AI / ML model may no longer output accurate positioning results or intermediate measurements and / or enhancement-related information for terminal devices located in the changed environment. In this case, it may be necessary to deactivate, fine-tune, or replace the AI / ML model with a new one.

[0056] The propagation environment can change due to various factors, such as the movement of terminal devices and the presence of new obstacles. Significant changes in the propagation environment can dramatically degrade the performance of AI / ML-based positioning. To avoid prolonged performance degradation, quality monitoring of the AI / ML model is necessary, and some action should be taken if the AI / ML model becomes invalid.

[0057] In some embodiments, model monitoring may be performed on the device where the AI / ML model is deployed, for example, terminal device 110 for AI / ML model 112, network device 120 for AI / ML model 122, or LMF 130 for AI / ML model 132. In some embodiments, model monitoring may be performed on a device other than the device where the AI / ML model is deployed. For example, network device 120 may be responsible for model monitoring of both AI / ML models 112 and 122. In some embodiments, a separate entity, such as a model monitoring entity 170 configured to monitor AI / ML models 112, 122, and / or 132 deployed within the communication environment 100, may exist within the communication environment 100. The model monitoring entity 170 may include a network device in the RAN or CN, or it may be a third-party entity.

[0058] For each use case, several metrics / methods for monitoring AI / ML models in lifecycle management are proposed. In some embodiments, model monitoring may be based on inference accuracy, including metrics related to key performance indicators (KPIs). In some embodiments, model monitoring may be based on system performance, including metrics related to system performance KPIs. In some other embodiments, model monitoring may be based on data distribution. Data distribution may be input-based, for example, to monitor the effectiveness of AI / ML inputs, and may be simpler, such as out-of-distribution detection, input data drift detection, or checks for signal-to-noise ratio (SNR) and delay spread. Data distribution may be output-based, for example, to perform drift detection of output data from the model.

[0059] Some model monitoring solutions require the collection of ground truth data (also known as ground truth labels) or approximations thereof for the AI / ML model being monitored. Positioning-related measurements and associated ground truth labels are used to monitor (and train) the AI / ML model. In direct AI / ML positioning, the ground truth label is the location of the target terminal device. In AI / ML-assisted positioning, the ground truth label is the ideal or actual information of the measurement / report (e.g., LOS / NLOS identification, RSTD, etc.). Obtaining ground truth labels is crucial, and it is clear that non-AI means such as base station / satellite positioning are required to obtain them.

[0060] Currently, too many methods are being implemented, some within the 3GPP (registered trademark) framework and others outside of it. Furthermore, each positioning method has its own unique application scenarios where positioning accuracy is optimal. For model monitoring, it is expected that reliable model monitoring sources from reliable field data will be utilized.

[0061] By selecting the appropriate method based on the conditions that trigger the AI / ML model, field data for model monitoring can be acquired efficiently and accurately. If the entities involved in model monitoring understand why the AI / ML model is triggered, it is beneficial to collect more reliable field data to generate ground truth labels for model monitoring.

[0062] Exemplary embodiments of this disclosure provide an improved solution for ground truth data acquisition for model monitoring. According to this solution, a first communication device (which may be configured for model monitoring of an AI / ML model) receives a model trigger condition instruction or model description related to the AI / ML model from a second communication device. The model trigger condition instruction indicates a trigger condition for triggering the AI / ML model to be applied in direct or assisted positioning of a terminal device. The model description includes attribute information for ground truth acquisition. The first communication device determines a target ground truth acquisition mode for the AI / ML model based on the model trigger condition instruction or model description, and, based on the target ground truth acquisition mode, acquires ground truth data for monitoring the AI / ML model. This solution allows for the use of some additional information to assist field data acquisition, particularly for model monitoring. This improves the reliability of model monitoring by indicating why the AI / ML model is triggered for positioning and applying that reason to the initiation of positioning procedures for field data acquisition purposes.

[0063] The principles and embodiments of this disclosure will be described in detail below with reference to the drawings.

[0064] Figure 2 shows a signaling flow 200 for communication according to several embodiments of the present disclosure. The signaling flow 200 involves a first communication device 201 and a second communication device 202.

[0065] The first communication device 201 may be an entity configured for model monitoring of one or more AI / ML models. The monitored AI / ML models may be transferred to any suitable communication device for direct or assisted positioning of the terminal device. In some examples, the monitored AI / ML models may be AI / ML model 112 deployed on terminal device 110, AI / ML model 122 deployed on network device 120, or AI / ML model 132 deployed on LMF 130, or any other AI / ML model deployed in the communication environment. The first communication device 201 may be a device on which the AI / ML models are deployed (e.g., terminal device 110, network device 120, or LMF 130), another device such as AMF 140, or a separate entity configured for monitoring, such as model monitoring entity 170. The actual entity implemented as the first communication device 201 may depend on the actual application. In some embodiments, there may be two or more communication devices involved in ground truth data collection for monitoring the AI / ML models. In this case, each communication device involved may be considered as the first communication device 201 and may perform operations similar to those described below with respect to the first communication device 201.

[0066] In the signaling flow 200, the first communication device 201 receives a model trigger condition instruction or model description related to the AI / ML model from the second communication device 202 (210). The model trigger condition instruction indicates a trigger condition for triggering the AI / ML model to be applied in direct positioning or assisted positioning of the terminal device. The model description includes attribute information for ground truth collection.

[0067] As described above, AI / ML-based positioning generally has two sub-use cases, including direct AI / ML positioning, where the output of the AI / ML model inference is the UE location, and AI / ML-assisted positioning, where the output of the AI / ML model is measurement and / or enhancement-related information used to assist in determining the UE location. In the case of model monitoring based on inference accuracy, the monitoring data may include both the model input and the ideal model output, i.e., pairs of {input, output} in AI / ML mode. As some examples, in direct AI / ML positioning, the {input, output} pairs may include new measurements such as {CIR / PDP, UE location} or existing measurements such as {RSRP / RSRPP / RSTD, UE location}. As some other examples, in AI / ML-assisted positioning, a {input, output} pair may include a new measurement report such as {CIR / PDP / RSRP / RSRPP / RSTD / ...,ToA / path phase}, an existing measurement report such as {CIR / PDP / RSRP / RSRPP / RSTD / ...,RSTD / (LOS / NLOS indicator) / RSRPP}, or an extension of an existing measurement report such as {CIR / PDP / RSRP / RSRPP / RSTD / ...,RSTD soft information / high resolution}.

[0068] Each of the sub-use cases (i.e., direct AI / ML positioning and AI / ML-assisted positioning) requires UE location (or related) information for model monitoring, which requires validation by the AI / ML model. As shown in Figure 3A, the exemplary workflow 300 may include providing model inputs to the AI / ML model 310 for inference, and thus outputting location or related information as ground truth output required for validation of the AI / ML model 310.

[0069] Unlike other use cases, monitoring an AI / ML model with field data collected offline is not feasible because it is difficult to "match" the model input (e.g., CIR / PDP) with the model output (e.g., location) when the target terminal device is in an unpredictable physical environment. UE location (or related) information can be obtained based on measurements using several radio access technology (RAT)-dependent and RAT-independent methods, such as Observed Time Difference of Arrival (OTDOA), multi-Round-Trip Time (multi-RTT), or Global Navigation Satellite System (GNSS). However, the reliability of model monitoring is questionable because ground truth data may also be unreliable if a non-AI modeling method is one of the reasons the AI / ML model is triggered to estimate the UE's location or improve positioning accuracy. This is reasonable if the reason for triggering the AI / ML model is not due to the unreliability of the non-AI modeling method.

[0070] In view of the foregoing, at least in the embodiments of the present disclosure, it is proposed to implicitly provide some additional information to assist in field data collection, particularly for model monitoring. Thus, model trigger condition instructions are communicated to the first communication device 201 to indicate trigger conditions for triggering an AI / ML model in direct positioning or assisted positioning. Additionally or alternatively, explicit methods are also provided in combination with current agreement cases. Thus, the model description is defined to include attribute information for ground truth collection. Thus, the model trigger condition instructions or attribute information for ground truth collection may help the first communication device 201 determine how reliable the ground truth data that can be collected for the AI / ML model is.

[0071] The second communication device 202, which transmits a model trigger condition instruction or model description to the first communication device 201 (205), may be any entity within the communication device that knows the trigger conditions of the AI / ML model monitored by the first communication device 201, or any entity that can access and transfer the model description.

[0072] The first communication device 201 determines the target ground truth acquisition mode for the AI / ML model based on a model trigger condition instruction or model description (215), and collects ground truth data for monitoring the AI / ML model based on the target ground truth acquisition mode (220).

[0073] Before discussing the decision on the target ground truth collection, let's first introduce the candidate ground truth collection modes.

[0074] There are various positioning methods that can be employed to collect the location or related information of a terminal device as ground truth data. These positioning methods can be considered candidates for ground truth collection modes. These positioning methods can be classified according to their application scenario or implementation. When positioning methods are classified by application scenario, there are several outdoor positioning methods and indoor positioning methods.

[0075] Outdoor positioning methods may include satellite-based positioning methods such as the Global Positioning System (GPS), Galileo satellite navigation system, GLONAS, and Beidou Navigation Satellite System (BDS), as well as base station positioning methods such as Location Based Services (LBS). The principles of satellite-based positioning methods may be based on triangulation, multilateral positioning, etc. Indoor positioning methods may include Wi-Fi positioning, radio frequency identification (RFID) positioning, infrared positioning, ultrasonic positioning, Bluetooth positioning, inertial navigation positioning, ultra-wideband (UWB) positioning, visible light positioning, geomagnetically matched positioning, and visual positioning. The principles of these positioning methods may be based on proximity detection, centrouding, triangulation, multilateral positioning, fingerprinting, dead reckoning, etc.

[0076] When positioning methods are classified by their implementation, there may be several RAT-dependent and RAT-independent positioning methods.

[0077] RAT-dependent positioning schemes refer to positioning schemes based on radio access technology and may therefore include one or more RAN entities. Examples of RAT-dependent positioning schemes include those based on OTDOA, Uplink Time Difference of Arrival (UTDOA), Downlink Angle-of-Departure (DL-AoD), Uplink Angles of Arrival (UL-AoA), and multi-RTT. RAT-independent positioning schemes refer to positioning schemes independent of radio access technology and may therefore not include one or more RAN entities. Examples of RAT-independent positioning schemes may include those based on satellite positioning.

[0078] Exemplary embodiments related to model trigger condition indication and model description are described below.

[0079] Model trigger condition directives can be used implicitly to determine how field data is implicitly collected. In some embodiments, the model trigger condition directive may indicate at least one of the following trigger conditions for triggering an AI / ML model, including a first trigger condition related to power saving factors, a second trigger condition related to physical environment factors, a third trigger condition related to communication network factors, or a fourth trigger condition related to positioning latency factors. Each of these four factors may include one or more subcategories.

[0080] Under the first trigger condition, the AI / ML model may be triggered for power-saving reasons. In the case of RAT-dependent positioning, the average method is currently triangulation, which requires the transmission and / or reception of reference signaling. Meanwhile, additional monitoring and calculation of UE power consumption is unavoidable. Therefore, it is reasonable to trigger the AI / ML model to infer the UE position. In some embodiments, there may be several subcategories of power-saving factors, including the disabling of Sounding Reference Signal (SRS) transmission, and the disabling of Positioning Reference Signal (PRS), including monitoring, measuring, and reporting of reference signals.

[0081] In the second trigger condition, since positioning accuracy is greatly influenced by the physical environment, the AI / ML model can be triggered by physical environmental factors. In some embodiments, there may be several subcategories of physical environmental factors, including indoor scenarios (which determine the probability of Loss of Service routes) and poor reception performance of reference signaling.

[0082] In a third trigger condition, the AI / ML model may be triggered due to communication network factors. In some examples, as shown in exemplary scenario 302 in Figure 3B, it is assumed that the positioning method is triangulation-based and the position (xn,yn) of terminal device 110 is determined through the observed arrival time difference between three network devices 120-1, 120-2, and 120-3 whose positions (x1,y1), (x2,y2), and (x3,y3) are known. If there is a network synchronization error between the network devices involved, an error in UL / DL measurement may occur in the terminal device, which may result in inaccurate positioning. In this case, the AI / ML model may be triggered. In addition to network synchronization errors, communication network factors may include other subcategories such as network device-terminal device synchronization errors, network device position errors, and poor UL hearing.

[0083] In the fourth trigger condition, the AI / ML model can be triggered by several latency factors. Latency factors may include subcategories such as sparse reference signal (RS) resources and long measurement reporting periods.

[0084] In some embodiments, the model trigger condition indication may include placeholders for each of the trigger conditions (or factors) described above. The positioning scheme may include multiple entities in the current RAT-dependent method, e.g., terminal devices, network devices, and LMFs. The AI / ML model for direct or assisted positioning is very likely to be triggered by one of the entities and, at the same time, by one of the factors dominated by either entity. Using placeholders for each factor can help indicate two or more factors to the first communication device 201 for data acquisition, helping to determine whether RAT-dependent or RAN-independent positioning can be used to acquire efficient and reliable field data and improve the reliability of monitoring. In some embodiments, each bit of the placeholder may represent whether the corresponding trigger condition is an incentive to trigger the AI / ML model.

[0085] Figure 4A shows a format 400 of an example of a model trigger condition instruction according to some embodiments of the present disclosure. As shown, if a first trigger condition 410 related to a power saving factor is 1, reliable field data for model monitoring can still be collected with additional power consumption. If a second trigger condition 420 related to a physical environment factor or a third trigger condition 430 related to a communication network factor is 1, reliable field data for model monitoring can still be collected by other mechanisms such as a RAT-independent method. If a fourth trigger condition 440 related to a latency factor or a network factor or a third trigger condition 430 related to a communication network factor is 1, reliable field data for model monitoring can still be collected on demand. In some examples, the model trigger condition instruction may include other bits, such as bits indicating trigger conditions related to other factors (other than the four factors described above). Such a model trigger condition instruction can indicate trigger conditions for an AI / ML model by combining multiple factors. That is, the model trigger condition instruction may indicate that an AI / ML model can be triggered by two or more factors.

[0086] In some embodiments, one or more trigger conditions are shown (e.g., an AI / ML model is triggered by a power-saving factor), but not all trigger conditions are shown. In such cases, the first communication device 201 may determine a target ground truth acquisition mode that enables field data acquisition without considering the limitations of these conditions (e.g., ignoring power-saving reasons during positioning). It will be understood that the target ground truth acquisition mode may be determined in any other way with respect to the shown trigger conditions.

[0087] In some embodiments, if all trigger conditions are indicated, for example, if the model trigger condition instruction is 1111, indicating that currently, reliable field data for monitoring the AI / ML model cannot be collected by non-AI means, the first communication device 201 may send a request to at least one third communication device from which field data is collected for model monitoring to deactivate or suspend the AI / ML model for direct or assisted positioning of the terminal device. It is reasonable to deactivate or suspend this AI / ML model because model monitoring of this AI / ML model is inefficient and its performance cannot be guaranteed. Direct or assisted positioning may be performed by switching to a new AI / ML model. In some embodiments, the first communication device 201 may send a request to replace it with another method for monitoring the AI / ML model.

[0088] In some embodiments, if no trigger conditions are indicated, for example, if the model trigger condition indication is 0000, representing confidence that field data for monitoring the AI / ML model can currently be collected by any ground truth collection mode, then the priority of the current positioning method, such as OTDOA, multi-RTT, or GNSS, can be predefined for selecting a ground truth collection mode for ground truth data collection. The first communication device 201 can select a target ground truth collection mode from a plurality of ground truth collection mode candidates based on the priority of each of the plurality of ground truth collection mode candidates, and each ground truth collection mode candidate is associated with at least one of the first, second, third, or fourth trigger conditions. For example, if all methods are available, the first communication device 201 may prioritize using the OTDOA positioning method to obtain ground truth labels for the AI / ML model.

[0089] In some embodiments, the model trigger condition indication may indicate the activation or deactivation of a RAT-dependent or RAT-independent ground truth acquisition mode. For example, one bit in the model trigger condition indication may be used to indicate whether a particular positioning method (classified as RAT-dependent or RAT-independent) can be activated to collect field data for monitoring an AI / ML model about positioning. In addition, in some embodiments, the confidence level of the RAT-dependent positioning may also be provided in association with the activation / deactivation indicator. Such indications are simple and have low overhead. The first communication device 201 can determine the target ground truth acquisition mode as a ground truth acquisition mode based on such a model trigger condition indication, whether RAT-dependent or RAT-independent.

[0090] In some embodiments, one or more of the first, second, third, or fourth trigger conditions may each include multiple subcategories of trigger conditions. In this case, the model trigger condition instruction may be designed to specifically indicate at least one of the multiple subcategories of trigger conditions for triggering the AI / ML model. In some embodiments, the model trigger condition instruction may include a bitmap relating to a particular trigger condition. A predefined table may be used to indicate a specific trigger condition that triggers the AI / ML model toward positioning. For example, one or more bits that are the most significant bits (MSB) may represent a category (e.g., the four trigger conditions described above), and one or more bits that are the least significant bits (LSB) may represent a specific subcategory of trigger conditions.

[0091] Figure 4B shows format 402, an example of model trigger condition indication according to some embodiments of the present disclosure. As shown, two MSB bits 412, 422 are used to indicate four trigger conditions related to power saving factors, physical environment factors, communication network factors, and positioning latency factors. Two LSB bits 432, 442 are used to indicate specific trigger conditions related to the corresponding trigger conditions indicated by the two MSB bits. In this way, specific conditions can be indicated to help determine how the first communication device collects field data for model monitoring. It will be understood that any other format may be defined to indicate specific conditions.

[0092] In some embodiments, there may be various opportunities for the first communication device 201 to receive trigger conditions that trigger an AI / ML model for direct or assisted positioning. In some embodiments, if the monitored AI / ML model is also deployed in the first communication device 201 for inference, the model trigger condition instruction may be associated with the transfer of the AI / ML model, for example, tracked using AI / ML model transfer. For example, additional bits of the model trigger condition instruction may be attached to the AI / ML model to indicate why this model is needed. The method of transmitting the trigger condition depends on the method of model transfer. In some embodiments, the model trigger condition instruction may be associated with the reception of an activation request that activates the AI / ML model deployed in the first communication device 201, for example, tracked using an AI / ML model activation / switching indicator. Thus, the method of transmitting the trigger condition depends on a model activation method, for example, Downlink Control Information (DCI), Media Access Control Element (MAC CE), or Radio Resource Control (RRC) signaling.

[0093] In some embodiments, model trigger condition instructions may be associated with the reception of AI / ML model monitoring requests, which may be tracked using associated support signaling or configured AI / ML model monitoring, including both aperiodic and periodic model monitoring. Figure 5A shows an exemplary model monitoring opportunity pattern 500, i.e., aperiodic model monitoring, where aperiodic opportunities 510 and 520 for model monitoring may be indicated by support signaling 1 and support signaling 2, respectively, attached to the model trigger condition instruction. Figure 5B shows an exemplary model monitoring opportunity pattern 502, i.e., periodic model monitoring, where periodic opportunities 512, 522, and 532 for model monitoring may be indicated by a single support signaling attached to the model trigger condition instruction.

[0094] As an alternative to or addition to model trigger condition indications, the first communication device 201 receives a model description to determine a target ground truth acquisition mode for field data acquisition. The model description is used to configure or transfer an AI / ML model for inference and generally includes a configuration profile associated with the AI / ML model. In embodiments of this disclosure, the model description further includes attribute information for ground truth acquisition to explicitly indicate the target ground truth acquisition mode of the AI / ML model. That is, the target ground truth acquisition mode for acquiring field data may also be considered an attribute of the AI / ML model, included in the model description, and shown together with the model description to a specific entity, namely the first communication device 201. In some embodiments, the first communication device 201 may be the entity into which the AI / ML model is deployed for inference. In some embodiments, the first communication device 201 may receive a model description from a second communication device 202 in a model transfer of the AI / ML model, which may be tracked, for example, using an AI / ML model transfer. For example, additional bits of model trigger condition indications may be attached to the AI / ML model to indicate why this model is needed.

[0095] As an example, the model description can be as shown in Table 1 below. [Table 1]

[0096] Once the target ground truth acquisition mode is determined and ground truth data has been acquired, the first communication device 201 may use the acquired ground truth data to perform AI / ML model monitoring. Entities other than the first communication device 201 may also be involved in ground truth acquisition and model monitoring.

[0097] In some embodiments, other information may be included in the model description signaling. Specifically, the model description may further include attribute information of a model monitoring mode, such as an input-based model monitoring mode or an output-based model monitoring mode. In response to a model monitoring mode, the first communication device 201 may perform monitoring of the AI / ML model using the collected ground truth data based on the model monitoring mode.

[0098] In some embodiments, in addition to entities for performing model monitoring (i.e., the first communication device 201), other entities may also require conditions for triggering the AI / ML model or attribute information for ground truth collection, including entities for model inference such as a fourth communication device where the AI / ML model is deployed for inference. Wherever the entities for model monitoring reside, the fourth communication device where the AI / ML model is deployed for inference can always be informed of the reason why the AI / ML model for positioning is triggered. In such embodiments, the second communication device 202 may also transmit model trigger condition instructions and / or model descriptions to the fourth communication device. In some embodiments, other entities, such as entities for model training, model selection, updating, deactivation, selection, switching, fallback, etc., may also be provided with model trigger condition instructions and / or model descriptions.

[0099] In some embodiments, an entity for model monitoring (i.e., a first communication device 201) is assumed to simply point to a location where a comparison between model inference and ground truth is performed and no decision is involved.

[0100] As mentioned above, there are several use cases for improving positioning accuracy using AI / ML. In some embodiments, if the first communication device 201 is a network device or location management function (LMF) on which an AI / ML model is deployed for direct or assisted positioning of a terminal device, the first communication device 201 can send a location service request related to the terminal device to the AMF (e.g., AMF150).

[0101] In some embodiments, in the first use case of UE-based positioning using a UE-side model, model inference is performed entirely on the terminal device side, so that wherever the entities for model monitoring are located, the terminal device (e.g., terminal device 110 in a communication environment 100) can always be notified of the reason why the AI / ML model for positioning is triggered, for example, by receiving model trigger condition instructions and / or model descriptions. In the case of direct AI / ML positioning, ground truth data in field data, i.e., UE location, can be supplied from RAN-dependent and / or RAN-independent architectures when triggered. In some embodiments, according to the model trigger conditions, terminal device 110 can determine whether to request any location service (e.g., positioning or delivery of support data) from the service provider AMF 140, if RAT-dependent positioning is considered. In some embodiments, according to the model trigger conditions, terminal device 110 can determine whether to initiate a procedure (e.g., request location-related information from a specific hardware module) if RAT-independent positioning is available. In some embodiments, AI / ML-assisted AI / ML positioning may, according to model trigger conditions, convert ground truth data in field data, such as Los / NLoS indicators, RSRP, etc., from UE-based positioning results or measurements from other UL / DL reference signaling.

[0102] In some embodiments, a second use case of UE-assisted / LMF-based positioning using a UE-side model is that model inference is performed entirely on the terminal device side. Therefore, regardless of where the entities for model monitoring are located, the terminal device (e.g., terminal device 110 in a communication environment 100) can always be notified of the reason why the AI / ML model for positioning is triggered, for example, by receiving model trigger condition instructions and / or model descriptions. Ground truth data in field data, such as Los / NLoS indicators, RSRP, etc., can be converted from the results of UE-based positioning or from measurements from other UL / DL reference signaling.

[0103] In some embodiments, in a fourth use case of NG-RAN node-assisted positioning using a gNB-side model (model in network device), model inference is performed entirely on the network device side, so that wherever the entities for model monitoring are located, the network device (e.g., network device 120 in communication environment 100) can always recognize why the AI / ML model for positioning is triggered from itself or any additional notification from terminal devices 110 or 130. Ground truth data in field data, such as Los / NLoS indicators, RSRP, etc., can be converted from UE-based or UE-assisted positioning results requested from the gNB side, from RAN-independent positioning results transmitted from the UE side, or from measurements from other UL / DL reference signaling.

[0104] Figure 6 shows the signaling flow 600 of an exemplary positioning procedure using the AI / ML model 122 on the network device 120 side, i.e., in the fourth use case. During operation, positioning of the terminal device 110 is triggered for various reasons. For example, several LCS entities 15 request some location service (e.g., positioning) for the terminal device 110 from the service provider AMF 140 (610a). Alternatively, the service provider AMF 140 for the terminal device 110 determines the need for some location service (e.g., to locate the terminal device 110 for an emergency call) (610b). Alternatively, the terminal device 110 requests some location service (e.g., positioning or delivery of support data) from the service provider AMF 140 at the non-access layer (NAS) level (610c). Alternatively, the network device 120 of the target terminal device 110 to be positioned requests some location service (e.g., positioning or delivery of support data) from the service provider AMF 140 (610d). AMF140 forwards the location service request to LMF130 (620).

[0105] In some embodiments, a third use case involves UE-assisted / LMF-based positioning using an LMF-side model, where model inference is performed entirely on the LMF side. Therefore, regardless of the entity for model monitoring, the LMF (e.g., LMF130 in communication environment 100) can always recognize why the AI / ML model for positioning is triggered. In the case of direct AI / ML positioning, ground truth in field data, i.e., UE location, can be sourced from RAN-dependent and / or RAN-independent architectures if triggered. In some embodiments, according to model trigger conditions, LMF130 can determine whether to request any location service (e.g., delivery of positioning or support data) from the service-providing AMF, given RAN-dependent positioning. In some embodiments, according to model trigger conditions, LMF130 can determine whether to indicate the UE to initiate a procedure (e.g., request location-related information from a specific hardware module) if RAN-independent positioning is available via a new LTE Positioning Protocol (LPP) message. In AI / ML-assisted AI / ML positioning, ground truth data in field data sources, such as Los / NLoS indicators and RSRP, can be converted from UE-reported positioning results or from measurements from other UL / DL reference signaling from network devices, according to model trigger conditions.

[0106] Figure 7 shows a signaling flow 700 of an exemplary positioning procedure using the AI / ML model 132 on the LMF 130 side, according to some further embodiments of the present disclosure. During operation, positioning of the terminal device 110 is triggered for various reasons. For example, some LCS entities 15 request some location service (e.g., positioning) for the terminal device 110 from the service provider AMF 140 (710a). Alternatively, the service provider AMF 140 for the terminal device 110 determines the need for some location service (e.g., to locate the terminal device 110 for an emergency call) (710b). Alternatively, the terminal device 110 requests some location service (e.g., positioning or delivery of support data) from the service provider AMF 140 at the non-access layer (NAS) level (710c). Alternatively, the service provider LMF 130 of the target terminal device 110 to be positioned requests some location service (e.g., positioning or distribution of support data) from the service provider AMF 140 (710d). The AMF 140 forwards the location service request to the LMF 130 (720).

[0107] In some embodiments, the fifth use case involves NG-RAN node-assisted positioning using an LMF-side model, where model inference is performed entirely on the LMF side, so that regardless of the entity for model monitoring, the LMF130 can always recognize why the AI / ML model for positioning is triggered. In the case of direct AI / ML positioning, ground truth in field data, i.e., UE location, can be supplied from RAN-dependent and / or RAN-independent architectures when triggered, and measurements are reported by service-providing network devices. In some embodiments, according to model trigger conditions, the LMF130 can determine whether to request any location service (e.g., positioning or delivery of support data) from the service-providing AMF130 when considering RAN-dependent positioning where positioning is performed by UL SRS positioning. In the case of AI / ML-assisted AI / ML positioning, according to model trigger conditions, ground truth in field data sources, e.g., Los / NLoS indicator, RSRP, etc., can be converted from the results of UE-based RAN-independent positioning or from measurements from other UL / DL reference signaling.

[0108] Several methods are provided for collecting field data using both AI / ML direct positioning and AI / ML-assisted AI / ML positioning in different use cases.

[0109] Figure 8 shows a flowchart of a communication method 800 implemented in a first communication device according to several embodiments of the present disclosure. For the purpose of discussion, the method 800 will be described from the perspective of the first communication device 201 in Figure 2.

[0110] In block 810, the first communication device 201 receives from the second communication device (for example, the second communication device 202 in Figure 2) a model trigger condition instruction or model description related to an artificial intelligence / machine learning (AI / ML) model, wherein the model trigger condition instruction indicates a trigger condition for triggering an AI / ML model applied in direct positioning or assisted positioning of the terminal device, and the model description includes attribute information for ground truth collection.

[0111] In block 820, the first communication device 201 determines the target ground truth acquisition mode of the AI / ML model based on a model trigger condition instruction or model description.

[0112] In block 830, the first communication device 201 collects ground truth data for monitoring the AI / ML model based on the target ground truth acquisition mode.

[0113] In some exemplary embodiments, the model trigger condition instruction indicates at least one of the following trigger conditions for triggering an AI / ML model: a first trigger condition related to power saving factors, a second trigger condition related to physical environment factors, a third trigger condition related to communication network factors, or a fourth trigger condition related to positioning latency factors.

[0114] In some exemplary embodiments, method 800 further includes sending a request to at least one third communication device from which field data is collected for model monitoring to deactivate the AI / ML model for direct or assisted positioning of the terminal device, in accordance with the determination that all of the first, second, third, and fourth trigger conditions are met.

[0115] In some exemplary embodiments, determining a target ground truth acquisition mode includes selecting a target ground truth acquisition mode from a plurality of ground truth acquisition mode candidates based on the priority of each of the plurality of ground truth acquisition mode candidates, in accordance with the determination that none of the first, second, third, or fourth trigger conditions are indicated, wherein the ground truth acquisition mode candidates are associated with at least one of the first, second, third, or fourth trigger conditions.

[0116] In some exemplary embodiments, at least one of the first, second, third, or fourth trigger conditions includes a plurality of subcategories of trigger conditions, and the model trigger condition instruction further indicates at least one of a plurality of subcategories of trigger conditions for triggering an AI / ML model applied in direct or assisted positioning of a terminal device.

[0117] In some exemplary embodiments, the model trigger condition instruction further indicates the activation or deactivation of a ground truth acquisition mode based on radio access technology (RAT)-dependent or RAT-independent positioning. In some exemplary embodiments, determining the target ground truth acquisition mode includes determining the target ground truth acquisition mode as a ground truth acquisition mode based on RAT-dependent or RAT-independent positioning, based on the model trigger condition instruction.

[0118] In some exemplary embodiments, the AI / ML model is deployed in a first communication device. In some exemplary embodiments, receiving a model trigger condition instruction includes receiving a model trigger condition instruction from a second communication device in at least one of the following: a model transfer of the AI / ML model, an activation request to activate the AI / ML model deployed in the first communication device, or a monitoring request for the AI / ML model.

[0119] In some exemplary embodiments, the AI / ML model is deployed in a first communication device, and the model description further includes attribute information for the model monitoring mode. In some exemplary embodiments, receiving the model description includes receiving the model description from a second communication device in the model transfer of the AI / ML model. In some exemplary embodiments, method 800 further includes performing monitoring of the AI / ML model using collected ground truth data based on the model monitoring mode.

[0120] In some exemplary embodiments, the first communication device is a network device or location management function (LMF) on which an AI / ML model is deployed for direct or assisted positioning of the terminal device. In some exemplary embodiments, method 800 further includes transmitting a location service request related to the terminal device to an access and mobility management function (AMF).

[0121] Figure 9 shows a flowchart of a communication method 900 implemented in a second communication device according to several embodiments of the present disclosure. For the purpose of discussion, the method 900 will be described from the perspective of the second communication device 202 in Figure 2.

[0122] In block 910, the second communication device 202 transmits a model trigger condition instruction or model description related to an artificial intelligence / machine learning (AI / ML) model to the first communication device (e.g., the first communication device 201 in Figure 2), where the model trigger condition instruction indicates a trigger condition for triggering the AI / ML model to be applied in direct or assisted positioning of the terminal device, and the model description includes attribute information for ground truth collection. The first communication device is configured to monitor the AI / ML model, and the target ground truth collection mode for collecting ground truth data of the AI / ML model is determined based on the model trigger condition instruction or model description.

[0123] In some exemplary embodiments, method 900 further includes transmitting model trigger condition instructions or model descriptions to a fourth communication device on which the AI / ML model is deployed.

[0124] In some exemplary embodiments, the model trigger condition instruction indicates at least one of the following trigger conditions for triggering the AI / ML model: a first trigger condition related to power saving factors, a second trigger condition related to physical environment factors, a third trigger condition related to communication network factors, or a fourth trigger condition related to positioning latency factors.

[0125] In some exemplary embodiments, at least one of the first, second, third, or fourth trigger conditions includes a plurality of subcategories of trigger conditions, and the model trigger condition instruction further indicates at least one of a plurality of subcategories of trigger conditions for triggering an AI / ML model applied in direct or assisted positioning of a terminal device.

[0126] In some exemplary embodiments, the model trigger condition indication further indicates the activation or deactivation of a ground truth acquisition mode based on radio access technology (RAT)-dependent or RAT-independent positioning.

[0127] In some exemplary embodiments, the AI / ML model is deployed in a first communication device. In some exemplary embodiments, sending a model trigger condition instruction includes sending a model trigger condition instruction to the first communication device in at least one of the following: a model transfer of the AI / ML model, an activation request to activate the AI / ML model deployed in the first communication device, or a monitoring request for the AI / ML model.

[0128] In some exemplary embodiments, the AI / ML model is deployed on a first communication device. In some exemplary embodiments, transmitting the model description includes transmitting the model description to the first communication device in the model transfer of the AI / ML model. In some exemplary embodiments, a model monitoring mode for the AI / ML model is further inserted into the model description.

[0129] Figure 10 is a simplified block diagram of device 1000 suitable for carrying out embodiments of the present disclosure. Device 1000 can be considered a further exemplary embodiment of any of the devices shown in Figures 1A and 2.

[0130] As shown in the figure, device 1000 includes a processor 1010, a memory 1020 coupled to the processor 1010, a suitable transceiver 1040 coupled to the processor 1010, and a communication interface coupled to the transceiver 1040. The memory 1010 stores at least a portion of program 1030. The transceiver 1040 may be for bidirectional or unidirectional communication depending on the requirements. The transceiver 1040 may include at least one of a transmitter 1042 and a receiver 1044. The transmitter 1042 and receiver 1044 may be functional modules or physical entities. The transceiver 1040 has at least one antenna to assist in communication, but in practice, the access node referred to in this application may have several antennas. The communication interface may represent any interface necessary for communication with other network elements, such as the X2 / Xn interface for bidirectional communication between eNBs / gNBs, the S1 / NG interface for communication between Mobility Management Entities (MMEs) / Access and Mobility Management Functions (AMFs) / SGWs / UPFs and eNBs / gNBs, the Un interface for communication between eNBs / gNBs and Relay Nodes (RNs), or the Uu interface for communication between eNBs / gNBs and terminal devices.

[0131] Program 1030 is assumed to include program instructions, which, when executed by the associated processor 1010, enable device 1000 to operate according to embodiments of the present disclosure, as discussed herein with reference to Figures 2 to 9. Embodiments of the present disclosure may be implemented by computer software, hardware, or a combination of software and hardware that can be executed by the processor 1010 of device 1000. The processor 1010 may be configured to implement various embodiments of the present disclosure. Furthermore, a combination of the processor 1010 and memory 1020 may form processing means 1050 adapted to implement various embodiments of the present disclosure.

[0132] Memory 1020 can be any type suitable for a local technology network and, in non-limiting examples, may be implemented using any suitable data storage technology such as non-temporary computer-readable storage media, semiconductor-based memory devices, magnetic memory devices and systems, optical memory devices and systems, fixed memory and removable memory. Although only one memory 1020 is shown within device 1000, several physically separate memory modules may be present within device 1000. Processor 1010 can be any type suitable for a local technology network and, in non-limiting examples, may include one or more processors based on general-purpose computers, dedicated computers, microprocessors, digital signal processors (DSPs), and multi-core processor architectures. Device 1000 may have multiple processors, for example, application-specific integrated circuit chips that are temporally dependent on a clock synchronized with the main processor.

[0133] According to embodiments of the present disclosure, a first communication device is provided comprising a circuit. The circuit is configured to receive from a second communication device a model trigger condition instruction or model description relating to an artificial intelligence / machine learning (AI / ML) model, wherein the model trigger condition instruction indicates a trigger condition for triggering an AI / ML model applied in direct or assisted positioning of a terminal device, and the model description includes attribute information for ground truth collection, and to determine a target ground truth collection mode for the AI / ML model based on the model trigger condition instruction or model description, and to collect ground truth data to monitor the AI / ML model based on the target ground truth collection mode. According to embodiments of the present disclosure, the circuit may be configured to perform any method performed by the first communication device as described above.

[0134] According to embodiments of the present disclosure, a second communication device is provided comprising a circuit. The circuit is configured to transmit a model trigger condition instruction or model description related to an artificial intelligence / machine learning (AI / ML) model to a first communication device, the model trigger condition instruction indicating a trigger condition for triggering the AI / ML model to be applied in direct or assisted positioning of a terminal device, the model description including attribute information for ground truth collection, the first communication device is configured to monitor the AI / ML model, and a target ground truth collection mode for collecting ground truth data of the AI / ML model is determined based on the model trigger condition instruction or model description. According to embodiments of the present disclosure, the circuit may be configured to perform any method carried out by the second communication device as described above.

[0135] As used herein, the term “circuit” may refer to a hardware circuit and / or a combination of a hardware circuit and software. For example, a circuit may be a combination of an analog and / or digital hardware circuit and software / firmware. As a further example, a circuit may be any part of a hardware processor having software including a digital signal processor, software and memory, which work together to cause a device such as a terminal device or network device to perform various functions. As yet another example, a circuit may be a hardware circuit and / or processor, such as a microprocessor or a part of a microprocessor, which requires software / firmware for operation, although the software may not be present when not required for operation. As used herein, the term “circuit” also includes embodiments of a hardware circuit or processor or a part of a hardware circuit or processor and its (or their) accompanying software and / or firmware.

[0136] In summary, embodiments of this disclosure provide the following aspects:

[0137] In one embodiment, a first communication device is proposed, which includes a processor, which is configured to receive from a second communication device a model trigger condition instruction or model description relating to an artificial intelligence / machine learning (AI / ML) model, wherein the model trigger condition instruction indicates a trigger condition for triggering an AI / ML model applied in direct positioning or assisted positioning of a terminal device, and the model description includes attribute information for ground truth collection, and to determine a target ground truth collection mode for the AI / ML model based on the model trigger condition instruction or model description, and to collect ground truth data to monitor the AI / ML model based on the target ground truth collection mode.

[0138] In some embodiments, the model trigger condition instruction indicates at least one of the following trigger conditions for triggering the AI / ML model: a first trigger condition related to power saving factors, a second trigger condition related to physical environment factors, a third trigger condition related to communication network factors, or a fourth trigger condition related to positioning latency factors.

[0139] In some embodiments, the processor is further configured to cause a first communication device to send a request to at least one third communication device from which field data is collected for model monitoring to deactivate the AI / ML model for direct or assisted positioning of the terminal device, based on a determination that all of the first, second, third, and fourth trigger conditions are indicated to the first communication device.

[0140] In some embodiments, the processor is further configured to cause a first communication device to select a ground truth acquisition mode from a plurality of ground truth acquisition mode candidates based on the priority of each of the plurality of ground truth acquisition mode candidates, according to a determination that none of the first, second, third, or fourth trigger conditions are indicated, and the ground truth acquisition mode candidates are associated with at least one of the first, second, third, or fourth trigger conditions.

[0141] In some embodiments, at least one of the first, second, third, or fourth trigger conditions includes a plurality of subcategories of trigger conditions, and the model trigger condition instruction further indicates at least one of a plurality of subcategories of trigger conditions for triggering an AI / ML model applied in direct or assisted positioning of a terminal device.

[0142] In some embodiments, the model trigger condition instruction further indicates enabling or disabling a ground truth acquisition mode based on radio access technology (RAT)-dependent or RAT-independent positioning, and the processor is configured to cause a first communication device to determine, based on the model trigger condition instruction, the target ground truth acquisition mode as a ground truth acquisition mode based on RAT-dependent or RAT-independent positioning.

[0143] In some embodiments, the AI / ML model is deployed in a first communication device, and the processor is configured to cause the first communication device to receive a model trigger instruction condition from a second communication device in at least one of the following: a model transfer of the AI / ML model, an activation request to activate the AI / ML model deployed in the first communication device, or a monitoring request for the AI / ML model.

[0144] In some embodiments, the AI / ML model is deployed to a first communication device, the model description further includes attribute information for a model monitoring mode, the processor is configured to cause the first communication device to receive the model description from a second communication device in the model transfer of the AI / ML model, and the processor is further configured to cause the first communication device to perform monitoring of the AI / ML model using ground truth data collected based on the model monitoring mode.

[0145] In some embodiments, the first communication device is a network device or location management function (LMF) on which an AI / ML model is deployed for direct or assisted positioning of a terminal device, and the processor is further configured to cause the first communication device to transmit location service requests related to the terminal device to an access and mobility management function (AMF).

[0146] In one embodiment, a second communication device is proposed, which includes a processor configured to cause the second communication device to transmit a model trigger condition instruction or model description related to an artificial intelligence / machine learning (AI / ML) model to a first communication device, wherein the model trigger condition instruction indicates a trigger condition for triggering an AI / ML model applied in direct or assisted positioning of a terminal device, and the model description includes attribute information for ground truth collection, the first communication device is configured to monitor the AI / ML model, and a target ground truth collection mode for collecting ground truth data of the AI / ML model is determined based on the model trigger condition instruction or model description.

[0147] In some embodiments, the processor is further configured to cause a second communication device to transmit model trigger condition instructions or model descriptions to a fourth communication device on which the AI / ML model is deployed.

[0148] In some embodiments, the model trigger condition instruction indicates at least one of the following trigger conditions for triggering the AI / ML model: a first trigger condition related to power saving factors, a second trigger condition related to physical environment factors, a third trigger condition related to communication network factors, or a fourth trigger condition related to positioning latency factors.

[0149] In some embodiments, at least one of the first, second, third, or fourth trigger conditions includes a plurality of subcategories of trigger conditions, and the model trigger condition instruction further indicates at least one of a plurality of subcategories of trigger conditions for triggering an AI / ML model applied in direct or assisted positioning of a terminal device.

[0150] In some embodiments, the model trigger condition indication further indicates the activation or deactivation of a ground truth acquisition mode based on radio access technology (RAT)-dependent or RAT-independent positioning.

[0151] In some embodiments, the AI / ML model is deployed to a first communication device, and the processor is configured to cause a second communication device to send a model trigger condition instruction to the first communication device in the form of at least one of the following: model transfer of the AI / ML model, activation request to activate the AI / ML model deployed in the first communication device, or monitoring request for the AI / ML model.

[0152] In some embodiments, the AI / ML model is deployed to a first communication device, the processor is configured to cause a second communication device to transmit the model description to the first communication device in the model transfer of the AI / ML model, and a model monitoring mode for the AI / ML model is further inserted into the model description.

[0153] In one embodiment, the first communication device includes at least one processor and at least one memory coupled to the at least one processor for storing instructions, wherein when an instruction is executed by the at least one processor, the device causes the device to perform the method performed by the first communication device described above.

[0154] In one embodiment, the second communication device includes at least one processor and at least one memory coupled to the at least one processor for storing instructions, wherein when an instruction is executed by the at least one processor, the device causes the device to perform the method performed by the second communication device described above.

[0155] In one embodiment, a computer-readable medium storing instructions, wherein, when executed on at least one processor, the instructions cause at least one processor to perform the method performed by the first communication device described above.

[0156] In one embodiment, a computer-readable medium storing instructions, wherein, when executed on at least one processor, the instructions cause at least one processor to perform the method performed by the second communication device described above.

[0157] In one embodiment, a computer program including instructions, the instructions, when executed on at least one processor, cause at least one processor to perform the method performed by the first communication device described above.

[0158] In one embodiment, a computer program including instructions, the instructions, when executed on at least one processor, cause at least one processor to perform the method performed by the second communication device described above.

[0159] In general, various embodiments of the present disclosure may be implemented in hardware or dedicated circuitry, software, logic, or any combination thereof. Some embodiments may be implemented in hardware, while others may be implemented in firmware or software that can be executed by a controller, microprocessor, or other computing device. Various embodiments of the present disclosure are illustrated and described using block diagrams, flowcharts, or any other graphical representation, but it should be understood that the blocks, apparatus, systems, techniques, or methods described herein may be implemented, in non-limiting examples, in hardware, software, firmware, dedicated circuitry or logic, general-purpose hardware or controllers or other computing devices, or any combination thereof.

[0160] This disclosure also provides at least one computer program product tangibly stored on a non-temporary computer-readable storage medium. The computer program product includes computer-executable instructions, such as those contained in a program module, which are executed on a device on a target real or virtual processor to perform the processes or methods described above with reference to Figures 1 to 10. Generally, a program module includes routines, programs, libraries, objects, classes, components, data structures, etc., that perform a particular task or implement a particular abstract data type. The functionality of a program module may be combined or divided among program modules as needed in various embodiments. The machine-executable instructions of a program module may be executed on a local device or a distributed device. In a distributed device, the program module may reside on both local and remote storage media.

[0161] Program code for performing the methods of this disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a dedicated computer, or other programmable data processing device, so that when executed by the processor or controller, the program code performs functions / operations specified in flowcharts and / or block diagrams. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0162] The above program code may be embodied on a machine-readable medium, which may be any tangible medium capable of storing or storing a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. The machine-readable medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. More specific examples of machine-readable storage media include one or more wires, portable computer diskettes, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, compact disc read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or electrical connections having any suitable combination thereof.

[0163] Furthermore, although the operations are shown in a specific order, this should not be understood as requiring that such operations be performed in a specific or sequential order, or that all shown operations be performed, in order to achieve the desired result. In certain circumstances, multitasking and parallel processing may be advantageous. Similarly, details of several specific embodiments are included in the above description, but these should not be interpreted as limitations on the scope of this disclosure, but rather as descriptions of features that may be specific to a particular embodiment. Certain features described in relation to a separate embodiment may also be implemented in combination in a single embodiment. Conversely, various features described in relation to a single embodiment may be implemented separately or in any suitable partial combination in multiple embodiments.

[0164] While this disclosure has been described using terminology specific to structural features and / or methodological behavior, it should be understood that this disclosure, as defined by the attached claims, is not necessarily limited to the specific features or behaviors described above. Rather, the specific features and behaviors described above are disclosed as exemplary forms for implementing the claims.

Claims

1. A first communication device, Including a processor, the processor provides the first communication device, Receiving a model trigger condition instruction or model description related to an artificial intelligence / machine learning (AI / ML) model from a second communication device, wherein the model trigger condition instruction indicates a trigger condition for triggering the AI / ML model applied in direct positioning or assisted positioning of the terminal device, and the model description includes attribute information for ground truth collection, Based on the aforementioned model trigger condition instruction or the aforementioned model description, the target ground truth acquisition mode of the AI / ML model is determined. Based on the aforementioned target ground truth acquisition mode, ground truth data is collected for monitoring the AI / ML model. A first communication device configured to perform the following action.

2. The aforementioned model trigger condition instruction is the following trigger condition for triggering the AI / ML model, namely, The first trigger condition related to power saving factors, A second trigger condition related to physical environmental factors, A third trigger condition related to communication network factors, or A fourth trigger condition related to positioning latency factors The device according to claim 1, which shows at least one of the following.

3. The processor provides the first communication device with In response to the determination that all of the first trigger condition, the second trigger condition, the third trigger condition, and the fourth trigger condition are met, The device according to claim 2, further configured to cause at least one third communication device from which field data is collected for model monitoring to transmit a request to deactivate the AI / ML model for direct or assisted positioning of the terminal device.

4. The processor provides the first communication device with In response to the determination that none of the first trigger condition, the second trigger condition, the third trigger condition, or the fourth trigger condition are met, The device according to claim 2, which further comprises selecting a target ground truth acquisition mode from a plurality of ground truth acquisition mode candidates based on the priority of each of the plurality of ground truth acquisition mode candidates, wherein the ground truth acquisition mode candidates are further configured to perform the selection related to at least one of the first trigger condition, the second trigger condition, the third trigger condition, or the fourth trigger condition.

5. At least one of the first trigger condition, the second trigger condition, the third trigger condition, or the fourth trigger condition includes a plurality of subcategories of trigger conditions, The device according to claim 2, wherein the model trigger condition instruction further indicates at least one of the plurality of subcategories of trigger conditions for triggering the AI / ML model applied in direct positioning or assisted positioning of the terminal device.

6. The aforementioned model trigger condition indication further indicates the activation or deactivation of a ground truth acquisition mode based on radio access technology (RAT) dependent positioning or RAT-independent positioning. The processor provides the first communication device with The device according to claim 1, configured to determine the target ground truth acquisition mode as a ground truth acquisition mode based on RAT-dependent positioning or RAT-independent positioning, based on the model trigger condition instruction.

7. The AI / ML model is deployed to the first communication device, and the processor is deployed to the first communication device. From the second communication device mentioned above, the following: In the model transfer of the aforementioned AI / ML model, In an activation request for activating the AI / ML model deployed in the first communication device, or In the monitoring request for the AI / ML model The device according to claim 1, wherein at least one of the devices is configured to receive the model trigger condition instruction.

8. The AI / ML model is deployed to the first communication device, the model description further includes attribute information for model monitoring mode, and the processor deploys to the first communication device. The second communication device is configured to receive the model description in the model transfer of the AI / ML model, The processor provides the first communication device with The device according to claim 1, further configured to perform monitoring of the AI / ML model using the ground truth data collected based on the model monitoring mode.

9. The first communication device is a network device or location management function (LMF) on which the AI / ML model is deployed for direct or assisted positioning of the terminal device, and the processor provides the first communication device with The device according to claim 1, further configured to cause the Access and Mobility Management Function (AMF) to transmit location service requests related to the terminal device.

10. A second communication device, Including a processor, the processor provides the second communication device, The first communication device is configured to transmit a model trigger condition instruction or model description related to an artificial intelligence / machine learning (AI / ML) model, wherein the model trigger condition instruction indicates a trigger condition for triggering the AI / ML model applied in direct positioning or assisted positioning of the terminal device, and the model description includes attribute information for ground truth collection. The first communication device is configured to monitor the AI / ML model, and the second communication device determines a target ground truth acquisition mode for collecting ground truth data of the AI / ML model based on the model trigger condition indication or the model description.

11. The processor connects to the second communication device, The device according to claim 10, further configured to transmit the model trigger condition instruction or the model description to a fourth communication device on which the AI / ML model is deployed.

12. The aforementioned model trigger condition instruction is the following trigger condition for triggering the AI / ML model, namely, The first trigger condition related to power saving factors, A second trigger condition related to physical environmental factors, A third trigger condition related to communication network factors, or A fourth trigger condition related to positioning latency factors The device according to claim 10, which shows at least one of the following.

13. At least one of the first trigger condition, the second trigger condition, the third trigger condition, or the fourth trigger condition includes a plurality of subcategories of trigger conditions, The device according to claim 12, wherein the model trigger condition instruction further indicates at least one of the plurality of subcategories of trigger conditions for triggering the AI / ML model applied in direct positioning or assisted positioning of the terminal device.

14. The device according to claim 10, wherein the model trigger condition indication further indicates enabling or disabling a ground truth acquisition mode based on radio access technology (RAT) dependent positioning or RAT-independent positioning.

15. The AI / ML model is deployed to the first communication device, and the processor is deployed to the second communication device. The following is provided to the first communication device: In the model transfer of the aforementioned AI / ML model, In an activation request for activating the AI / ML model deployed in the first communication device, or In the monitoring request for the AI / ML model The device according to claim 10, wherein at least one of the devices is configured to transmit the model trigger condition instruction.

16. The AI / ML model is deployed to the first communication device, and the processor is deployed to the second communication device. The AI / ML model transfer is configured to transmit the model description to the first communication device. The device according to claim 10, wherein the model monitoring mode for the AI / ML model is further inserted into the model description.

17. A method of communication, The first communication device receives a model trigger condition instruction or model description related to an artificial intelligence / machine learning (AI / ML) model from a second communication device, wherein the model trigger condition instruction indicates a trigger condition for triggering the AI / ML model applied in direct positioning or assisted positioning of the terminal device, and the model description includes attribute information for ground truth collection. Based on the aforementioned model trigger condition instruction or the aforementioned model description, the target ground truth acquisition mode of the AI / ML model is determined. Based on the aforementioned target ground truth acquisition mode, ground truth data is collected for monitoring the AI / ML model. A communication method that includes this.

18. A method of communication, The second communication device transmits to the first communication device a model trigger condition instruction or model description related to an artificial intelligence / machine learning (AI / ML) model, wherein the model trigger condition instruction indicates a trigger condition for triggering the AI / ML model applied in direct positioning or assisted positioning of the terminal device, and the model description includes attribute information for ground truth collection. Includes, A communication method wherein the first communication device is configured to monitor the AI / ML model, and a target ground truth acquisition mode for acquiring ground truth data of the AI / ML model is determined based on the model trigger condition instruction or the model description.

19. A computer-readable medium storing instructions, wherein, when executed by at least one processor, the instructions cause the at least one processor to execute the method according to claim 17 or the method according to claim 18.

Images