Method and apparatus for NLOS identification in ai / ML model selection and NR positioning estimation enhancement

Abstract

The present application discloses a method for artificial intelligence (AI) / machine learning (ML) model selection for non-line-of-sight (NLOS) identification in new radio (NR) positioning estimation enhancement, performed by a communication network system, including flexible selection between an AI / ML-based positioning method, an AI / ML assisted positioning method, and a non-AI / ML positioning method, for example, the AI / ML-based positioning method, the AI / ML assisted positioning method, and the non-AI / ML positioning method. According to the characteristics of the positioning method and the positioning quality service (QoS) requirements of the application level; lOS / NLOS AI / ML identification model parameters are activated and / or transmitted to a node responsible for providing LOS / NLOS indication based on mapping indication from an application layer to a positioning method; and / or transmitting the same or similar LOS / NOLS classification AI / ML model parameters to the node responsible for providing the LOS / NLOS indication, so as to improve the indication credibility of the node to the final UE position estimation.

Description

Technical Field

[0001] The present application relates to the field of wireless communication systems, and more specifically, to methods and devices for artificial intelligence (AI) / machine learning (ML) model selection and non-light of sight (NLOS) identification in new radio (NR) positioning estimation enhancement. For example, the present application relates to enhancing current NR radio access network (RAN) signaling and procedures to support AI / ML-based positioning methods, thereby improving the accuracy of user equipment (UE) position estimation. Background Art

[0002] At the 3GPP RAN#94 meeting, a study item (SI) on AI / ML for NR air interface was approved, with the main goal of exploring the enhancement of air interface functions to improve support for AI / ML-based algorithms, thereby improving performance and / or reducing complexity and system overhead. The goal of this study item is to focus on a number of carefully selected use cases, such as CSI feedback, beam management, and positioning accuracy enhancement, to determine the application areas of AI / ML in improving air interface functional performance.

[0003] Regarding AI positioning accuracy enhancement, it was agreed at the RAN1#109 meeting to study two AI-based positioning enhancement solutions, namely direct AI positioning and AI-assisted positioning methods. In the direct AI positioning solution, the AI ​​model will directly replace the existing positioning method and be used for the final UE position estimation. In contrast, in the AI-assisted positioning solution, the AI ​​model is used to assist the existing positioning method to improve the accuracy of UE positioning estimation. At the RAN1#109 meeting, some companies reported that although the accuracy of the AI-assisted positioning method is slightly lower than that of the direct AI positioning method, it is compatible with existing non-AI-based positioning methods, has good model generalization capabilities, and has lower system overhead and complexity. In addition, this method is particularly suitable for indoor scenarios with severe non-light of sight (NLOS) environments, because it relies on AI / ML models to improve the accuracy of NLOS recognition, thereby improving those technologies that rely on light of sight (LOS) measurements for UE positioning estimation. For this type of positioning technology, the 3GPP Release 17 standard has introduced the UE or positioning transmission / reception point (TRP) to provide LOS / NLOS indications to the node responsible for UE position estimation to help improve the positioning estimate. However, the standard does not clearly specify how to obtain LOS / NLOS indications. Without a clear definition, different nodes responsible for providing LOS / NLOS indications may use different estimation methods, and due to different estimation errors, LOS / NLOS identification may eventually become unreliable, affecting the final UE position estimation accuracy.

[0004] In addition to the above factors, according to the application-level positioning accuracy requirements provided by TR 38.857, it can be observed that different NR positioning application categories have different levels of positioning QoS accuracy requirements. For example, some applications have low positioning accuracy requirements, which can be met using only traditional non-AI positioning methods. Some applications require medium accuracy, and AI-assisted methods can guarantee the accuracy requirements. Some applications require higher accuracy, and only AI-based positioning methods can meet the requirements. Therefore, in the R18 positioning enhancement research, it is very likely that both AI-based and AI-assisted methods will be supported on the basis of existing positioning solutions to adapt to the differences in positioning QoS and accuracy requirements of different positioning applications. In this case, it is still unclear how to choose between AI / ML-based positioning methods, AI / ML-assisted positioning methods, and non-AI positioning methods, because each of the above methods has different characteristics, and there are also differences in the QoS requirements and accuracy requirements of positioning applications. In addition, it is not clear how to determine the NLOS / LOS state when the AI / ML-assisted method is selected as the preferred method for UE positioning estimation. The present application provides a mechanism for flexibly selecting a priority positioning method, which is based on the QoS requirements of the positioning application and the characteristics of the positioning estimation method, and introduces a mechanism for reliably obtaining LOS / NLOS indications to improve the accuracy and credibility of positioning estimation. Summary of the invention

[0005] The purpose of the present application is to propose a method and apparatus for non-light of sight (NLOS) identification in artificial intelligence (AI) / machine learning (ML) model selection and new radio (NR) positioning estimation enhancement.

[0006] In a first aspect of the present application, a method for artificial intelligence (AI) / machine learning (ML) model selection for non-light of sight (NLOS) identification in new radio (NR) positioning estimation enhancement is performed by a communication network system, including flexibly selecting between AI / ML-based positioning methods, AI / ML-assisted positioning methods and non-AI / ML positioning methods, based on the characteristics of the positioning method and the positioning quality of service (QoS) requirements at the application layer; activating and / or transmitting LOS / NLOS AI / ML identification model parameters to a node responsible for providing LOS / NLOS indications, based on a mapping indication from the application layer to the positioning method; and / or transmitting the same or similar LOS / NOLS classification AI / ML model parameters to the node responsible for providing the LOS / NLOS indication to improve the credibility of the node's indication of the final UE position estimate.

[0007] In a second aspect of the present application, a communication network system includes a memory, a transceiver, and a processor coupled to the memory and the transceiver. The processor is configured to execute the above method.

[0008] In a third aspect of the present application, a non-transitory machine-readable storage medium stores instructions, which, when executed by a computer, cause the computer to perform the above method.

[0009] In a fourth aspect of the present application, a chip includes a processor configured to call and run a computer program stored in a memory so that a device having the chip installed therein executes the above method.

[0010] In a fifth aspect of the present application, a computer readable storage medium stores a computer program, which enables a computer to execute the above method.

[0011] In a sixth aspect of the present application, a computer program product comprises a computer program, which enables a computer to execute the above method.

[0012] In a seventh aspect of the present application, a computer program enables a computer to execute the above method. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the embodiments of the present application or related technologies, the following briefly introduces the drawings to be described in the embodiments. Obviously, the drawings are only some embodiments of the present application, and ordinary technicians in this field can obtain other drawings based on these drawings without paying any price.

[0014] Figure 1 It is a block diagram of one or more user equipment (UE) communicating with a first and a second RAN node in a communication network system according to an embodiment of the present application.

[0015] Figure 2 It is a flowchart describing a method for artificial intelligence (AI) / machine learning (ML) model selection according to an embodiment of the present application, which is used for non-light of sight (NLOS) identification in new radio (NR) positioning estimation enhancement, and is executed by a communication network system.

[0016] Figure 3 It is an example of the NR positioning system architecture according to an embodiment of the present application, showing different AI positioning model positions and collaboration options.

[0017] Figure 4 This is an example of a positioning method directly based on AI / ML and an AI / ML-assisted positioning method according to an embodiment of the present application.

[0018] Figure 5 This is an example of an AI / ML-assisted positioning method for LOS / NLOS identification according to an embodiment of the present application.

[0019] Figure 6 This is a possible signaling example for exchanging positioning method selection mapping or indication according to an embodiment of the present application.

[0020] Fig. 7A This is a signaling example for exchanging mapping or indication of AI-assisted measurement configuration and LOS / NLOS identification according to an embodiment of the present application.

[0021] Figure 7B This is a signaling example for exchanging AI-assisted / AI-based measurement configuration, LOS / NLOS identification and / or model parameter transmission according to an embodiment of the present application.

[0022] Figure 8 This is a signaling example for exchanging AI-assisted / AI-based measurement configuration, LOS / NLOS identification and / or model parameter transmission according to an embodiment of the present application.

[0023] Fig. 9 This is an example of signaling and processes related to AI / ML model parameter request or transmission according to an embodiment of the present application, assuming that the model is located in a location server (AMF / LMF / 5GC).

[0024] Fig.10 is a system block diagram for wireless communication according to an embodiment of the present application. DETAILED DESCRIPTION

[0025] The technical content, structural features, objectives and effects of the present application are described in detail below in conjunction with the accompanying drawings. Specifically, the terms in the embodiments of the present application are only used to describe the purpose of specific embodiments, and are not intended to limit the present application.

[0026] Figure 1 It is illustrated that in some embodiments, a communication network system 40 is composed of a target device 10, a RAN node (e.g., a gNB) 20, and a network node (e.g., an LMF or an AMF, a location server (LCS)) 30, for communicating in the communication network system, and the example is based on an embodiment of the present application. The communication network system 40 includes a target device 10, a RAN node 20, and a network node 30. The target device 10 may include a memory 12, a transceiver 13, and a processor 11, wherein the processor 11 is connected to the memory 12 and the transceiver 13. The RAN node (e.g., a gNB) 20 may include a memory 22, a transceiver 23, and a processor 21, wherein the processor 21 is connected to the memory 22 and the transceiver 23. The network node (e.g., an LMF or an AMF, a location server (LCS)) 30 may include a memory 32, a transceiver 33, and a processor 31, wherein the processor 31 is connected to the memory 32 and the transceiver 33. The processor 11, 21 or 31 may be configured to perform the proposed functions, processes and / or methods described in this specification. The radio interface protocol layer may be implemented on the processor 11, 21 or 31. The memory 12, 22 or 32 is operatively connected to the processor 11, 21 or 31 and stores a variety of information to support the operation of the processor 11, 21 or 31. The transceiver 13, 23 or 33 is operatively connected to the processor 11, 21 or 31 for sending and / or receiving wireless signals.

[0027] The processor 11, 21 or 31 may include an application specific integrated circuit (ASIC), other chipsets, logic circuits and / or data processing devices. The memory 12, 22 or 32 may include a read only memory (ROM), a random access memory (RAM), flash memory, a memory card, a storage medium and / or other storage devices. The transceiver 13, 23 or 33 may include a baseband circuit for processing radio frequency signals. When these embodiments are implemented in software, the techniques described herein may be implemented by modules (e.g., programs, functions, etc.) that perform the functions described herein. These modules may be stored in the memory 12, 22 or 32 and executed by the processor 11, 21 or 31. The memory 12, 22 or 32 may be integrated inside the processor 11, 21 or 31, or may be located outside the processor 11, 21 or 31, in which case they may be connected to the processor 11, 21 or 31 by various communication methods, such as methods known in the art. In some embodiments, the communication network system 40 is configured to perform the following functions: flexibly select between AI / ML-based positioning methods, AI / ML-assisted positioning methods and non-AI / ML positioning methods, and optimize the positioning service quality (QoS) requirements based on the characteristics of the positioning method and the application layer; activate and / or transmit the LOS / NLOS AI / ML recognition model parameters according to the mapping indication from the application layer to the positioning method, and transmit them to the node responsible for providing the LOS / NLOS indication; transmit the same or similar LOS / NLOS classification AI / ML model parameters to the node responsible for providing the LOS / NLOS indication to improve the node's credibility in the indication in the final UE position estimation.

[0028] Figure 2 It is a flowchart describing a method for artificial intelligence (AI) / machine learning (ML) model selection according to an embodiment of the present application, which is used for non-light of sight (NLOS) identification in new radio (NR) positioning estimation enhancement, and is executed by a communication network system. Figure 3The present invention describes an AI / ML model selection method performed by a communication network system in some embodiments for NLOS identification in NR positioning estimation enhancement, the method comprising the following steps: flexibly selecting between different positioning methods, including AI / ML-based positioning methods, AI / ML-assisted positioning methods and non-AI / ML positioning methods, the selection being based on the characteristics of the positioning method and the positioning quality of service (QoS) requirements at the application layer; activating and / or transmitting LOS / NLOS AI / ML identification model parameters to a node responsible for providing LOS / NLOS indications, based on a mapping indication from the application layer to the positioning method; transmitting the same or similar LOS / NLOS classification AI / ML model parameters to a node responsible for providing LOS / NLOS indications to improve the credibility of the indications of the node in the final UE position estimation.

[0029] At the 3GPP RAN#94 meeting, a study item (SI) on AI / ML for NR air interface was approved. The main goal of this study item is to explore the advantages of enhancing air interface features to improve support for AI / ML-based algorithms to improve system performance and / or reduce complexity and system overhead. The goal of this study item is to focus on several carefully selected use cases, such as CSI feedback, beam management, and positioning accuracy enhancement, to identify areas where AI / ML can improve air interface functional performance. In addition, the project is also committed to studying a general AI / ML framework, including the functional requirements of the AI / ML architecture and the specification adjustments required to introduce AI / ML technology in 3GPP specifications to improve air interface performance. Regarding NR AI positioning accuracy enhancement, it was agreed at the RAN1#109 meeting to further study the scheme of AI-based positioning models at different locations, for example, the AI ​​model is located on a single side or a single entity (UE, gNB or LMF), where training and inference are done on the same side; the AI ​​model is partially deployed on one side, where training and inference are done on the network or UE side, but additional signaling or program enhancements are required between the two sides, which may be compatible with the existing signaling framework. In addition, a consensus was reached at the meeting to include different levels of network and UE collaboration solutions into further research, including: no collaboration mode; signaling-based collaboration mode (not involving model transmission); signaling-based collaboration mode (involving model transmission). These different levels of collaboration solutions are shown in Figure 3 , and the meeting also agreed to further study the generalization capabilities of AI / ML models in NR positioning.

[0030] In addition to the above, a consensus was reached at the 3GPP RAN1#109 meeting to further study the following AI / ML-based positioning accuracy enhancement use cases, including:

[0031] Direct AI / ML based positioning: This method uses AI / ML based algorithms to replace existing positioning methods and refers to new measurement types and / or measurement reports to estimate the UE position. These measurement types include, for example, amplitude, time of arrival (TOA), angle of arrival (AoA), channel impulse response (CIR), beam index (Beam Index), transmission reception point index (TRP Index), signal power, and signal reflection order. This data can be collected from multiple transmission and reception points (TRP) and used for AI to calculate UE position estimates (see examples). Figure 4 ).

[0032] Indirect or AI / ML-assisted positioning: This method uses AI technology to assist existing positioning methods to improve the accuracy of UE position estimation. AI extracts intermediate features through measurement data and / or measurement reports, and then estimates the final UE position based on the extracted intermediate features. Measurement data that can be used for intermediate feature extraction include, for example, downlink reference signal timing difference (DL-RSTD), downlink transmission angle (DL-angle of departure, DL-AoD), uplink relative time of arrival (UL-RTOA), uplink sounding reference signal-reference signal received power (SRS-RSRP), downlink positioning reference signal RSRP (DL positioningreference signal RSRP, DL PRS-RSRP), UE transmit-receive time difference (UE Rx-Tx TimeDifference), gNB transmit-receive time difference (gNB Rx-Tx Time Difference), uplink angle of arrival (UL-AOA) or transmission angle (AoD), zenith angle of arrival (ZOA) (calculated for each signal path). Then, AI estimates the final UE position through these extracted intermediate features (see example Figure 4 and Figure 5 ).

[0033] In addition to the above considerations, according to the application-level positioning accuracy requirements provided by TR 38.857 (Table 1), which shows the maximum allowable positioning error for each positioning method, it can be observed that different NR positioning application categories have different levels of positioning QoS accuracy requirements. For example, some positioning applications have low positioning accuracy requirements, which can be met using only traditional non-AI positioning methods. Some positioning applications require medium accuracy, which can be achieved by AI-assisted positioning methods. While some other positioning applications require higher accuracy, which can only be met by AI-based positioning methods. Therefore, in the R18 positioning enhancement research, it is likely that both AI-based and AI-assisted methods will be supported on the basis of existing positioning solutions to meet the differences in positioning QoS and accuracy requirements of different positioning applications.

[0034] Based on the above considerations, it is not clear how to choose between AI / ML-based positioning methods, AI / ML-assisted positioning methods, and non-AI positioning methods, because these methods each have different characteristics, and different positioning applications also have different requirements for QoS and accuracy. In addition, when the AI / ML-assisted method is selected as the preferred method for UE positioning estimation, how to determine the NLOS / LOS state is also an unresolved issue. The present application provides a mechanism for flexibly selecting a priority positioning method based on the QoS requirements of the positioning application and the characteristics of the positioning estimation method, and introduces a mechanism for reliably obtaining LOS / NLOS indications to improve the accuracy and credibility of NR positioning estimates.

[0035] Table 1: NR positioning application categories.

[0036]

[0037] In some embodiments of the present application, in order to solve the problems related to the selection of positioning models, a positioning method selection mechanism based on QoS or QoS assistance is proposed. This mechanism allows flexible selection between AI-based, AI-assisted and non-AI-based positioning methods, based on the characteristics of these positioning methods (such as advantages and disadvantages) and the positioning QoS requirements at the application level, which refer to the recommendations of TR 38.857. In addition, in order to help the UE determine the NLOS / LOS state, or provide reliable NLOS / LOS indications when selecting an AI-assisted positioning method for positioning estimation to improve the positioning accuracy estimation, the method also proposes to use application-level indications and / or mapping information as trigger conditions for providing, requesting or transmitting appropriate AI / ML classification model parameters. These parameters can assist the entity responsible for providing LOS / NLOS indications to perform NLOS / LOS identification or classification at the UE or TRPs end, thereby optimizing the reliability and accuracy of positioning estimates.

[0038] The present application provides a positioning method for AI / ML model selection, which is used for non-line-of-sight recognition in NR positioning estimation enhancement.

[0039] The main advantages of this approach include:

[0040] 1. The new method introduces a flexible selection concept, which allows users to choose between AI / ML-based positioning methods, AI / ML-assisted positioning methods, and non-AI / ML positioning methods, based on the characteristics of these positioning methods and the positioning QoS requirements at the application level (refer to TR 38.857). This mechanism facilitates the flexible integration of AI / ML with existing traditional positioning methods.

[0041] 2. The new method introduces an activation and / or transmission mechanism of the LOS / NLOS AI / ML recognition model, which is used to transmit the recognition model to the node (UE or TRP) responsible for providing LOS / NLOS indications, and the transmission is based on the mapping indication of the application layer to the positioning method. This mechanism can solve the problem in the prior art, that is, how the node responsible for providing LOS / NLOS indications determines or estimates the NLOS / LOS state.

[0042] 3. The new method proposes the concept of unified transmission of LOS / NLOS classification AI / ML model parameters, that is, transmitting the same or similar AI / ML recognition model parameters to the node (UE or TRP) responsible for providing LOS / NLOS indications. This mechanism helps to unify the trust of the nodes that calculate the final position on the LOS / NLOS indications and solve the reliability problem of LOS / NLOS indications, thereby improving the accuracy of the final position estimation.

[0043] Example 1: Detailed description:

[0044] In order to solve the problems related to positioning model selection, the present application proposes a positioning method selection mechanism based on QoS or QoS assistance. This mechanism allows flexible selection between AI-based, AI-assisted and non-AI-based positioning methods, based on the characteristics of these positioning methods (such as advantages and disadvantages) and the positioning QoS requirements at the application level, which refer to the recommendations of TR38.857. In this method, the mapping or indication of application QoS level and positioning method can be exchanged between the entity / node / device requesting UE position estimation (such as UE, target device or location server) and the entity / node responsible for the final position estimation (such as target device, location server, location management entity (LMF) or signaling reference source). The purpose of this exchange is to help the target entity / node select an appropriate positioning method between AI / ML-based positioning methods, AI / ML-assisted positioning methods or non-AI positioning methods, based on the exchanged indication information or positioning QoS mapping information provided by the upper layer (see Figure 6 ). In addition, in order to solve the problem of how UE determines the NLOS / LOS state or how to provide reliable NLOS / LOS indication to improve the positioning accuracy estimation when the AI-assisted positioning method is selected as the positioning estimation method, the method also proposes to use application-level indications and / or mapping information as trigger conditions to provide, request or transmit the same AI / ML classification model parameters to all entities / nodes responsible for providing LOS / NLOS indications (such as target device UE or reference source TRPs). This mechanism helps to assist the target entity in classification during the NLOS / LOS identification process and optimize the reliability of LOS / NLOS identification (see Fig. 7A and Figure 7B The detailed steps of this method will be further described below (see Figure 8 ).

[0045] The target device (UE) or the location server receives indication information including a mapping relationship between positioning application QoS requirements and a preferred positioning method from the application layer (see Table 2), or receives indication information of a preferred positioning method derived from the mapping relationship (see Table 3).

[0046] The target device (UE) and the location server may exchange requests and / or responses with each other for transmitting information related to the exchange of the indication information.

[0047] The target device (UE) and the location server may directly exchange requests and / or responses carrying the indication information via an LTE location protocol (LLP) signaling procedure.

[0048] The target device (UE) and the location server can also exchange information indirectly through the reference source (gNB) through the LTE location protocol (LLP) signaling procedure, which is carried out through NR positioning protocol a (NRPPa) signaling and NR radio interface signaling.

[0049] The target device (UE) and the location server may also exchange requests and / or responses carrying the indication information to the reference source (gNB) through NR positioning protocol a (NR positioning protocola, NRPPa) signaling and NR radio interface signaling.

[0050] The reference source (gNB) can exchange the mapping indication between the positioning application QoS requirements and the preferred positioning method, or the indication information of the preferred positioning method derived from the mapping, to other adjacent reference sources (gNB) participating in the position estimation through the Xn or X2 interface.

[0051] The target device (UE), location server and / or reference source (gNB) can use the mapping indication between the positioning application QoS requirements and the preferred positioning method, or the indication information of the preferred positioning method derived based on the mapping, to select and operate in the following ways: locally select between AI-based, AI-assisted and non-AI-based positioning methods; request or provide activation / deactivation of a certain positioning method; request or transmit trained AI / ML model parameters for use in AI-based or AI-assisted positioning methods; transmit measurement configurations related to AI-based, AI-assisted or existing positioning methods to achieve data sharing between different nodes.

[0052] After receiving the exchanged mapping indication information or response, the target device (UE) and the core network (CN) entity (eg, a 5G location server) may select a method according to Table 4.

[0053] Trained AI / ML model parameters can be categorized based on AI-based or AI-assisted methods as follows:

[0054] For AI-based methods, the trained AI / ML model parameters may include neural network (NN) model parameters, such as: NN input and output types, number of input / output layers, number of hidden layers, number of neurons in hidden layers, activation function, initial weight, learning rate. Input types may include Channel Impulse Response (CIR), Power Delay Profile (PDP), Layer 1 Reference Signal Received Power (L1-RSRP) (based on existing positioning measurement data over time). Output types may include final UE position, UE coordinates (X, Y, Z).

[0055] For AI-assisted methods, the trained AI / ML model parameters may include simple AI / ML classifier model parameters, such as: input and output types of the classifier, initial weights, learning rate parameters, binary variables (0 or 1) (for binary classification), enumeration variables (for multi-level classification). Input types may include: Channel Impulse Response (CIR), Power Delay Profile (PDP), Layer 1 Reference Signal Received Power (L1-RSRP), Channel Frequency Response in Frequency Domain (CFR). Output types may include: LOS / NLOS probabilities for different channel paths, Downlink Reference Signal Time Difference (DL-RSTD), UE Transmit-Receive Time Difference (UE Rx-Tx timedifference), PRS Reference Signal Received Power (PRS RSRPP), and / or Downlink Transmit Angle (DL-AoD / ZoD) or Arrival Angle (DL-AoA / ZoA).

[0056] For signaling and procedures related to requesting or transmitting trained AI-based or AI-assisted model parameters, in some embodiments, it is assumed that the trained AI / ML positioning model is stored in the reference source (gNB). In other embodiments, it is assumed that the trained AI / ML positioning model is stored in the location server (LMF, AMF or 5GC).

[0057] Table 2: Direct mapping between positioning application QoS levels and preferred positioning methods.

[0058]

[0059] Table 3: Indication of preferred positioning method based on application-level QoS.

[0060] Application Category Application level indication 0.5(0.2)m to <3m 1 ≥1.5m to <5m 0 >1.5m to <62.5m This field is not configured in the signaling message.

[0061] Table 4: Actions of the target device / location server / reference source entity upon receiving the indication / mapping.

[0062]

[0063]

[0064] Embodiment 2: Signaling and process of requesting or transmitting AI / ML model parameters (assuming that the trained AI model is stored at the reference source or gNB end).

[0065] When the trained AI / ML positioning model is stored in the reference source or gNB, the signaling and procedures for exchanging AI / ML model parameters for AI-assisted or AI-based positioning methods are described in detail in Figure 8 , the specific process is as follows:

[0066] The target device (UE) or the location server receives from the application layer an indication of a mapping relationship between the positioning application QoS requirements and the preferred positioning method, or indication information of the preferred positioning method derived based on the mapping. The target device (UE) sends the indication or mapping information to the location server through a NAS signaling or LPP signaling exchange process.

[0067] Based on the received indication or mapping information, the location server (LMF / AMF) forwards the positioning measurement request and / or the AI / AI-assisted model parameter transmission request to the reference source or gNB. This transmission process is completed through the NR-PPa signaling exchange process.

[0068] The reference source or gNB performs the following operations based on the instructions provided by the location server (LMF / AMF): transmits AI / AI-assisted model parameters; initiates a measurement request or provides a measurement gap configuration to support a specific AI / AI-assisted model, and sends it to the target device (UE) via NR / LTE wireless signaling.

[0069] The target device (UE) performs AI / AI-assisted measurements using the received model parameters, and then sends the measurement results and / or LOS / NLOS indication information to the location server (LMF / AMF). The data transmission method is directly transmitted through NAS or LLP signaling, or indirectly through NR / LTE wireless signaling and NR-PPa signaling to complete the final UE position estimation.

[0070] Example 3: Signaling and process for requesting or transmitting AI / ML model parameters (assuming that the trained AI model is stored in the location server AMF / LMF / 5GC).

[0071] When the trained AI / ML positioning model is stored on the location server (AMF / LMF / 5GC), the signaling and process for exchanging AI / ML model parameters supporting AI-assisted or AI-based positioning methods are detailed in Fig. 9 , the specific process is as follows:

[0072] The target device (UE) or the location server receives, from the application layer, indication information including a mapping relationship between the positioning application QoS requirement and the preferred positioning method, or indication information of the preferred positioning method derived based on the mapping.

[0073] The target device (UE) sends the indication or mapping information to the location server through the NAS signaling or LPP signaling exchange process.

[0074] Based on the received indication or mapping information, the location server (LMF / AMF) starts the AI / AI-assisted model parameter processing process and forwards the positioning measurement request to the target device (UE) to support the transmitted base AI / AI-assisted model. The transmission method of the measurement request includes direct transmission through NAS or LLP signaling, or indirect transmission through NR-PPa signaling and NR / LTE wireless signaling.

[0075] The target device (UE) performs AI / AI-assisted measurements using the received model parameters and completes the measurement task as requested. Subsequently, the target device (UE) sends the measurement results and / or LOS / NLOS indication information to the location server (LMF / AMF) for final UE position estimation. Data transmission methods include direct transmission through NAS or LLP signaling, or indirect transmission through NR / LTE wireless signaling and NR-PPa signaling.

[0076] Fig.10 7 is a block diagram of an exemplary wireless communication system 700 according to an embodiment of the present application. The embodiments described herein may be implemented into a system using any appropriately configured hardware and / or software. Fig.10 A system 700 is shown, which includes a radio frequency (RF) circuit 710, a baseband circuit 720, an application circuit 730, a memory / storage device 740, a display 750, a camera 760, a sensor 770, and an input / output (I / O) interface 780, which are coupled to each other at least as shown. The application circuit 730 may include circuits, such as but not limited to one or more single-core or multi-core processors. The processor may include any combination of general-purpose processors and special-purpose processors, such as a graphics processor, an application processor. The processor may be coupled to a memory / storage device and configured to execute instructions stored in the memory / storage device to enable various applications and / or operating systems to run on the system.

[0077] While the present disclosure has been described in connection with what is considered to be the most practical and preferred embodiment, it is to be understood that the present disclosure is not limited to the disclosed embodiment, but is intended to cover various arrangements made without departing from the scope of the broadest interpretation of the appended claims.

Claims

1. A method for artificial intelligence AI / machine learning ML model selection for non-line-of-sight NLOS identification in new radio NR positioning estimation enhancement, performed by a communication network system, characterized in that: include: Flexible selection between AI / ML-based positioning methods, AI / ML-assisted positioning methods, and non-AI / ML positioning methods, based on the characteristics of the positioning methods and the positioning quality service QoS requirements at the application level; Activate and / or transmit LOS / NLOS AI / ML identification model parameters to the node responsible for providing LOS / NLOS indications based on the mapping indication of the application layer to the positioning method; and / or The same or similar LOS / NOLS classification AI / ML model parameters are transmitted to the node responsible for providing the LOS / NLOS indication to improve the indication confidence of the node for the final UE position estimate.

2. The method according to claim 1, characterized in that The mapping or indication of application QoS levels to positioning methods is exchanged between the node requesting the UE position estimate and the node responsible for the final position estimate to assist the node in making appropriate selections between AI / ML-based positioning methods, AI / ML-assisted positioning methods, or non-AI / ML positioning methods, the selection being based on the exchanged indications or the positioning QoS mapping information provided by the upper layer.

3. The method according to claim 1 or 2, characterized in that: The application level indication and / or mapping information is used as a trigger to provide, request or transmit the same AI / ML classification model parameters to all the nodes responsible for providing the LOS / NLOS indication to assist the nodes in identifying or classifying NLOS / LOS conditions or performing the LOS / NLOS identification process.

4. The method according to any one of claims 1 to 3, characterized in that The target device or the location server receives from the application level an indication containing a mapping between positioning application QoS requirements and positioning methods, or an indication of another positioning method derived from the mapping.

5. The method according to any one of claims 1 to 4, characterized in that The target device and the location server exchange requests and / or responses carrying information indicating the exchange of relevant information.

6. The method according to any one of claims 1 to 5, characterized in that The target device and the location server directly exchange requests and / or responses carrying information indicating the exchange of relevant information through the LTE Positioning Protocol LLP signaling procedure.

7. The method according to any one of claims 1 to 6, characterized in that The target device and the location server exchange requests and / or responses that carry instructions for exchanging relevant information indirectly through the reference source via the LTE Positioning Protocol LLP signaling procedure, using the NR Positioning Protocol a NRPPa signaling and NR radio interface signaling.

8. The method according to any one of claims 1 to 7, characterized in that The target device and the location server exchange requests and / or responses carrying information indicating the exchange of relevant information to the reference source through NR Positioning Protocol a NRPPa signaling and NR radio interface signaling.

9. The method according to any one of claims 1 to 7, characterized in that Reference sources exchange indications about the mapping between positioning application QoS requirements and positioning methods, or indicate other nearby reference sources involved in position estimation based on the mapped positioning method. The exchange is carried out over the Xn or X2 interface.

10. The method according to any one of claims 1 to 9, characterized in that The target device and / or location server and / or reference source utilizes the indication of the mapping between the positioning application QoS requirements and the positioning method, or based on the mapped positioning method indication, to make a local selection between AI / ML-based positioning methods, AI / ML-assisted positioning methods and non-AI / ML positioning methods and / or request or provide activation / deactivation of positioning methods, and / or request or transmit AI / ML model parameters for use in AI-based positioning methods or AI-assisted positioning methods, or transmit measurement configurations related to AI-based, AI-assisted or existing positioning methods between each other.

11. The method according to any one of claims 1 to 10, characterized in that After the exchanged mapping or indication response, the target device and the core network CN entity perform method selection according to a table indicating that the information type is AI / ML-based information or AI / ML-assisted information.

12. The method according to claim 11, characterized in that The information type is the AI / ML-based information and includes neural network NN model parameters, including NN input type and output type, the number of input / output layers, the number of hidden layers, the number of neurons in the hidden layer, activation function, initial weight and learning rate.

13. The method according to claim 12, characterized in that The input types include channel impulse response CIR, power delay profile PDP, and data on the time variation of layer 1 reference signal received power L1 RSRP based on existing positioning measurements.

14. The method according to claim 12 or 13, characterized in that The output type is the final UE position or UE coordinates.

15. The method according to claim 11, characterized in that The information type is the AI / ML auxiliary information and includes AI / ML classifier model parameters, including classifier input and output types, initial weights, learning rate parameters, binary variables (0, 1) for binary classification, or enumeration variables for multi-level classification.

16. The method according to claim 15, characterized in that The input types include CIR, PDP, L1 RSRP and / or a channel frequency response CFR in the frequency domain.

17. The method according to claim 15 or 16, characterized in that The output types are LOS / NLOS probabilities and / or downlink reference signal time difference DL RSTD of different channel paths, UE reception and transmission time difference, PRS RSRPP, and downlink arrival angle / departure angle of the path or downlink arrival angle / departure angle.

18. A communication network system, characterized in that: include: Memory; Transceiver; as well as a processor connected to the memory and the transceiver; The processor is configured to execute the method according to any one of claims 1 to 17.

19. A non-transitory machine-readable storage medium, characterized in that Instructions are stored which, when executed by a computer, cause the computer to perform the method of any one of claims 1 to 17.

20. A chip, characterized in that: include: The processor is configured to call and run the computer program stored in the memory so that the device on which the chip is mounted executes the method according to any one of claims 1 to 17.

21. A computer-readable storage medium, characterized in that: A computer program is stored, the computer program causing a computer to execute the method according to any one of claims 1 to 17.

22. A computer program product, characterized in that It comprises a computer program causing a computer to execute the method according to any one of claims 1 to 17.

23. A computer program, characterized in that The computer program causes a computer to execute the method according to any one of claims 1 to 17.

Images