A method for ai / ML positioning based on dynamic TRP activation
By dynamically selecting and incorporating TRP location and channel delay profiles into AI/ML models, the method addresses performance loss and complexity issues in AI/ML positioning, achieving improved positioning accuracy.
Patent Information
- Application Number
- PCT/CN2023/110265
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2023-07-31
- Publication Date
- 2025-02-06
AI Technical Summary
Existing AI/ML positioning techniques face performance loss and increased complexity due to dynamic TRP activation, which affects model input size and signaling overhead.
The method involves dynamically selecting and activating a subset of TRPs, using their location, index, or bitmap information along with associated channel delay profiles as model input, to reduce input size while maintaining performance.
This approach enhances AI/ML positioning performance by allowing the model to learn accurate associations between channel delay profiles and TRP activation patterns, thereby reducing performance loss and complexity.
Smart Images

Figure CN2023110265_06022025_PF_FP_ABST
Abstract
Description
A METHOD FOR AI / ML POSITIONING BASED ON DYNAMIC TRP ACTIVATIONFIELD
[0001] This present disclosure relates generally to wireless communications, and more specifically, to techniques of positioning a user equipment (UE) with Artificial Intelligence (AI) / Machine Learning (ML) .BACKGROUND
[0002] The 3GPP (Third Generation Partnership Program) approved a study item on AI / ML based positioning. There are several approaches for the evaluation of model input due to different number of TRPs, which will impact the model performance, signalling overhead (including training data collection and model inference) , model complexity and computational complexity. It is very meaningful to reduce the performance loss while reducing the AI / ML complexity and signalling overhead.SUMMARY
[0003] The following presents a simplified summary of one or more aspects in order to provide a basic understanding of such aspects. This summary is not an extensive overview of all contemplated aspects, and is intended to neither identify key or critical elements of all aspects nor delineate the scope of any or all aspects. Its sole purpose is to present some concepts of one or more aspects in a simplified form as a prelude to the more detailed description that is presented later.
[0004] Typically, downlink PRS or uplink SRS signals for positioning are used to estimate channel delay profiles as AI / ML model input to obtain the intermediate metrics (e.g., time of arrival (TOA) of the channel path between TRP and UE, LOS or NLOS indicator of the channel path between TRP and UE) for AI / ML assisted positioning or obtain the UE location for direct AI / ML positioning. In order to reduce complexity and signaling overhead, part of the TRPs will be activated in a fixed or dynamic manner for positioning. Dynamic change of the TRPs can reduce the input size of the model, but will lead to poor performance, because for different UEs, channel delay profiles corresponding to different set of TRPs will be activated as the model input, and this activation mechanism is not fully provided to the model with the reduced input size. However, if the TRP pattern selected for positioning is fixed for all UEs, better performance will be achieved. Therefore, this disclosure is mainly aimed at the situation where the TRP selection is dynamic and TRP dimension of model input is reduced. This disclosure provides the dynamic TRP selection method to reduce the input size and proposes to use the active TRP information, such as TRP location, index or bitmap, together with the associated channel delay profiles, as the model input for AI / ML positioning. In this way, TRP can be dynamically activated to reduce the size of the model input while improving the performance of the model.
[0005] To the accomplishment of the foregoing and related ends, the one or more aspects comprise the features hereinafter fully described and particularly pointed out in the claims. The following description and the annexed figures set forth in detail certain illustrative features of the one or more aspects. These features are indicative, however, of but a few of the various ways in which the principles of various aspects may be employed, and this description is intended to include all such aspects and their equivalents.BRIEF DESCRIPTION OF THE DRAWINGS
[0006] Figure 1 illustrates indoor factory deployment scenario.
[0007] Figure 2 illustrates high-level process of AI / ML model training and inference based on dynamic changes of TRP.
[0008] Figure 3 illustrates the AI / ML models for positioning with TRP information as an additional model input.DETAILED DESCRIPTION
[0009] The detailed description set forth below in connection with the appended drawings is intended as a description of various configurations and is not intended to represent the only configurations in which the concepts described herein may be practiced. The detailed description includes specific details for the purpose of providing a thorough understanding of various concepts. However, it will be apparent to those skilled in the art that these concepts may be practiced without these specific details. In some instances, well known structures and components are shown in block diagram form in order to avoid obscuring such concepts.
[0010] Several aspects of telecommunication systems will now be presented with reference to various apparatus and methods. These apparatus and methods will be described in the following detailed description and illustrated in the accompanying drawings by various blocks, components, circuits, processes, algorithms, etc. (collectively referred to as “elements” ) . These elements may be implemented using electronic hardware, computer software, or any combination thereof. Whether such elements are implemented as hardware or software depends upon the particular application and design constraints imposed on the overall system.
[0011] Figure 1 shows an example of indoor factory deployment scenario. There are 18 TRPs (NTRP =18) and the location of all TRPs are fixed. UEs are random distributed in the factory. PRUs (Positioning Reference Unit) as special UEs are deployed in known location for AI / ML training data collection.
[0012] If AI / ML model is deployed in UE / PRU side, TRP will send positioning reference signal to UE / PRU. UE / PRU will estimate the channel delay profiles between TRP and UE / PRU based on received positioning reference signal. If AI / ML model is deployed in network side (TRP or LMF) , UE / PRU will send sounding reference signal to TRP. And TRP will estimate the channel delay profiles between TRP and UE / PRU based on received sounding reference signal. These measured channel delay profiles can be used as AI / ML model input. There is different measurement result for different channel path between TRP and UE / PRU.
[0013] The most complex case is that for UE / PRU, the channel delay profiles between UE / PRU and all N TRPs (TRP0, TRP1, TRP2, …, TRP15, TRP16, TRP17) are used as model input for AI / ML based positioning, which will result in high signaling overhead and complexity in phases such as data collection, model training and inference.
[0014] To reduce signaling overhead and complexity, the following approaches in 3GPP RAN1#113 is agreed. For the evaluation of AI / ML based positioning, the study of model input due to different number of TRPs include the following approaches. Proponent of each approach provide analysis for model performance, signalling overhead (including training data collection and model inference) , model complexity and computational complexity.
[0015] 1. Approach 1: Model input size stays constant as NTRP=18. The number of TRPs (N’TRP) that provide measurements to model input varies. When N’TRP < NTRP, the remaining (NTRP -N’TRP) TRPs do not provide measurements to model input, i.e., measurement value is set such that the (NTRP -N’TRP) TRPs do not affect model output.
[0016] i. Approach 1-A. The set of TRPs (N’TRP) that provide measurements is fixed.
[0017] ii. Approach 1-B. The set of TRPs (N’TRP) that provide measurements can change dynamically.
[0018] iii. Note: for Approach 1, one model is provided to cover the entire evaluation area.
[0019] 2. Approach 2: The TRP dimension of model input is equal to the number of TRPs (N’TRP) that provide measurements as model input. When N’TRP < NTRP, the remaining (NTRP -N’TRP) TRPs are ignored by the given model.
[0020] i. Approach 2-A. The set of active TRPs (N’TRP) that provide measurements is fixed.
[0021] For Approach 2-A: one model can be provided to cover the entire evaluation area, which is equivalent to deploying N’TRP TRPs in the evaluation area for positioning if ignoring the potential inference from the remaining (18 -N’TRP) TRPs.
[0022] ii. Approach 2-B. The set of active TRPs (N’TRP) that provide measurements can change dynamically.
[0023] For Approach 2-B, one model is developed to handle various patterns of active TRPs.
[0024] iii. For Approach 2, if Nmodel (Nmodel >1) models are provided to cover the entire evaluation area, the total complexity (model complexity is the summation of the Nmodel models.
[0025] The set of active TRPs (N’TRP) that provide measurements for Approach 1-A and Approach 2-A is fixed, except that the size of the TRP dimension of model input is NTRP for Approach 1-A and N’TRP and Approach 2-A, respectively. Compared to Approach 1-A, AI / ML complexity of Approach 2-Awill be reduced.
[0026] This invention mainly focusses on the Approach 1-B and Approach 2-B that the set of active TRPs (N’TRP) changes dynamically. Figure 2 shows the proposed a dynamic TRP selection method performed at UE side. The TRP dynamic changes can be performed by LMF or by UE. When dynamic TRP change is performed by UE, UE selects N’TRP TRPs for each measurement as model input according to a certain selection mechanism, N’TRP and selection method is configured by NW. Different selection mechanisms will be associated to different AI / ML models. Similarly, different number of active TRPs is also associated to different models. The selected measurements will be reported to NW if in the data collection phase or the model inference is performed at UE side, else, if model inference is performed at UE side, the model output will be reported to NW.
[0027] In the stage of model training, when dynamic TRP change is performed on LMF, the network will select N’TRP TRPs by itself, and send the selected data to the UE to reduce the signaling overhead. At the same time, the network informs UE of the selection mechanism used, so that in the model inference stage, the UE can use the same TRP selection mechanism as that of training for AI / ML based positioning.
[0028] The selection mechanism used can be based on power (RSRP / RSSI) or timing, for example, N’TRP TRPs with the strongest power or with the shortest estimated first path delay is selected for AI / ML based positioning.
[0029] For Approach 2-B, the TRP dimension of model input is reduced from NTRP to N’TRP, the loss of performance from the diversification of TRP activation patterns is much greater than the decrease in the number itself, and even leads to model failure, because it’s difficult for the model to learn accurate associations between channel delay profiles of different TRPs without information about the TRP activation pattern.
[0030] This invention proposes to use the activated TRP information, such as TRP location, index or bitmap, together with the associated channel delay profiles, as the input of models in which the model input size is reduced (e.g., Approach 2-B) for AI / ML positioning. With the TRP information, the model can more easily learn the relationship between the input N’TRP channel delay profiles and the relationship between the input and output, thus enhancing the AI / ML based positioning performance.
[0031] Figure 3 shows the AI / ML model with TRP information as model input. The set of activated TRPs can be indicated by a length of NTRP bitmap. If TRP0, TRP2 and TRP4 are activated, the bitmap is ‘000, 000, 000, 000, 010, 101’ if the least significant bit is TRP0 and ‘101, 010, 000, 000, 000, 000’ if the most significant bit is TRP0. The TRP index (e.g., 0 ~ NTRP -1) , index differential or TRP coordinates can also be used as model input. Similarly, other TRP coding formats can also be used as model input.
[0032] Additionally, a data pre-processing AI / ML model with TRP information as model input can be used to learn the TRP representations or relationship between activated TRP information, as shown in Figure 3. The pre-processing AI / ML model is much simpler than positioning AI / ML model and the effect on complexity can be ignored. The positioning AI / ML model and pre-processing AI / ML model are trained together to improve positioning performance.
[0033] The previous description is provided to enable any person skilled in the art to practice the various aspects described herein. Various modifications to these aspects will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other aspects. Thus, the claims are not intended to be limited to the aspects shown herein, but is to be accorded the full scope consistent with the language claims, wherein reference to an element in the singular is not intended to mean “one and only one” unless specifically so stated, but rather “one or more. ” The word “exemplary” is used herein to mean “serving as an example, instance, or illustration. ” Any aspect described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other aspects. Unless specifically stated otherwise, the term “some” refers to one or more. Combinations such as “at least one of A, B, or C, ” “one or more of A, B, or C, ” “at least one of A, B, and C, ” “one or more of A, B, and C, ” and “A, B, C, or any combination thereof” include any combination of A, B, and / or C, and may include multiples of A, multiples of B, or multiples of C. Specifically, combinations such as “at least one of A, B, or C, ” “one or more of A, B, or C, ” “at least one of A, B, and C, ” “one or more of A, B, and C, ” and “A, B, C, or any combination thereof” may be A only, B only, C only, A and B, A and C, B and C, or A and B and C, where any such combinations may contain one or more member or members of A, B, or C. All structural and functional equivalents to the elements of the various aspects described throughout this disclosure that are known or later come to be known to those of ordinary skill in the art are expressly incorporated herein by reference and are intended to be encompassed by the claims. Moreover, nothing disclosed herein is intended to be dedicated to the public regardless of whether such disclosure is explicitly recited in the claims. The words “module, ” “mechanism, ” “element, ” “UE, ” and the like may not be a substitute for the word “means. ” As such, no claim element is to be construed as a means plus function unless the element is expressly recited using the phrase “means for. ”
[0034] While aspects of the present disclosure have been described in conjunction with the specific embodiments thereof that are proposed as examples, alternatives, modifications, and variations to the examples may be made. Accordingly, embodiments as set forth herein are intended to be illustrative and not limiting. There are changes that may be made without departing from the scope of the claims set forth below.
Claims
1.A method of wireless communication of User Equipment (UE) , comprising,receiving, the assistance data, from NW,receiving, the reference signal, from NW,selecting, the measurements between the UE and TRPs, andreporting, the result to NW.2.The method of claim 1, wherein the UE is a UE with known position by the network.3.The method of claim 1, wherein the selecting measurements is associated to N selected TRPs, the number of selected TRPs N and the selection method are configured by NW.4.The method of claim 1 or claim 3, wherein the TRP selection is performed based on power, N TRPs with strongest power will be selected.5.The method of claim 1 or claim 3, wherein the TRP selection is performed based on timing, N TRPs with the shortest estimated first path delay will be selected.6.The method of claim 1, wherein the measurements is CIR / PDP / DP.7.The method of claim 1, wherein the result is selected measurements and associated TRP information if AI / ML model inference is performed at NW side or in the data collection phase.8.The method of claim 7, wherein the TRP information is: bitmap, index, and coordinates.9.The method of claim 7, wherein the TRP information is used as the input of the AI / ML model.10.The method of claim 1, wherein the result is the AI / ML model output when the model inference is performed at UE side.11.The method of claim 7 or claim 10, wherein AI / ML model is direct AI / ML model or AI / ML assisted model.12.The method of claim 7 or claim 10, wherein the AI / ML model includes the AI / ML model for positioning and AI / ML model for TRP information pre-processing.13.The method of claim 7 or claim 10, wherein the AI / ML model is associated with the number of TRP selected.14.The method of claim 7 or claim 10, wherein the AI / ML model is associated with the TRP selection method.15.The method of claim 10, wherein the AI / ML model output is UE location if it’s a direct AI / ML model.16.The method of claim 10, wherein the AI / ML model output is TOA or LOS / NLOS indicator with associated soft information if it’s an AI / ML assisted model.17.The method of claim 1, wherein the assistance data includes training data after TRP selection and TRP information if in the model training phase.18.The method of claim 1, wherein the reference signal is PRS.
Citation Information
Patent Citations
Measurement model based on uplink signals with reciprocity to downlink beam
US20220312151A1
Geometric dilution of precision based transmit / receive point selection for wireless positioning
US20220322277A1
Method for angle based positioning measurement and apparatus therefor
US20220330041A1
Methods for reference signal configuration in wireless systems
WO2022031974A1