Methods and apparatus for localization using artificial intelligence and machine learning (AI / ML)
By selecting and switching AI/ML models using WTRU, the performance degradation of the RAT positioning method in NLOS environments is resolved, achieving higher accuracy and more reliable location measurement and positioning.
Patent Information
- Application Number
- CN202580027526.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2024-02-14
- Filing Date
- 2025-02-14
- Publication Date
- 2026-11-17
AI Technical Summary
Existing radio access technology (RAT)-based positioning methods degrade in performance in non-line-of-sight (NLOS) environments and cannot effectively locate wireless transceiver units (WTRUs).
The WTRU sends AI/ML positioning capability information to the base station, receives positioning reference signal (PRS) configuration and auxiliary information, selects or switches AI/ML models for positioning, and uses AI/ML model sets and back-off positioning methods for location measurement and positioning.
The positioning accuracy and reliability of WTRU in NLOS environment are improved, and the ability to measure and locate positions between different cells is enhanced.
Smart Images

Figure CN123125508A_ABST
Abstract
Description
[0001] Cross-references to related applications This application claims the benefit of U.S. Provisional Patent Application No. 63 / 553,307, filed February 14, 2025, which is incorporated herein by reference. Background Technology
[0002] This application relates to the fields of communications, software, and coding, including, for example, methods, architectures, apparatus, and systems relating to the process of locating wireless transmit / receive unit (WTRU) using artificial intelligence / machine learning (AI / ML) models. More particularly, this disclosure relates to the selection and switching of AI / ML models.
[0003] Current 3GPP specifications define positioning methods that rely on Radio Access Technology (RAT). These methods require the WTRU to be in a line-of-sight (LOS) environment relative to the Transmit / Receive Point (TRP). In non-line-of-sight (NLOS) environments, the performance of RAT-based positioning methods degrades. For applicable use cases of WTRU-based positioning, exceeding this limitation would be beneficial. Summary of the Invention
[0004] In summary, in one embodiment, the WTRU can send capability information indicating that it supports AI / ML positioning to a base station associated with the first cell and based on the WTRU's movement from the second cell to the first cell. The WTRU can send a request to the base station for auxiliary information regarding Positioning Reference Unit (PRU) inference. The WTRU can receive configuration information indicating a fallback positioning method and resources associated with a Positioning Reference Signal (PRS) set. The WTRU can measure the PRS set. The WTRU can receive auxiliary information indicating the PRU's location and / or a set of measurement information associated with the PRU's location inference. The WTRU may send a report to the base station based on a request from the base station, the report indicating any of the following: (i) a first AI / ML model used at the WTRU for positioning in the second cell to be used in the first cell, (ii) fine-tuning of the first AI / ML model used at the WTRU for positioning in the second cell, (iii) a second AI / ML model to be used at the WTRU for positioning in the second cell, (iv) a back-off positioning method to be used at the WTRU for positioning in the second cell, and / or (v) inference generated based on the indicated measurement information set and other measurement information associated with the measured PRS set.
[0005] In one embodiment, the WTRU can determine its location. Based on the determined location, the WTRU can (i) select an AI / ML model from the AI / ML model set or (ii) select a back-off localization method. Each AI / ML model in the AI / ML model set can be associated with a corresponding area. The WTRU can send information to the base station indicating the selected AI / ML model or the selected back-off localization method.
[0006] In one embodiment, the WTRU may receive information indicating a first PRS configuration. The first PRS configuration may be associated with an AI / ML model. When the WTRU moves from a first cell to a second cell, it may receive information indicating a second PRS configuration associated with the second cell. Based on whether the second PRS configuration is included in the first PRS configuration, the WTRU may select (i) an AI / ML model or (ii) a fallback positioning method. The WTRU may send information associated with the selected AI / ML model or the selected fallback positioning method.
[0007] In one embodiment, the WTRU can receive information indicating a set of AI / ML models. Each AI / ML model can be associated with a corresponding timing advance (TA) value range. The WTRU can receive information indicating a timing advance (TA) value associated with a base station. The WTRU can select an AI / ML model from the set of AI / ML models based on the indicated TA value falling within the TA value range associated with the AI / ML model. The WTRU can transmit information associated with the selected AI / ML model.
[0008] In one embodiment, the WTRU may receive information indicating a first set of measurement results associated with the PRU. The WTRU can use the first set of measurement results to obtain location inference from an AI / ML model. The WTRU may then send information associated with the obtained inference to the network. Attached Figure Description
[0009] The following detailed description will be better understood when read in conjunction with the accompanying drawings, which illustrate examples of one or more embodiments of the present disclosure. However, it should be understood that the embodiments described herein are not limited to the precise arrangements and means shown in the drawings. In the drawings: Figure 1A This is a system diagram illustrating an example communication system according to one or more embodiments of the present disclosure; Figure 1B This illustrates one or more embodiments of the present disclosure that can be implemented... Figure 1A A system diagram of an example wireless transmit / receive unit (WTRU) used in the communication system shown; Figure 1C This illustrates one or more embodiments of the present disclosure that can be implemented... Figure 1A System diagram of an example radio access network (RAN) and an example core network (CN) used in the communication system shown; Figure 1D This illustrates one or more embodiments of the present disclosure that can be implemented... Figure 1A The diagram shows another example RAN and another example CN used in the communication system shown. Figure 2 This is a block diagram illustrating an example of a neural network according to one or more embodiments of the present disclosure; Figure 3 This is a block diagram illustrating an example of estimating WTRU location using an AI / ML model according to one or more embodiments of the present disclosure; Figure 4 This is a block diagram illustrating an example hierarchical structure of a positioning reference signal (PRS) configuration according to one or more embodiments of the present disclosure; Figure 5 This is a block diagram illustrating an example association between an artificial intelligence / machine learning (AI / ML) model and PRS configuration information according to one or more embodiments of this disclosure; Figure 6 This is a signaling diagram illustrating an example of communication for WTRU-based AI / ML positioning according to one or more embodiments of this disclosure; Figure 7 This is a signaling diagram illustrating an example of communication for updating the association between an AI / ML model and a PRS configuration according to one or more embodiments of this disclosure; Figure 8 This is a signaling diagram illustrating an example of communication for initiating AI / ML-based positioning according to one or more embodiments of this disclosure; Figure 9 This is a signaling diagram illustrating another example of communication for initiating AI / ML-based location-based communication according to one or more embodiments of this disclosure; Figure 10 This is a block diagram of an example AI / ML model illustrating the optimal AI / ML model for predicting WTRU location and WTRU location according to one or more embodiments of this disclosure; Figure 11 This is an example of an operation area diagram illustrating the operation areas of different AI / ML models according to one or more embodiments of this disclosure; Figure 12 This is a flowchart illustrating an example of a model switching process based on a WTRU trajectory according to one or more embodiments of this disclosure; Figure 13This is a block diagram of an example AI / ML model illustrating the prediction of WTRU location, the optimal AI / ML model for WTRU location, and the future AI / ML model according to one or more embodiments of this disclosure; Figure 14 This is a signaling diagram illustrating an example of initiating AI / ML-based location communication when an AI / ML model is unavailable, according to one or more embodiments of this disclosure; Figure 15 This is a system diagram illustrating an example of measurement forwarding between a PRU and a WTRU according to one or more embodiments of this disclosure; Figure 16 This is a signaling diagram illustrating an example of model verification following a mobility event according to one or more embodiments of this disclosure; Figure 17 This is a signaling diagram illustrating another example of model verification following a mobility event according to one or more embodiments of this disclosure; Figure 18 This is a system diagram illustrating an example of verifying measurement results using a PRU according to one or more embodiments of this disclosure; Figure 19 This is a system diagram illustrating an example of verifying measurement results using PRS according to one or more embodiments of this disclosure; Figure 20 This is a signaling diagram illustrating an example of model validation using CIR; Figure 21 This is a signaling diagram illustrating an example of truth value acquisition according to one or more embodiments of this disclosure; Figure 22 This is a signaling diagram illustrating an example of provisioning repeated truth information according to one or more embodiments of this disclosure; Figure 23 This is a signaling diagram illustrating an example of validating an AI / ML model using PRU measurement results according to one or more embodiments of this disclosure; Figure 24 This is a system diagram illustrating an example of a WTRU moving from one PRS set to another according to one or more embodiments of this disclosure; Figure 25 This is a block diagram illustrating an example configuration of the association between an AI / ML model, an SSB, and a reference according to one or more embodiments of this disclosure; Figure 26 This is a block diagram illustrating another example configuration of the association between an AI / ML model, an SSB, and a reference according to one or more embodiments of this disclosure; Figure 27 This is a signaling diagram illustrating an example determined by an AI / ML model based on a PRS region according to one or more embodiments of this disclosure; Figure 28 This is a system diagram illustrating examples of TRP and timing advance (TA) associated with an AI / ML model according to one or more embodiments of this disclosure; Figure 29 This is a system diagram illustrating an example of determining an AI / ML model due to mobility according to one or more embodiments of this disclosure; Figure 30 This is a system diagram illustrating another example of determining an AI / ML model due to mobility according to one or more embodiments of this disclosure; Figure 31 This is a system diagram illustrating another example of determining an AI / ML model due to mobility according to one or more embodiments of this disclosure; Figure 32 This is a flowchart illustrating an example of AI / ML model verification according to one or more embodiments of this disclosure; Figure 33 This is a flowchart illustrating an example of switching AI / ML models according to one or more embodiments of this disclosure; Figure 34 This is a flowchart illustrating another example of switching AI / ML models according to one or more embodiments of this disclosure; Figure 35 This is a flowchart illustrating another example of switching AI / ML models according to one or more embodiments of this disclosure; Figure 36 This is a flowchart illustrating another example of switching AI / ML models according to one or more embodiments of this disclosure; Figure 37 This is a flowchart illustrating an example process according to one or more embodiments of this disclosure; Figure 38 This is a flowchart illustrating an example process according to one or more embodiments of this disclosure; Figure 39 This is a flowchart illustrating an example process according to one or more embodiments of this disclosure; Figure 40 This is a flowchart illustrating an example process according to one or more embodiments of this disclosure; and Figure 41 This is a flowchart illustrating an example process according to one or more embodiments of the present disclosure. Detailed Implementation
[0010] In describing various embodiments of this disclosure, certain terms are used herein for convenience only and should not be construed as limiting such embodiments. In the accompanying drawings, the same reference numerals are used throughout the drawings and this description to denote the same elements.
[0011] In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments and / or examples disclosed herein. However, it will be understood that such embodiments and examples may be practiced without some or all of the specific details set forth herein. In other instances, well-known methods, processes, components, and circuits have not been described in detail so as not to obscure the following description. Furthermore, embodiments and examples not specifically described herein may be practiced in place of or in combination with the embodiments and other examples described, disclosed, or otherwise explicitly, implicitly, and / or inherently provided (collectively, the “Provided”). Although various embodiments are described and / or claimed herein in which apparatuses, systems, devices, etc., and / or any elements thereof implement operations, processes, algorithms, functions, etc., and / or any part thereof, it should be understood that any embodiment described and / or claimed herein configures any apparatus, system, device, etc., and / or any element thereof to implement any operation, process, algorithm, function, etc., and / or any part thereof.
[0012] Example Communication System The methods, apparatus, and systems provided herein are well-suited for communications involving both wired and wireless networks. Regarding Figure 1A-1D This provides an overview of various types of wireless devices and infrastructures, in which various elements of the network can utilize the methods, apparatuses, and systems provided herein, perform the methods, apparatuses, and systems provided herein, are arranged according to the methods, apparatuses, and systems provided herein, and / or are adapted and / or configured for the methods, apparatuses, and systems provided herein.
[0013] Figure 1A This is a system diagram illustrating an example communication system 100 in which one or more of the disclosed embodiments may be implemented. The communication system 100 may be a multiple access system that provides content (such as voice, data, video, messaging, broadcasting, etc.) to multiple wireless users. The communication system 100 enables multiple wireless users to access such content by sharing system resources (including wireless bandwidth). For example, the communication system 100 may employ one or more channel access methods, such as Code Division Multiple Access (CDMA), Time Division Multiple Access (TDMA), Frequency Division Multiple Access (FDMA), Orthogonal FDMA (OFDMA), Single Carrier FDMA (SC-FDMA), Zero-Tail (ZT) Unique Word (UW) Discrete Fourier Transform (DFT) Extended OFDM (ZT UW DTS-s OFDM), Unique Word OFDM (UW-OFDM), Resource Block Filtered OFDM, Filter Bank Multicarrier (FBMC), etc.
[0014] like Figure 1AAs shown, the communication system 100 may include wireless transmit / receive units (WTRUs) 102a, 102b, 102c, 102d, radio access networks (RANs) 104 / 113, core networks (CNs) 106 / 115, public switched telephone networks (PSTNs) 108, the Internet 110, and other networks 112. Although it will be appreciated, the disclosed embodiments contemplate any number of WTRUs, base stations, networks, and / or network elements. Each of the WTRUs 102a, 102b, 102c, 102d can be any type of device configured to operate and / or communicate in a wireless environment. As an example, WTRUs 102a, 102b, 102c, and 102d (any of which may be referred to as a “station” and / or “STA”) may be configured to transmit and / or receive wireless signals and may include / or be user equipment (UE), mobile stations, fixed or mobile subscriber units, subscription-based units, pagers, cellular phones, personal digital assistants (PDAs), smartphones, laptops, netbooks, personal computers, wireless sensors, hotspots or Mi-Fi devices, Internet of Things (IoT) devices, watches or other wearable devices, head-mounted displays (HMDs), vehicles, drones, medical devices and applications (e.g., remote surgery), industrial devices and applications (e.g., robots and / or other wireless devices operating in industrial and / or automated processing chain scenarios), consumer electronics devices, devices operating on commercial and / or industrial wireless networks, etc. Any of WTRUs 102a, 102b, 102c, and 102d may be interchangeably referred to as a UE.
[0015] The communication system 100 may also include base station 114a and / or base station 114b. Each of base stations 114a and 114b may be any type of device configured to wirelessly interface with at least one of WTRUs 102a, 102b, 102c, and 102d, for example, to facilitate access to one or more communication networks, such as CN 106 / 115, Internet 110, and / or Network 112. As an example, base stations 114a and 114b may be any of a base transceiver station (BTS), Node-B (NB), eNode-B (eNB), Home Node-B (HNB), Home eNode-B (HeNB), gNode-B (gNB), NR Node-B (NR NB), site controller, access point (AP), wireless router, etc. Although base stations 114a and 114b are depicted as single elements, it will be appreciated that base stations 114a and 114b may include any number of interconnected base stations and / or network elements.
[0016] Base station 114a may be part of RAN 104 / 113, which may also include other base stations and / or network elements (not shown), such as base station controllers (BSCs), radio network controllers (RNCs), relay nodes, etc. Base station 114a and / or base station 114b may be configured to transmit and / or receive radio signals on one or more carrier frequencies, which may be referred to as cells (not shown). These frequencies may be in licensed spectrum, unlicensed spectrum, or a combination of licensed and unlicensed spectrum. A cell may provide coverage for a specific geographic area for radio services, which may be relatively fixed or may change over time. A cell may be further divided into cell sectors. For example, the cell associated with base station 114a may be divided into three sectors. Therefore, in an embodiment, base station 114a may include three transceivers, i.e., one transceiver for each sector of the cell. In an embodiment, base station 114a may employ multiple-input multiple-output (MIMO) technology, and multiple transceivers may be used for each or any sector of the cell. For example, beamforming can be used to transmit and / or receive signals in a desired spatial direction.
[0017] Base stations 114a and 114b can communicate with one or more of WTRUs 102a, 102b, 102c, and 102d via air interface 116, which can be any suitable wireless communication link (e.g., radio frequency (RF), microwave, centimeter wave, millimeter wave, infrared (IR), ultraviolet (UV), visible light, etc.). Air interface 116 can be established using any suitable radio access technology (RAT).
[0018] More specifically, as noted above, the communication system 100 can be a multiple access system and can employ one or more channel access schemes, such as CDMA, TDMA, FDMA, OFDMA, SC-FDMA, etc. For example, base station 114a in RAN 104 / 113, and WTRUs 102a, 102b, and 102c can implement radio technologies such as Universal Mobile Telecommunications System (UMTS) Terrestrial Radio Access (UTRA), which can use Wideband CDMA (WCDMA) to establish the air interface 116. WCDMA may include communication protocols such as High-Speed Packet Access (HSPA) and / or Evolved HSPA (HSPA+). HSPA may include High-Speed Downlink Packet Access (HSDPA) and / or High-Speed Uplink Packet Access (HSUPA).
[0019] In the embodiment, base station 114a and WTRUs 102a, 102b, 102c can implement radio technologies such as Evolved UMTS Terrestrial Radio Access (E-UTRA), which can use Long Term Evolution (LTE) and / or Advanced LTE (LTE-A) and / or Advanced LTE Pro (LTE-A Pro) to establish air interface 116.
[0020] In the embodiments, base station 114a and WTRUs 102a, 102b, 102c can implement radio technologies such as NR radio access, which can use a new radio (NR) to establish an air interface 116.
[0021] In the embodiments, base station 114a and WTRUs 102a, 102b, and 102c can implement multiple radio access technologies. For example, base station 114a and WTRUs 102a, 102b, and 102c can jointly implement LTE radio access and NR radio access, for example, using the dual connectivity (DC) principle. Therefore, the air interface utilized by WTRUs 102a, 102b, and 102c can be characterized by multiple types of radio access technologies and / or transmissions sent to / from multiple types of base stations (e.g., eNBs and gNBs).
[0022] In the embodiments, base station 114a and WTRUs 102a, 102b, and 102c can implement the following radio technologies, such as IEEE 802.11 (i.e., Wi-Fi), IEEE 802.16 (i.e., WiMAX), CDMA2000, CDMA2000 1X, CDMA2000 EV-DO, Provisional Standard 2000 (IS-2000), Provisional Standard 95 (IS-95), Provisional Standard 856 (IS-856), Global System for Mobile Communications (GSM), Enhanced Data Rate GSM Evolution (EDGE), and GSMEDGE (GERAN).
[0023] Figure 1ABase station 114b can be, for example, a wireless router, a home Node-B, a home eNode-B, or an access point, and can utilize any suitable RAT to facilitate wireless connectivity in a local area, such as a commercial area, home, vehicle, campus, industrial facility, air corridor (e.g., for drone use), road, etc. In embodiments, base station 114b and WTRUs 102c, 102d can implement radio technologies such as IEEE 802.11 to establish a wireless local area network (WLAN). In embodiments, base station 114b and WTRUs 102c, 102d can implement radio technologies such as IEEE 802.15 to establish a wireless personal area network (WPAN). In embodiments, base station 114b and WTRUs 102c, 102d can utilize cellular-based RATs (e.g., WCDMA, CDMA2000, GSM, LTE, LTE-A, LTE-A Pro, NR, etc.) to establish any of the following: small cell, pico cell, or femtocell. Figure 1A As shown, base station 114b may have a direct connection to Internet 110. Therefore, base station 114b may not be required to access Internet 110 via CN 106 / 115.
[0024] RAN 104 / 113 can communicate with CN 106 / 115, which can be any type of network configured to provide voice, data, application, and / or Voice over Internet Protocol (VoIP) services to one or more of WTRUs 102a, 102b, 102c, and 102d. Data may have different Quality of Service (QoS) requirements, such as different throughput requirements, latency requirements, error tolerance requirements, reliability requirements, data throughput requirements, mobility requirements, etc. CN 106 / 115 can provide call control, billing services, location-based services, prepaid calling, internet connectivity, video distribution, etc., and / or perform advanced security functions, such as user authentication. Although... Figure 1A Although not shown, it will be understood that RAN104 / 113 and / or CN 106 / 115 can communicate directly or indirectly with other RANs that use the same RAT as or a different RAT than RAN 104 / 113. For example, in addition to being connected to RAN 104 / 113, which can utilize NR radio technology, CN106 / 115 can also communicate with another RAN (not shown) that uses any of the following radio technologies: GSM, UMTS, CDMA 2000, WiMAX, E-UTRA, or Wi-Fi.
[0025] CN 106 / 115 may also act as a gateway for WTRU 102a, 102b, 102c, 102d to access PSTN 108, the Internet 110, and / or other networks 112. PSTN 108 may include a circuit-switched telephone network providing Common Old-Style Telephone Service (POTS). The Internet 110 may include a global system of interconnected computer networks and devices using common communication protocols such as Transmission Control Protocol (TCP), User Datagram Protocol (UDP), and / or Internet Protocol (IP) from the TCP / IP Internet Protocol suite. Network 112 may include wired and / or wireless communication networks owned and / or operated by other service providers. For example, network 112 may include another CN connected to one or more RANs, which may use the same RAT as RAN 104 / 114 or a different RAT.
[0026] Some or all of the WTRUs 102a, 102b, 102c, and 102d in communication system 100 may include multi-mode capabilities (e.g., WTRUs 102a, 102b, 102c, and 102d may include multiple transceivers for communicating with different wireless networks via different wireless links). For example, Figure 1A The WTRU 102c shown can be configured to communicate with a base station 114a that can use cellular-based radio technology and a base station 114b that can use IEEE 802 radio technology.
[0027] Figure 1B This shows a system diagram of an example WTRU 102. (See diagram below.) Figure 1B As shown, WTRU 102 may include a processor 118, a transceiver 120, a transmit / receive element 122, a speaker / microphone 124, a keypad 126, a display / touchpad 128, non-removable memory 130, removable memory 132, a power supply 134, a Global Positioning System (GPS) chipset 136, and / or other components / peripherals 138, etc. It will be appreciated that WTRU 102 may include any sub-combination of the above components while remaining consistent with the embodiments.
[0028] Processor 118 may be a general-purpose processor, a special-purpose processor, a conventional processor, a digital signal processor (DSP), multiple microprocessors, one or more microprocessors associated with a DSP core, a controller, a microcontroller, an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) circuit, any other type of integrated circuit (IC), a state machine, etc. Processor 118 may perform signal encoding, data processing, power control, input / output processing, and / or any other functions that enable WTRU 102 to operate in a wireless environment. Processor 118 may be coupled to transceiver 120, which may be coupled to transmitting / receiving element 122. Although... Figure 1B The processor 118 and transceiver 120 are depicted as separate components, but it will be understood that the processor 118 and transceiver 120 can be integrated together, for example, in an electronic package or chip.
[0029] Transmitting / receiving element 122 can be configured to transmit signals to or receive signals from a base station (e.g., base station 114a) via air interface 116. For example, in one embodiment, transmitting / receiving element 122 can be an antenna configured to transmit and / or receive RF signals. In another embodiment, transmitting / receiving element 122 can be a transmitter / detector configured to transmit and / or receive, for example, IR, UV, or visible light signals. In yet another embodiment, transmitting / receiving element 122 can be configured to transmit and / or receive both RF and optical signals. It will be appreciated that transmitting / receiving element 122 can be configured to transmit and / or receive any combination of wireless signals.
[0030] Although the transmitting / receiving element 122 is in Figure 1B While depicted as a single element, WTRU 102 may include any number of transmitting / receiving elements 122. For example, WTRU 102 may employ MIMO technology. Therefore, in an embodiment, WTRU 102 may include two or more transmitting / receiving elements 122 (e.g., multiple antennas) for transmitting and receiving wireless signals via air interface 116.
[0031] Transceiver 120 can be configured to modulate signals to be transmitted by transmitting / receiving element 122 and demodulate signals received by transmitting / receiving element 122. As noted above, WTRU 102 can have multi-mode capability. Thus, for example, transceiver 120 may include multiple transceivers for enabling WTRU 102 to communicate via multiple RATs (such as NR and IEEE 802.11).
[0032] The processor 118 of WTRU 102 can be coupled to the speaker / microphone 124, keypad 126, and / or display / touchpad 128 (e.g., a liquid crystal display (LCD) unit or an organic light-emitting diode (OLED) display unit), and can receive user input data from them. The processor 118 can also output user data to the speaker / microphone 124, keypad 126, and / or display / touchpad 128. Additionally, the processor 118 can access information from any type of suitable memory (such as non-removable memory 130 and / or removable memory 132), and store data in that memory. Non-removable memory 130 may include random access memory (RAM), read-only memory (ROM), hard disk, or any other type of memory storage device. Removable memory 132 may include a subscriber identity module (SIM) card, memory stick, secure digital storage (SD) card, etc. In other embodiments, processor 118 may access information from memory that is not physically located on WTRU 102 (such as on a server or home computer (not shown)) and store data in that memory.
[0033] The processor 118 can receive power from the power supply 134 and can be configured to distribute and / or control the power going to other components in the WTRU 102. The power supply 134 can be any suitable device for powering the WTRU 102. For example, the power supply 134 may include one or more dry cell batteries (e.g., nickel-cadmium (NiCd), nickel-zinc (NiZn), nickel metal hydride (NiMH), lithium-ion (Li-ion), etc.), solar cells, fuel cells, etc.
[0034] The processor 118 may also be coupled to a GPS chipset 136, which may be configured to provide location information (e.g., longitude and latitude) about the current location of the WTRU 102. In addition to, or instead of, information from the GPS chipset 136, the WTRU 102 may receive location information from base stations (e.g., base stations 114a, 114b) via air interface 116 and / or determine its location based on the timing of signals received from two or more nearby base stations. It will be understood that the WTRU 102 may acquire location information using any suitable location determination method, while remaining consistent with the embodiments.
[0035] The processor 118 may be further coupled to other components / peripherals 138, which may include one or more software and / or hardware modules / units providing additional features, functions, and / or wired or wireless connectivity. For example, components / peripherals 138 may include accelerometers, electronic compasses, satellite transceivers, digital cameras (e.g., for photos and / or videos), Universal Serial Bus (USB) ports, vibration devices, television transceivers, hands-free headsets, Bluetooth® modules, FM radio units, digital music players, media players, video game player modules, internet browsers, virtual reality and / or augmented reality (VR / AR) devices, activity trackers, etc. Components / peripherals 138 may include one or more sensors, which may be one or more of the following: gyroscopes, accelerometers, Hall effect sensors, magnetometers, orientation sensors, proximity sensors, temperature sensors, time sensors, geolocation sensors, altimeters, light sensors, touch sensors, magnetometers, barometers, gesture sensors, biometric sensors, and / or humidity sensors.
[0036] WTRU 102 may include a full-duplex radio, for which the transmission and reception of some or all signals (e.g., associated with specific subframes for both uplink (e.g., for transmission) and downlink (e.g., for reception)) may be concurrent and / or simultaneous. The full-duplex radio may include an interference management unit to reduce and / or substantially eliminate self-interference via hardware (e.g., a choke) or via signal processing (e.g., a separate processor (not shown) or via processor 118). In embodiments, WTRU 102 may include a half-duplex radio, for which the transmission and reception of some or all signals (e.g., associated with specific subframes for either uplink (e.g., for transmission) or downlink (e.g., for reception)) may be concurrent and / or simultaneous.
[0037] Figure 1C This diagram illustrates a system diagram of RAN 104 and CN 106 according to an embodiment. As noted above, RAN 104 may employ E-UTRA radio technology to communicate with WTRUs 102a, 102b, and 102c via air interface 116. RAN 104 may also communicate with CN 106.
[0038] RAN 104 may include eNode-Bs 160a, 160b, and 160c, although it will be understood that RAN 104 may include any number of eNode-Bs while remaining consistent with the embodiments. eNode-Bs 160a, 160b, and 160c may each include one or more transceivers for communicating with WTRUs 102a, 102b, and 102c via air interface 116. In the embodiments, eNode-Bs 160a, 160b, and 160c may implement MIMO technology. Therefore, eNode-B 160a may, for example, use multiple antennas to transmit radio signals to and receive radio signals from WTRU 102a.
[0039] Each of eNode-B 160a, 160b, and 160c can be associated with a specific cell (not shown) and can be configured to handle radio resource management decisions, handover decisions, and user scheduling in the uplink (UL) and / or downlink (DL), etc. Figure 1C As shown, eNode-B 160a, 160b, and 160c can communicate with each other via the X2 interface.
[0040] Figure 1C The CN 106 shown may include a Mobility Management Entity (MME) 162, a Serving Gateway (SGW) 164, and a Packet Data Network (PDN) Gateway (PGW) 166. Although each of the foregoing elements is depicted as part of CN 106, it will be understood that any of these elements may be owned and / or operated by an entity other than a CN operator.
[0041] The MME 162 can connect to each of the eNode-Bs 160a, 160b, and 160c in RAN 104 via the S1 interface and can act as a control node. For example, the MME 162 can be responsible for authenticating users of WTRUs 102a, 102b, and 102c, bearer activation / deactivation, selecting a specific serving gateway during the initial attachment of WTRUs 102a, 102b, and 102c, etc. The MME 162 can provide control plane functions for handover between RAN 104 and other RANs (not shown) employing other radio technologies such as GSM and / or WCDMA.
[0042] The SGW 164 can connect to each of the eNode Bs 160a, 160b, and 160c in RAN 104 via the S1 interface. The SGW 164 can typically route and forward user data packets to / from WTRUs 102a, 102b, and 102c. The SGW 164 can perform other functions, such as anchoring the user plane during inter-eNode B handover, triggering paging when DL data is available for WTRUs 102a, 102b, and 102c, and managing and storing the context of WTRUs 102a, 102b, and 102c.
[0043] SGW 164 can be connected to PGW 166, which can provide WTRU 102a, 102b, 102c with access to a packet-switched network (such as Internet 110) to facilitate communication between WTRU 102a, 102b, 102c and IP-enabled devices.
[0044] CN 106 facilitates communication with other networks. For example, CN 106 can provide WTRUs 102a, 102b, and 102c with access to a circuit-switched network (such as PSTN 108) to facilitate communication between WTRUs 102a, 102b, and 102c and conventional terrestrial line communication equipment. For example, CN 106 may include or communicate with an IP gateway (e.g., an IP Multimedia Subsystem (IMS) server), which acts as an interface between CN 106 and PSTN 108. Additionally, CN 106 can provide WTRUs 102a, 102b, and 102c with access to other networks 112, which may include other wired and / or wireless networks owned and / or operated by other service providers.
[0045] Despite WTRU in Figures 1A to 1D While described as a wireless terminal, it is envisioned that, in some representative embodiments, such a terminal may use (e.g., temporarily or permanently) a wired communication interface with a communication network.
[0046] In a representative embodiment, the other network 112 may be a WLAN.
[0047] In an Infrastructure Basic Services Set (BSS) mode, a WLAN may have an access point (AP) for the BSS and one or more stations (STAs) associated with the AP. The AP may have access or an interface to a distribution system (DS) or another type of wired / wireless network that carries traffic into and / or out of the BSS. Traffic originating outside the BSS destined for a STA can be delivered to the STA via the AP. Traffic from a STA to a destination outside the BSS can be sent to the AP for delivery to the appropriate destination. Traffic between STAs within the BSS can be sent via the AP, for example, where a source STA can send traffic to the AP, and the AP can deliver the traffic to the destination STA. Traffic between STAs within the BSS can be considered and / or referred to as peer-to-peer traffic. Peer-to-peer traffic can be sent between source and destination STAs (e.g., directly between them) using a direct link setup (DLS). In some representative embodiments, the DLS may use 802.11e DLS or 802.11z Tunneled DLS (TDLS). A WLAN using the Standalone BSS (IBSS) mode may not have an access point (AP), and STAs within or using the IBSS (e.g., all STAs) can communicate directly with each other. The IBSS communication mode may sometimes be referred to as a "self-organizing" communication mode in this document.
[0048] When using 802.11ac infrastructure operation mode or a similar operation mode, the AP can transmit beacons on a fixed channel, such as a primary channel. The primary channel can be of fixed width (e.g., a bandwidth of 20 MHz) or dynamically set via signaling. The primary channel can be the operating channel of the BSS and can be used by STAs to establish connections with the AP. In some representative embodiments, Carrier Sense Multiple Access with Collision Avoidance (CSMA / CA) can be implemented, for example, in an 802.11 system. For CSMA / CA, STAs including the AP (e.g., each STA) can sense the primary channel. If the primary channel is sensed / detected and / or determined to be busy by a particular STA, that STA can back off. A STA (e.g., only one station) can transmit at any given time within a given BSS.
[0049] High-throughput (HT) STAs can communicate using a 40MHz wide channel, for example, by combining a primary 20MHz channel with adjacent or non-adjacent 20MHz channels to form a 40MHz wide channel.
[0050] Very High Throughput (VHT) STAs can support channels with widths of 20MHz, 40MHz, 80MHz, and / or 160MHz. 40MHz and / or 80MHz channels can be formed by combining consecutive 20MHz channels. A 160MHz channel can be formed by combining eight consecutive 20MHz channels or by combining two non-consecutive 80MHz channels (which can be referred to as an 80+80 configuration). For the 80+80 configuration, after channel coding, data is transmitted via a segment resolver that divides the data into two streams. Inverse Fast Fourier Transform (IFFT) processing and time-domain processing can be performed separately on each stream. The streams can be mapped onto two 80MHz channels, and the data can be transmitted by the transmitting STA. At the receiver of the receiving STA, the above operations for the 80+80 configuration can be reversed, and the combined data can be sent to the Media Access Control (MAC) layer, entities, etc.
[0051] Operating modes below 1 GHz are supported by 802.11af and 802.11ah. The channel operating bandwidth and carrier are reduced in 802.11af and 802.11ah compared to those used in 802.11n and 802.11ac. 802.11af supports 5 MHz, 10 MHz, and 20 MHz bandwidths in the TV white space (TVWS) spectrum, while 802.11ah supports 1 MHz, 2 MHz, 4 MHz, 8 MHz, and 16 MHz bandwidths using non-TVWS spectrum. According to a representative embodiment, 802.11ah may support instrument-type control / machine-type communication (MTC), such as MTC devices in macro coverage areas. MTC devices may have certain capabilities, such as limited capabilities, including support for (e.g., only support) certain and / or limited bandwidths. MTC devices may include batteries with a battery life exceeding a threshold (e.g., to maintain a very long battery life).
[0052] WLAN systems that can support multiple channels and channel bandwidths (such as 802.11n, 802.11ac, 802.11af, and 802.11ah) include a channel that can be designated as the primary channel. The primary channel can have a bandwidth equal to the maximum common operating bandwidth supported by all STAs in the BSS. The bandwidth of the primary channel can be set and / or limited by the STA that supports the minimum bandwidth operating mode among all STAs operating in the BSS. In the 802.11ah example, for STAs that support (e.g., only support) the 1MHz mode (e.g., MTC type devices), the primary channel can be 1MHz wide, even if the AP and other STAs in the BSS support 2MHz, 4MHz, 8MHz, 16MHz, and / or other channel bandwidth operating modes. Carrier sensing and / or Network Allocation Vector (NAV) settings can depend on the status of the primary channel. If the primary channel is busy, for example because an STA (which only supports the 1MHz operating mode) is transmitting to the AP, the entire available band may be considered busy, even if most of the band is still idle and could be available.
[0053] In the United States, the available frequency band for 802.11ah is from 902MHz to 928MHz. In South Korea, the available frequency band is from 917.5MHz to 923.5MHz. In Japan, the available frequency band is from 916.5MHz to 927.5MHz. Depending on the country code, the total available bandwidth for 802.11ah is 6MHz to 26MHz.
[0054] Figure 1D This diagram illustrates a system diagram of RAN 113 and CN 115 according to an embodiment. As noted above, RAN 113 may employ NR radio technology to communicate with WTRUs 102a, 102b, and 102c via air interface 116. RAN 113 may also communicate with CN 115.
[0055] RAN 113 may include gNBs 180a, 180b, and 180c, although it will be understood that RAN 113 may include any number of gNBs while remaining consistent with the embodiments. gNBs 180a, 180b, and 180c may each include one or more transceivers for communicating with WTRUs 102a, 102b, and 102c via air interface 116. In embodiments, gNBs 180a, 180b, and 180c may implement MIMO technology. For example, gNBs 180a and 180b may utilize beamforming to transmit signals to and / or receive signals from WTRUs 102a, 102b, and 102c. Therefore, gNB 180a may, for example, use multiple antennas to transmit radio signals to and / or receive radio signals from WTRU 102a. In embodiments, gNBs 180a, 180b, and 180c can implement carrier aggregation technology. For example, gNB 180a can transmit multiple component carriers to WTRU 102a (not shown). A subset of these component carriers can be on unlicensed spectrum, while the remaining component carriers can be on licensed spectrum. In embodiments, gNBs 180a, 180b, and 180c can implement Cooperative Multipoint (CoMP) technology. For example, WTRU 102a can receive cooperative transmissions from gNBs 180a and 180b (and / or gNB 180c).
[0056] WTRUs 102a, 102b, and 102c can communicate with gNBs 180a, 180b, and 180c using transmissions associated with scalable digitization. For example, OFDM symbol spacing and / or OFDM subcarrier spacing can vary depending on different transmissions, different cells, and / or different portions of the radio transmission spectrum. WTRUs 102a, 102b, and 102c can communicate with gNBs 180a, 180b, and 180c using subframes of various or scalable lengths or transmission time intervals (TTIs) (e.g., including different numbers of OFDM symbols and / or absolute times of varying durations).
[0057] gNBs 180a, 180b, and 180c can be configured to communicate with WTRUs 102a, 102b, and 102c in standalone and / or non-standalone configurations. In standalone configuration, WTRUs 102a, 102b, and 102c can communicate with gNBs 180a, 180b, and 180c without also accessing other RANs (e.g., eNode-Bs 160a, 160b, and 160c). In standalone configuration, WTRUs 102a, 102b, and 102c can utilize one or more of gNBs 180a, 180b, and 180c as mobility anchors. In standalone configuration, WTRUs 102a, 102b, and 102c can communicate with gNBs 180a, 180b, and 180c using signals in unlicensed frequency bands. In a non-standalone configuration, WTRUs 102a, 102b, and 102c can communicate with / connect to gNBs 180a, 180b, and 180c, and simultaneously communicate with / connect to another RAN (such as eNode-Bs 160a, 160b, and 160c). For example, WTRUs 102a, 102b, and 102c can implement DC principles to communicate substantially simultaneously with one or more gNBs 180a, 180b, and 180c and one or more eNode-Bs 160a, 160b, and 160c. In a non-standalone configuration, eNode-B 160a, 160b, and 160c can act as mobility anchors for WTRU 102a, 102b, and 102c, and gNB 180a, 180b, and 180c can provide additional coverage and / or throughput to serve WTRU 102a, 102b, and 102c.
[0058] Each of gNBs 180a, 180b, and 180c can be associated with a specific cell (not shown) and can be configured to handle radio resource management decisions, handover decisions, user scheduling in UL and / or DL, network slicing support, dual connectivity, interoperability between NR and E-UTRA, routing of user plane data to User Plane Functions (UPF) 184a and 184b, routing of control plane information to Access and Mobility Management Functions (AMF) 182a and 182b, etc. Figure 1D As shown, gNB 180a, 180b, and 180c can communicate with each other via the Xn interface.
[0059] Figure 1DThe CN 115 shown may include at least one AMF 182a, 182b, at least one UPF 184a, 184b, at least one Session Management Function (SMF) 183a, 183b, and at least one Data Network (DN) 185a, 185b. Although each of the foregoing elements is depicted as part of the CN 115, it will be understood that any of these elements may be owned and / or operated by an entity other than the CN operator.
[0060] AMF 182a and 182b can connect to one or more of the gNBs 180a, 180b, and 180c in RAN 113 via the N2 interface and can act as control nodes. For example, AMF 182a and 182b can be responsible for authenticating users of WTRU 102a, 102b, and 102c, supporting network slicing (e.g., handling different Protocol Data Unit (PDU) sessions with different requirements), selecting specific SMF 183a and 183b, managing registration areas, terminating NAS signaling, mobility management, etc. Network slices can be used by AMF 182a and 182b, for example, to customize CN support for WTRU 102a, 102b, and 102c based on the service types utilized by WTRU 102a, 102b, and 102c. For example, different network slices can be built for different use cases, such as services that rely on Ultra Reliable Low Latency (URLLC) access, services that rely on Enhanced Massive Mobile Broadband (eMBB) access, and services for MTC access. AMF 162 can provide control plane functions for handover between RAN 113 and other RANs (not shown) that employ other radio technologies, such as LTE, LTE-A, LTE-A Pro, and / or non-3GPP access technologies, such as Wi-Fi.
[0061] SMFs 183a and 183b can connect to AMFs 182a and 182b in CN 115 via the N11 interface. SMFs 183a and 183b can also connect to UPFs 184a and 184b in CN 115 via the N4 interface. SMFs 183a and 183b can select and control UPFs 184a and 184b, and configure traffic routing through UPFs 184a and 184b. SMFs 183a and 183b can perform other functions, such as managing and allocating UE IP addresses, managing PDU sessions, controlling policy enforcement and QoS, and providing downlink data notifications. PDU session types can be IP-based, non-IP-based, Ethernet-based, etc.
[0062] UPF 184a and 184b can connect to one or more of gNBs 180a, 180b, and 180c in RAN 113 via the N3 interface. These gNBs can provide WTRU 102a, 102b, and 102c with access to packet-switched networks (such as the Internet 110), for example, to facilitate communication between WTRU 102a, 102b, 102c and IP-enabled devices. UPF 184a and 184b can perform other functions, such as routing and forwarding packets, enforcing user plane policies, supporting multi-homed PDU sessions, handling user plane QoS, buffering downlink packets, and providing mobility anchoring.
[0063] CN 115 can facilitate communication with other networks. For example, CN 115 may include or communicate with an IP gateway (e.g., an IP Multimedia Subsystem (IMS) server) that acts as an interface between CN 115 and PSTN 108. Additionally, CN 115 can provide WTRUs 102a, 102b, and 102c with access to other networks 112, which may include other wired and / or wireless networks owned and / or operated by other service providers. In an embodiment, WTRUs 102a, 102b, and 102c can connect to local data networks (DNs) 185a and 185b via the N3 interface to UPFs 184a and 184b and the N6 interface between UPFs 184a and 184b and DNs 185a and 185b.
[0064] Given Figures 1A to 1D as well as Figures 1A to 1D The corresponding descriptions may be performed by one or more emulation components / devices (not shown) to perform one or more of the functions described herein with respect to any of the following: WTRU102a-d, base station 114a-b, eNode-B 160a-c, MME 162, SGW 164, PGW 166, gNB 180a-c, AMF 182a-b, UPF 184a-b, SMF 183a-b, DN 185a-b, and / or one or more other components / devices described herein. An emulation device may be one or more devices configured to emulate one or more of the functions described herein. For example, an emulation device may be used to test other devices and / or simulate network and / or WTRU functions.
[0065] Simulation devices can be designed to perform tests on one or more other devices in a laboratory environment and / or a carrier network environment. For example, the one or more simulation devices can perform one or more or all of their functions while being fully or partially implemented and / or deployed as part of a wired and / or wireless communication network to test other devices within the communication network. The one or more simulation devices can perform one or more or all of their functions while being temporarily implemented / deployed as part of a wired and / or wireless communication network. Simulation devices can be directly coupled to another device for testing purposes and / or can use over-the-air wireless communication to perform tests.
[0066] The one or more simulation devices can perform one or more (including all) functions without being implemented / deployed as part of a wired and / or wireless communication network. For example, the simulation devices can be used in test scenarios in a test laboratory and / or in non-deployed (e.g., testing) wired and / or wireless communication networks to perform testing on one or more components. The one or more simulation devices can be test rigs. Direct RF coupling and / or wireless communication via RF circuitry (e.g., which may include one or more antennas) can be used by the simulation devices to transmit and / or receive data.
[0067] introduction The following abbreviations and acronyms may be used in this article: ACK (Accept) AoA Angle of Arrival AoD (Leaving Angle) ARFCN Absolute Radio Frequency Channel Number BLER block error rate BW bandwidth BWP bandwidth portion CAP channel access priority CAPC Channel Access Priority Categories CCA Idle Channel Assessment CCE Control Channel Element CDF cumulative distribution function CE control elements CG configuration authorization or cell group CIR channel impulse response CORESET Control Resource Set CP cyclic prefix CP-OFDM vs. Traditional OFDM (dependent on cyclic prefix) CQI Channel Quality Indicator CRC Cyclic Redundancy Check CSI Channel State Information CW Competition Window CWS Competition Window Size CO channel occupancy DAI Downlink Allocation Index DCI Downlink Control Information DFI downlink feedback information DG Dynamic Licensing DL downlink DM-RS demodulation reference signal DP delay distribution DRB Data Radio Bearer DRX Discontinuous Reception ECID Enhanced Cell ID eLAA Enhanced Licensing Assisted Access eMBB Enhanced Mobile Broadband FeLAA Further Enhanced Licensed Assisted Access HARQ Hybrid Automatic Repeat Request IM Interference Measurement KPIs (Key Performance Indicators) LAA Licensed Assisted Access LBT Listen before you speak LCH Logical Channel LCM Lifecycle Management LCP Logical Channel Priority LBT Listen before you speak LOS (Location of View) NLOS (Non-line-of-sight) LMF location management function LPP LTE positioning protocol LTE Long Term Evolution, such as from 3GPP LTE Release 8 and later. MAC CE MAC control element MAC Media Access Control MCS modulation and coding scheme MIMO (Multiple Input Multiple Output) Mean Squared Error NACK (Negative) NAS Non-Access Layer NR New Radio OFDM (Orthogonal Frequency Division Multiplexing) OTDOA observed arrival time difference PDCCH (Physical Downlink Control Channel) PDSCH (Physical Downlink Shared Channel) PDP power delay distribution PDU (Packet Data Unit) PFL Positioning Frequency Layer PHY physical layer PID (Process ID) PO paging timing PRACH (Physical Random Access Channel) PRS Positioning Reference Signal PRU Positioning Reference Element PSS Master Synchronization Signal PTRS phase tracking reference signal PUCCH (Physical Uplink Control Channel) PUSCH Physical Uplink Shared Channel RA (Random Access or Procedure) RACH Random Access Channel RAR Random Access Response RAT Radio Access Technology RCU Radio Access Network Central Unit RE Resource Elements RF radio front end RLF radio link failure RLM Radio Link Monitoring RMSE (Root Mean Square Error) RNTI Radio Network Identifier RNA RAN Notification Region RO RACH timing RRC Radio Resource Control RRM Radio Resource Management RTT round trip time RP receiving point RS reference signal RSRP reference signal received power RSTD (Reference Signal Time Difference) RTT round trip time RSSI Received Signal Strength Indicator RTOA (Relative Time of Arrival) SDAP Service Data Adaptation Protocol SDU Service Data Unit SRB signaling radio bearer SRS Detection Reference Signal SRSp is an SRS used for positioning. SS synchronization signal SSB Synchronization Signal Block SSS auxiliary synchronization signal SWG switching interval (in a self-contained subframe) SPS (Semi-Persistent Scheduling) SUL supplements uplink TA scheduled in advance TB transfer block TBS (Transfer Block Size) TDoA Arrival Time Difference TEG Timing Error Group TRP Transmitter-Receiver Point TSC Time-Sensitive Communication TSN Time-Sensitive Networking TTI Transmission Time Interval UCI uplink control information UL uplink URLLC: Ultra-Reliable and Low-Latency Communication WBWP wide bandwidth portion WLAN (Wireless Local Area Network) and related technologies (IEEE 802.xx domain).
[0068] Positioning method In 3GPP Release 16, DL, UL, and DL and UL positioning methods were used (e.g., as described in TS 38.305).
[0069] In some representative embodiments, WTRU 102 and / or the network may use one or more positioning methods.
[0070] For example, "DL positioning method" can refer to any positioning method that uses a downlink reference signal (such as PRS). The WTRU 102 can receive multiple reference signals from one or more transmit points (TPs) and measure DL RSTD and / or RSRP. Examples of DL positioning methods are DL-AoD and DL-TDOA positioning.
[0071] For example, "UL positioning method" can refer to any positioning method that uses an uplink reference signal, such as an SRS for positioning. The WTRU 102 can transmit SRS to multiple receiver points (RPs), and the RPs can measure UL RTOA and / or RSRP. Examples of UL positioning methods are UL-TDOA and UL-AoA positioning.
[0072] For example, "DL and UL positioning method" can refer to any positioning method that uses both uplink and downlink reference signals for positioning. In one example, WTRU 102 can transmit one or more SRSs to multiple TRPs, and gNB can measure the Rx-Tx time difference calculated based on the arrival time of the DL RS (e.g., PRS). gNB can measure the RSRP of the received SRS. WTRU 102 can measure the Rx-Tx time difference of PRSs transmitted from multiple TRPs. WTRU 102 can measure the RSRP of the received PRS. The Rx-TX difference and / or RSRP measured at WTRU 102 and gNB can be used to calculate the round-trip time (RTT). As used herein, the phrase "UE Rx-Tx time difference" can refer to the difference between the arrival time of the reference signal transmitted by the TRP and the transmission time of the reference signal transmitted from WTRU 102. An example of DL and UL positioning method is multi-RTT positioning.
[0073] Machine Learning (ML) For example, machine learning can refer to a type of algorithm that solves problems based on learning from experience (e.g., data) without explicit programming (e.g., configuring a set of rules). Machine learning can be considered a subset of AI. Different machine learning paradigms can be envisioned based on the nature of the data and / or feedback available for learning the algorithm. For example, supervised learning methods can involve learning a function that maps inputs to outputs based on labeled training examples, where each training example can be a pair consisting of an input and a corresponding output. For example, unsupervised learning methods can involve detecting patterns in data that do not have pre-existing labels. For example, reinforcement learning methods can involve performing a series of actions in an environment to maximize cumulative rewards. In some embodiments, machine learning algorithms can be applied using a combination or interpolation of the above methods. For example, semi-supervised learning methods can use a combination of a small amount of labeled data and a large amount of unlabeled data during training. In this respect, semi-supervised learning falls between unsupervised learning (e.g., no labeled training data) and supervised learning (e.g., only labeled training data).
[0074] Artificial intelligence (AI) for location services Artificial intelligence can be broadly defined as the behavior of machines that mimic cognitive functions, including the ability to sense, reason, adapt, act, and / or provide pattern recognition.
[0075] Neural Networks Figure 2 This is a block diagram illustrating an example of a neural network 200. The goal of training is to apply the input and adjust the weights (in...). Figure 2 The values are designated as 202 and 204 (which can be called neuron weights or link weights), making the output from neural network 200 approximate the desired target value associated with the input value. Figure 2 In the example, neural network 200 consists of 3 layers. During training, for a given input, the difference between the output value and the expected value is calculated, and the difference is used to update the weights 202 and 204 in neural network 200. If a large difference is observed between the output value and the expected value, a large change in weights 202 and 204 can be expected, while a small difference may result in a small change in weights 202 and 204.
[0076] For example, for localization, the input could be reference signal parameters, and the output could be an estimated location. The desired value could be location information, such as location information that can be obtained with high accuracy by GNSS. Once the neural network 200 has completed its training (e.g., the difference between the output and the desired value is below a threshold), it can be applied to localization by feeding the input and using the output as the desired result associated with the input. For example, the output could be the estimated localization or location of WTRU 102.
[0077] Therefore, for training a neural network, it is important to identify the following: (i) the input of the neural network; (ii) the expected output associated with the input; and (iii) the actual output from the neural network compared with the target value.
[0078] As an example, a neural network model can be characterized by the following parameters: (i) the number of weights; (ii) the number of layers in the neural network; and (iii) the number of neurons per layer.
[0079] As an example, a neural network can be characterized by the following parameters: (i) the number and type of layers, and (ii) the parameter values (e.g., weights) associated with each layer.
[0080] Deep learning Deep learning refers to a class of machine learning algorithms that employ artificial neural networks (such as deep neural networks) loosely inspired by biological systems and include at least one hidden layer. Deep neural networks (DNNs) are a special type of machine learning model inspired by the human brain, where the input is linearly transformed and passed multiple times through a non-linear activation function. DNNs typically consist of multiple layers, each composed of a linear transformation and a given non-linear activation function. DNNs can be trained using training data via the backpropagation algorithm. Recently, DNNs have demonstrated state-of-the-art performance in various domains (e.g., speech, vision, natural language processing) and for a wide range of machine learning settings (e.g., supervised, unsupervised, and semi-supervised).
[0081] As used in this article, "event" and "timing" can be used interchangeably.
[0082] AI / ML localization methods are effective in both line-of-sight and non-line-of-sight environments because they use fingerprinting (e.g., associating RSRP and / or timing measurements with the WTRU location) when it is related to localization. However, the WTRU 102 needs to determine which AI / ML model to use. As the RS configuration or environment of the WTRU 102 (e.g., cell, nearby TRP) changes due to WTRU mobility, the performance of the AI / ML model used for localization by the WTRU 102 may degrade due to a mismatch between the trained AI / ML model and the changed environment. Providing a process that allows the WTRU 102 to determine whether to switch AI / ML models, fine-tune AI / ML models, and / or use fallback localization methods to maintain localization performance would be beneficial. Providing a process for the WTRU to notify the network of any changes to the AI / ML model would also be beneficial.
[0083] Overview Current 3GPP specifications define location methods that rely on the RAT (Range Atlas) system. These methods require the WTRU 102 to be in a line-of-sight environment relative to the TRP (Tracking Point Restriction Point), limiting applicable use cases. In non-line-of-sight environments, the performance of RAT-dependent location methods degrades.
[0084] In some representative embodiments, the AI / ML localization method can be effective in both line-of-sight and non-line-of-sight environments due to the use of fingerprint-based localization (e.g., associating RSRP and / or timing measurement results with the WTRU location). However, the WTRU 102 needs to determine which AI / ML model to use. As the RS configuration or environment of the WTRU 102 (e.g., cell, nearby TRP) changes (e.g., due to WTRU mobility), the performance of the AI / ML model used for localization by the WTRU 102 may degrade due to a mismatch between the trained AI / ML model and the changed environment. It would be beneficial to provide techniques for the WTRU 102 to determine whether to switch and / or fine-tune the localization method and / or use a fallback localization method to maintain localization performance. Furthermore, it would be beneficial to provide techniques that allow the WTRU 102 to notify the network of any changes to the AI / ML model used.
[0085] In some representative embodiments, the WTRU can perform location reference unit (PRU)-assisted AI / ML model switching.
[0086] For example, the AI / ML model can be associated with a cell. When WTRU 102 moves to a different cell, it can query (e.g., request) PRU measurements and their location. The network can provide the requested information to WTRU 102, and WTRU 102 can determine whether the current AI / ML model (e.g., the model used in a previous cell) can still be used. After the auxiliary information is provided, the WTRU can report (e.g., indicate) whether WTRU 102 is still using the same model.
[0087] For example, the WTRU can be moved to a new cell (e.g., due to mobility). The WTRU 102 can indicate WTRU capability information to the network (e.g., support for WTRU-based AI / ML-based positioning).
[0088] For example, WTRU 102 may send a request for auxiliary information (such as PRU location, PRU inference, timestamp, associated measurement results) that indicates (e.g., required) the level of uncertainty (e.g., standard deviation) for PRU inference, the desired associated information (e.g., the desired SSB), and / or the cell ID of the previous cell.
[0089] For example, WTRU 102 can be configured with a backtracking method (e.g., DL-TDOA).
[0090] For example, WTRU 102 can perform measurements on a configured PRS set within a configured time window (e.g., periodicity, duration).
[0091] For example, WTRU 102 may periodically receive auxiliary information (such as PRU measurement results, corresponding PRU inference, and PRU location) from the network. In some embodiments, the auxiliary information may include the duration of the provisioning of the auxiliary information, such as the duration being the same as the duration of a time window.
[0092] For example, at the end of the preparation of auxiliary information or if WTRU 102 receives a request from the network, WTRU 102 may report (e.g., send) information indicating any of the following: (i) performing fine-tuning using the current AI / ML model; (ii) the association information of the new model (e.g., cell ID); (iii) reusing the model used in the previous cell; and / or (iv) using a back-off positioning method.
[0093] Typical WTRU behavior In some representative embodiments, WTRU 102 may send requests to the network for configuration information (such as PRS configuration and / or SRSp configuration), such as via PUSCH, PUCCH, UCI, MAC-CE, RRC, or LPP messages. Requests from WTRU 102 may include configurations for measurement gaps, PRS processing windows, and / or windows for the transmission of SRS (SRSp) for positioning.
[0094] In some representative embodiments, WTRU 102 may send an affirmative response message (e.g., in PUSCH or PUCCH) for authorization received from the network.
[0095] In some representative embodiments, more than one condition and / or criterion can be combined. For example, WTRU 102 can be configured with multiple conditions and associated WTRU behaviors. WTRU 102 can determine which behavior WTRU 102 can or should use based on applicable conditions.
[0096] In some representative embodiments, WTRU 102 can measure DL-PRS inside or outside (e.g., an active) BWP. WTRU 102 can also emit SRSp inside or outside (e.g., an active) BWP.
[0097] In some representative embodiments, the WTRU 102 may be pre-configured with parameters (e.g., measurement gap, PRS processing window, PRS configuration, SRSp configuration) via semi-static messages (e.g., LPP, RRC).
[0098] In some representative embodiments, any action to be taken by WTRU 102 can be configured by the network. For example, WTRU 102 can be configured with rules, and based on those rules, WTRU 102 can determine the associated action to take.
[0099] In some representative embodiments, WTRU 102 may (e.g., in addition to measurements of PRS) perform (e.g., execute and / or report) any of the following cell-related measurements: (i) SSB RSRP and corresponding cell ID from the serving cell; (ii) SSB RSRP and corresponding cell ID from any neighboring cell; (iii) RSRP of CSI-RS with CSI-RS resource ID; and / or (iv) RSRS of DM-RS.
[0100] the term As used in this article, the phrase “network” can be used interchangeably with any of AMF, LMF, gNB and / or NG-RAN.
[0101] As used in this article, the phrases “pre-configured” and “configured” can be used interchangeably.
[0102] As used in this article, the phrases “non-service gNB” and “neighboring gNB” can be used interchangeably.
[0103] As used herein, the phrases “gNB” and “TRP” are used interchangeably in this disclosure.
[0104] As used herein, the phrases “PRS,” “SRS,” “SRS for location purposes,” and / or “SRS for location purposes” are used interchangeably.
[0105] As used in this article, the phrases “PRS” and “PRS resources” are used interchangeably.
[0106] As used in this article, the phrases “PRS” and “PRS resource” are used interchangeably. PRS and / or PRS resources may belong to different PRS resource sets.
[0107] As used in this article, the phrases “PRS,” “DL-PRS,” and / or “DL PRS” are used interchangeably.
[0108] As used herein, the phrases “measurement gap” and “measurement gap mode” are used interchangeably. A measurement gap mode may include parameters such as measurement gap duration, measurement gap repetition period, and / or measurement gap periodicity.
[0109] As used in this article, the phrases “location” and “position” can be used interchangeably.
[0110] In some embodiments, the PRU can be a WTRU 102 or TRP whose location (e.g., altitude, latitude, geographic coordinates, or local coordinates) is known to the network (e.g., gNB, LMF). The capabilities of the PRU can be the same as those of the WTRU 102 or TRP (e.g., capable of receiving PRS or transmitting SRS or SRSp, returning measurement results, and / or transmitting PRS). Any WTRU 102 acting as a PRU can be used by the network for calibration purposes (e.g., correcting unknown timing offsets, correcting unknown angular offsets).
[0111] As an example, LMF is a non-limiting example of a node or entity (such as a network node or entity) that can be used to or support location. Any other node or entity can replace LMF and still be consistent with the examples and embodiments described herein.
[0112] In some embodiments, WTRU 102 may receive pre-configured thresholds from a network (e.g., LMF, gNB).
[0113] In some embodiments, the LOS indicator can be a hard indicator (e.g., 1 or 0) or a soft indicator (e.g., 0, 0.1, 0.2..., 1), and it indicates the likelihood of a LOS path existing between the TRP and WTRU 102 or along the PRS. The LOS indicator can be associated with a TRP or PRS resource ID (e.g., an index). WTRU 102 can receive the LOS indicator from the network (e.g., by TRP and / or resource ID). In some embodiments, WTRU 102 can determine the LOS indicator (e.g., by TRP and / or resource ID) based on measurement results.
[0114] As used in this article, the phrases “ID” and “index” can be used interchangeably.
[0115] In some embodiments, the WTRU location can be expressed using any of altitude, latitude, geographic coordinates, and / or local coordinates. Altitude, latitude, geographic coordinates, and / or local coordinates can be used interchangeably.
[0116] As used in this article, “pre-configured,” “configured,” “pre-configured,” and “configured” can be used interchangeably.
[0117] Configuration of RS for positioning In some representative embodiments, WTRU 102 can receive PRS and / or SRS configurations for positioning purposes from a network (e.g., LMF). The LMF can forward the PRS and SRS configurations to the gNB, allowing the gNB to schedule PRS transmissions and / or SRS receptions at the TRP, TP, and / or RP.
[0118] PRS Configuration In some representative embodiments, the PRS configuration may include information indicating at least one of the following parameters: number of symbols, transmit power, number of PRS resources included in the PRS resource set, PRS silence mode (e.g., the silence mode may be expressed via a bitmap), periodicity, PRS type (e.g., periodic, semi-persistent, or aperiodic), time slot offset for periodic PRS transmission, vertical shift of the PRS mode in the frequency domain, time slot during repetition, repetition factor, RE (resource element) offset, comb pattern, comb size, spatial relationship, QCL information of the PRS (e.g., QCL target, QCL source), number of PRUs, number of TRPs, absolute radio frequency channel number (ARFCN), subcarrier spacing, desired RSTD, uncertainty in desired RSTD, starting physical resource block (PRB), bandwidth, BWP ID, number of frequency layers, start / end time of PRS transmission, PRS on / off indicator, TRPID, PRS ID, cell ID, global cell ID, PRU ID, and / or applicable time window. WTRU 102 can apply PRS configuration if the current time is within the applicable time window.
[0119] SRS configuration for positioning In some representative embodiments, the SRS (SRSp) and / or SRS configuration used for positioning may include information indicating at least one of the following: resource ID; comb offset value, cyclic shift value; starting position in the frequency domain; number of SRSp symbols; shift of the SRSp in the frequency domain; frequency hopping mode; SRSp type (e.g., aperiodic, semi-persistent, or periodic); sequence ID used to generate the SRSp, or other IDs used to generate the SRSp sequence; spatial relationship information indicating which reference signal (e.g., DL RS, UL RS, CSI-RS, SRS, DM-RS) or SSB (e.g., SSB ID, SSB cell ID) the SRSp is spatially associated with, wherein the SRSp and DL... RS can be spatially aligned; QCL information (e.g., QCL relationship between SRSp and other reference signals or SSBs); QCL type (e.g., QCL type A, QCL type B, QCL type C, QCL type D); resource set ID; list of SRSp resources in the resource set; transmit power related information; path loss reference information, which may include indexes of SSB, CSI-RS, or PRS; periodicity of SRSp transmission; and / or spatial information, such as spatial direction information of SRSp transmission (e.g., beam information, transmission angle) and / or spatial direction information of DL RS reception (e.g., beam ID, angle of arrival for receiving DL RS).
[0120] Measurement In some representative embodiments, the WTRU and / or gNB can perform timing-based measurements.
[0121] For example, RSTD can be defined by the time difference of arrival between PRS transmitted from a reference TRP and a target TRP. WTRU 102 can be configured with a reference TRP index and a target TRP index. WTRU 102 can be configured with a PRS resource index for performing measurements. WTRU 102 can determine the time of arrival from (e.g., the target) TRP based on one or more PRS resources associated with (e.g., the target) TRP. In another example, RSTD can be defined as the time difference of arrival between a reference PRS transmitted from a TRP and a target PRS transmitted from a TRP.
[0122] For example, "UE Rx-Tx time difference" can refer to the difference between the arrival time of a reference signal transmitted by the TRP and the transmission time of a reference signal transmitted from the WTRU 102. The WTRU Rx-Tx time difference can be associated with the PRS resource ID and / or the SRSp resource ID.
[0123] In some representative embodiments, the WTRU and / or gNB can perform phase-based measurements.
[0124] For example, RSCP (RS carrier phase) can be defined as the carrier phase measurement of a PRS. RSCP difference (RSCPD) can be defined as the difference in carrier phase measurements between two PRS resources.
[0125] In some representative embodiments, the WTRU and / or gNB can perform power-based measurements. For example, the RSRP per path can be defined as the RSRP per path in the case where the WTRU 102 observes a multipath channel during the measurement. The WTRU 102 can determine the RSRP of the DL RS resource. The RSRP or RSRPP can be reported using units of dBm and / or the relative power difference compared to a reference (e.g., the RSRP of the first path, such as in dB).
[0126] In some representative embodiments, the WTRU and / or gNB can perform impulse response-based measurements. The channel impulse response (CIR) can be associated with the RS configuration.
[0127] For example, the channel impulse response (e.g., for N paths) can be defined by the following equation: ,in h k ( t )and τ k These are time-varying complex-valued coefficients (e.g., from...) a + bj Expression, in which , respectively representing the second for the first kThe channel impulse response and delay are measured along a single path. The δ function is defined as: δ ( t ) = 1, for t = 0; and δ ( t ) = 0 (, for t ≠ 0. For simplicity, we assume the coefficient is constant over time (e.g., ...). h k ( t ) = h k ). WTRU 102 can report the information of each path k to the network. h k and τ k WTRU 102 can report the number of paths N to the network. Alternatively, WTRU 102 can receive the number of paths k for each path from the network and / or that number of paths. h k and τ k .
[0128] For example, WTRU 102 can obtain the CIR from the network. The network can indicate the PRS configuration, such as the PRS resource ID associated with the CIR. For example, the CIR can be associated with a resource ID. In this case, WTRU 102 can determine that the CIR is derived based on measurements made on the PRS resource associated with that ID. Alternatively, WTRU 102 can determine that the channel along the PRS transmit direction or the PRS receive direction corresponds to the CIR.
[0129] For example, the CIR can be associated with the TRP ID. In this case, WTRU 102 can determine that the CIR represents the channel between the TRP and WTRU 102.
[0130] For example, a CIR can be associated with more than one TRP or PRS resource ID. In this case, WTRU 102 can determine that the channel between the TRP and WTRU 102 corresponds to the CIR. Alternatively, WTRU 102 can determine that the channel along the transmit direction or the receive direction of the PRS associated with the ID corresponds to the CIR.
[0131] For example, more than one CIR can be associated with a single parameter from the PRS configuration (e.g., TRP ID, PRS resource ID, frequency layer ID). For instance, WTRU 102 can receive information from the network about two CIRs associated with a TRP (e.g., receiving information from the network). and Alternatively, the WTRU 102 can report information to the network about more than one CIR associated with a PRS configuration (e.g., TRP ID, PRS Resource ID) based on measurement results. Since the WTRU 102 or the network can observe different channel characteristics, for example, based on the AoA of DL RS or UL RS, there can be more than one CIR associated with a PRS configuration.
[0132] In some embodiments, CIR can be represented by a delay distribution or a power delay distribution. A power delay distribution can be defined as a set of delay and power distributions, such as... and ,in p k It can correspond to the first k The relative power at each path compared to the first path. Delay distribution can be defined as a set of delays... Its indication is higher than p threshold The path delay for each path. WTRU 102 can receive this from the network. p threshold The delay distribution is derived from the power delay distribution.
[0133] For example, WTRU 102 can receive instructions from the network on how to generate CIR, PDP, and / or DP based on timing, phase, and / or power measurements. In one example, WTRU 102 can send a request to the network to receive instructions on which methods to use to generate CIR, Power Delay Distribution (PDP), and / or Delay Distribution (DP) based on measurements made by WTRU 102. For example, WTRU 102 can receive a message from the network (e.g., via LPP, RRC, MAC-CE, DCI) indicating the PRS resource index and associated measurement type (e.g., RSTD, AOA) to be used to generate CIR, PDP, and / or DP. In one example, WTRU 102 can receive instructions from the network to generate CIR, PDP, and / or DP.
[0134] In some representative embodiments, WTRU 102 can send the measurement results in the report to the network (e.g., LMF, gNB) via semi-static (e.g., LPP, RRC) or dynamic messages (e.g., UCI, UL MAC-CE).
[0135] As used in this article, the phrases “PRS”, DL-RS (e.g., CSI-RS, DM-RS, TRS) and SSB are used interchangeably.
[0136] In some representative embodiments, the WTRU can obtain the CIR from the network.
[0137] For example, WTRU 102 can receive CIR, PDP, and / or DP from the network based on reported measurement results. For example, WTRU 102 can receive a request for measurement results from the network. This request can instruct WTRU 102 on the positioning method (e.g., DL TDOA) used to determine the desired measurement result (e.g., RSTD of DL-TDOA). WTRU 102 can report the measurement results to the network. WTRU 102 can receive CIR from the network based on the measurement results reported by WTRU 102.
[0138] For example, WTRU 102 can include information related to PRS configuration (such as PRS resource ID) in the measurement report. WTRU 102 can receive at least one of the reported PRS resource IDs associated with the CIR obtained from the network.
[0139] For example, WTRU 102 can receive requests to report measurement results periodically in a configured periodic manner. In another example, WTRU 102 can report measurement results non-periodically. In yet another example, WTRU 102 can be configured to report in a semi-persistent manner, such as periodically reporting during a configured duration or time window. WTRU 102 can be configured with start and end times (e.g., absolute time, symbol index, slot index, frame index, SFN). WTRU 102 can be configured with durations according to the number of symbols, slots, frames, subframes, milliseconds, or other transmission time intervals (TTIs).
[0140] For example, WTRU 102 can determine whether to activate or deactivate semi-persistent reporting or semi-persistent measurement (e.g., measurement during a time window) based on signaling from the network (e.g., MAC-CE). In another example, WTRU 102 can determine whether a semi-persistent PRS transmission or SRSp transmission is activated or deactivated by the network based on signaling (e.g., MAC-CE).
[0141] For example, WTRU 102 may send a request to the network for periodic auxiliary information (e.g., containing CIR) or non-periodic auxiliary information. In the WTRU request, WTRU 102 may include information indicating any of the following: (i) the PRS configuration (e.g., TRP ID, PRS resource ID) for requesting the CIR, DP, or PDP associated with the indicated PRS configuration; (ii) the maximum and / or minimum number of taps for the CIR, DP, and / or PDP; (iii) the requested / required CIR, DP, and / or PDP; and / or (iv) the requested periodicity or duration of the CIR, DP, and / or PDP provisioned from the network.
[0142] For example, WTRU 102 may be requested by the network to report CIR, DP, and / or PDP. WTRU 102 may determine CIR, DP, and / or PDP based on measurements taken on the configured PRS. In such a request, WTRU 102 may receive information from the network indicating any of the following: (i) PRS configuration (e.g., TRP ID, PRS Resource ID), wherein the network requests CIR, DP, or PDP associated with the indicated PRS configuration; (ii) the maximum and / or minimum number of taps for CIR, DP, and / or PDP; (iii) the requested / required CIR, DP, and / or PDP; and / or (iv) the periodicity and / or duration of CIR, DP, and / or PDP reported by WTRU 102.
[0143] In some representative embodiments, WTRU 102 can determine, and / or identify, requests for CIR, DP, and / or PDP based on channel conditions. For example, if a hard or soft LOS indicator associated with a PRS resource or TRP is NLOS or below a configured threshold, WTRU 102 can determine to send a request for CIR, DP, and / or PDP from the network. In another example, if a condition is met (e.g., a hard or soft LOS indicator associated with a PRS resource or TRP is NLOS or below a configured threshold), WTRU 102 can determine CIR, DP, and / or PDP based on measurements performed by WTRU 102.
[0144] Contents of measurement results and location reports In some representative embodiments, WTRU 102 may receive (e.g., from the network) requests to report its location and / or measurement results (e.g., those performed on the PRS). WTRU 102 may report (e.g., to the network) information indicating any of the following: (i) PRS ID; (ii) TRP ID; (iii) Cell ID; (iv) ARFCN; (v) PRS Resource ID; (vi) PRS Resource Set ID; (vii) Frequency Layer ID; (viii) Timestamp indicating when the measurement was performed or when the report was made; (ix) RSTD associated with the PRS Resource ID for each path in the multipath; (x) RSRP associated with the PRS Resource ID for each path in the multipath; (xi) Phase measurement results (e.g., RSCP, RSCPD) for each path in the multipath; (xii) Uncertainty information of the measurement results (e.g., expressed in range, such as ±2). us(xiii) Quality information of the measurement result (e.g., indicating whether the indicated measurement result is in 0.1 μs or 0.01 μs); (xiv) TEG (Timing Error Group) associated with the measurement result, PRS Resource ID, and / or PRS Resource Set ID; (xiv) LOS indicator associated with PRS Resource ID and / or TRP ID; (xv) WTRU location (e.g., absolute location with geographic coordinates expressed in x and y coordinates, relative location relative to a reference point (such as TRP, cell center), such as when requested by the network); (xvi) Uncertainty information of the determined WTRU location (e.g., expressed in range, such as ±2 meters) and / or quality information (e.g., indicating whether the indicated WTRU location is in 0.1 meters or 0.01 meters); (xvii) Method used to determine the WTRU location; (xviii) Channel impulse response and associated DL-RS configuration used to determine the CIR; and / or (xix) Information related to the AI / ML model used to determine the location of the WTRU (e.g., association information).
[0145] Artificial intelligence (AI) for location services In some representative embodiments, the AI model can be used by the WTRU to determine its location.
[0146] In some representative embodiments, the ML model can be used by WTRU to determine its location.
[0147] As stated in this article, the phrase “AI / ML model” can be used interchangeably with “AI model” and “ML model”.
[0148] AI for localization: input, output, association In some representative embodiments, AI / ML models can be used to obtain WTRU locations. Figure 3 This is a block diagram illustrating an example of estimating WTRU location using an AI / ML model 300. (See diagram for example.) Figure 3 As shown, the WTRU can provide inputs to the AI / ML model 300, such as measurement results (e.g., timing, phase, power measurement results, such as RSTD, time of flight, ToA, ToD, carrier phase measurement results, carrier phase difference measurement results, RSRP, and RSRP per path), and the WTRU 102 can obtain the WTRU location from the output of the AI / ML model 300. The output of the AI / ML model 300 can be referred to as "inference".
[0149] For example, as input to the AI / ML model 300, if the AI / ML model 300 is associated with or trained with measurements from multiple TRPs, the WTRU 102 can use measurements from more than one TRP or associated TRPs as input to the AI / ML model 300. If the AI / ML model 300 is trained with measurements from more than one TRP, the WTRU 102 can receive instructions or configurations from the network related to identification information (e.g., TRP ID, PRS ID) of the TRPs used for training the AI / ML model 300.
[0150] For example, WTRU 102 can receive from the network an indication that AI / ML model 300 is trained with measurement results from a TRP, and WTRU 102 can receive from the network an indication of identification information about the associated TRP.
[0151] In some representative embodiments, the input to the AI / ML model 300 (e.g., for localization) may include information indicating any of the following: (i) RSRP of PRS resources; (ii) RSRP statistics per PRS resource (e.g., mean, variance, etc.); (iii) maximum or minimum RSRP per PRS resource; (iv) RSRP of PRS resources per path; (v) RSRP of PRS resources per daily line port; (vi) RSCP of PRS resources per path; (vii) RSCP of PRS resources per daily line port; (viii) RSTD and / or RSCPD of PRS resources; (ix) RSTD statistics per PRS resource; (x) maximum or minimum RSTD of PRS resources; (xi) RSTD and / or RSCPD of PRS resources per path; (xii) RSTD and / or RSCPD of PRS resources per daily line port; (xiii) Time of arrival for each PRS resource; (xiv) Time of arrival for each PRS resource per path; (xv) Arrival time per PRS resource per port; (xvi) Statistical measure of arrival time per PRS resource; (xvii) Maximum or minimum arrival time per PRS resource; (xviii) CIR estimated based on DL-RS (e.g., PRS, CSI-RS, DM-RS), where CIR can be associated with one or more TRPs; (xix) PDP estimated based on DL-RS (e.g., PRS, CSI-RS, DM-RS), where CIR can be associated with one or more TRPs; and / or (xx) DP estimated based on DL-RS (e.g., PRS, CSI-RS, DM-RS), where CIR can be associated with one or more TRPs.
[0152] In some representative embodiments, an input mask can be executed.
[0153] For example, inference can be performed by inputting a subset of measurement results into one or more AI / ML models 300 while masking (e.g., inputting 0 or NULL values) any remaining measurement results. WTRU 102 can receive a minimum number of CIR measurement results to be input into the AI / ML model 300 for inference.
[0154] For example, AI / ML model 300 may allow WTRU 102 to mask a subset of CIR measurements (e.g., reporting only the first N paths for a CIR measurement and masking the remaining paths for the same measurement). Because there is a one-to-one mapping between acceptable CIR measurements and model input dimensions (e.g., the number of acceptable CIR measurements to AI / ML model 300 is the same as the number of CIR measurements simultaneously input to the AI / ML model), WTRU 102 may not provide PRS resource set numbers as input to the AI / ML model. WTRU 102 may receive additional validity conditions along with the trained AI / ML model 300. Examples of these validity conditions include: (i) presenting a minimum of N (e.g., N=3 CIR measurements) CIR measurements; (ii) presenting a minimum of X (e.g., X=128 paths) paths for each CIR measurement; and / or (iii) presenting CIR measurements associated with a PRS set where PRS-RSRP is above a threshold.
[0155] For example, the input dimension of the AI / ML model 300 may be smaller than the acceptable set of CIR measurement results. In this case, the WTRU 102 can input the PRS resource ID along with the CIR measurement results when performing inference operations. The WTRU 102 can then present or mask the measurement results based on validity conditions received along with the trained AI / ML model 300.
[0156] For example, WTRU 102 can receive a mask pattern from the network indicating which measurements should be used as input to AI / ML model 300 to generate inference. The mask pattern can be part of a PRS configuration or an AI / ML-related configuration. For example, AI / ML model 300 may require CIR estimates from N TRPs, where each CIR can be associated with a TRP. WTRU 102 can receive a mask pattern from the network indicating which CIRs to use. For example, if N=3, WTRU 102 can have 3 CIRs, each corresponding to a corresponding TRP. WTRU 102 can receive a mask pattern or bitmap indicating which CIRs to use (e.g., [1 10] corresponds to TRP1 and TRP2). WTRU 102 can replace CIRs (e.g., for TRP3) with 0 or null values as indicated, as input to AI / ML model 300.
[0157] Figure 4 This is a block diagram illustrating an example hierarchical structure of PRS configuration. (Example:) Figure 4 As shown, PRS parameters can be organized in a hierarchical manner 400. Parameters associated with higher layers can be used by parameters at lower layers. For example, if the frequency layer has the parameter value "comb factor = 2", then the PRS resource set, TRP, and PRS resources under the frequency layer can also use the parameter value "comb factor = 2". Parameters can be organized in a hierarchical manner 400 to reduce signaling overhead from the network.
[0158] In some representative embodiments, the AI / ML model may be associated with a higher-level parameter or a first parameter. WTRU 102 can determine that the AI / ML model can be applied to any parameter at a lower level than the first parameter. WTRU 102 can receive configuration information about the AI / ML model from the network. For example, the AI / ML model may be associated with frequency layer #1. WTRU 102 can determine that parameters associated with frequency layer #1 can also be associated with the AI / ML model. Therefore, WTRU 102 can determine the WTRU location by using measurements obtained from a TRP that utilizes or uses frequency layer #1 to transmit a PRS as input to the AI / ML model.
[0159] In some representative embodiments, WTRU 102 may explicitly (e.g., via model ID) or implicitly determine the association between configuration parameters (e.g., TRP ID) and AI / ML models. For example, if WTRU 102 receives an indication from the network to use AI / ML-based positioning based on TRP configuration, WTRU 102 may (e.g., implicitly) determine the AI / ML model to use based on TRP configuration (e.g., TRPID indicating which model WTRU 102 should use).
[0160] In some representative embodiments, WTRU 102 can be configured to associate an AI / ML model with one or more RS configurations. For example, the configuration may include one or more parameters of PRS and / or SRS. WTRU 102 can determine the AI / ML model used for inference and / or performance monitoring and / or validation based on the active RS configuration.
[0161] In some representative embodiments, the association between the configuration and the model indicates the inputs that WTRU 102 should use for the AI / ML model. For example, if the AI / ML model is associated with TRP ID#1, TRP ID#2, and TRP ID#3, WTRU 102 may determine to use measurements taken from PRS emitted from TRP ID#1, TRP ID#2, and TRP ID#3 as inputs to the AI / ML model for inference.
[0162] In some representative embodiments, WTRU 102 may be configured with an AI / ML model, which is trained using measurements from a TRP and is associated with a cell. This indicates that the model can be used anywhere within the cell, and WTRU 102 requires measurements from a TRP within the cell as input to obtain the WTRU location.
[0163] In some representative embodiments, the AI / ML model may be associated with at least one or a combination of the following: (i) parameters in the PRS configuration (e.g., PRS ID, cell ID, TRP ID, frequency information such as frequency layer ID, ARFCN); (ii) parameters in the SRS or SRSp configuration (e.g., SRS sequence ID); (iii) cell ID; (iv) an area that can be defined by more than one cell ID, and the area can be associated with an ID (e.g., area ID); (v) multiple TRPs or cell IDs indicating that the AI / ML model can be used with measurements made to associated TRPs or cell IDs; (vii) an ID related to an RS or signal (e.g., CSI-RS ID, DM-RS ID, SSB ID); (viii) input information and / or its type; (ix) output information and / or its type; (x) model version information; (xi) model format information and / or its type; (xii) AI / ML capabilities (e.g., AI / ML localization, AI / ML beam management), which can be defined at a finer granularity (e.g., AI / ML direct localization, AI / ML assisted localization, AI / ML spatial domain beam prediction, AI / ML temporal domain beam prediction, etc.); (xiii) Vendor information (e.g., vendor ID); (xiv) Applicable scenarios and / or configurations (e.g., frequency range, antenna configuration, site information, cell ID, etc., where the AI / ML model is suitable for inference, e.g., based on matching with the training scenario); (xv) Computational complexity (e.g., FLOPs, the level of preprocessing / postprocessing that may be required before deploying the model and / or before using the model for inference); (xvi) Complexity (e.g., the number of real-valued model parameters, the number of real-valued operations on the model, the number of models whose complexity is expressed by any of the above parameters, and / or IDs); (xvii) Size (e.g., a fixed size associated with some models, ≤ 5 MB, ≤ 50 MB, ≤ 100 MB, etc.); (xviii) Performance metrics (e.g., accuracy, bias, variance, and / or the number / ID associated with any of them); (xix) functions and / or sub-functions and / or use cases and / or sub-use cases (e.g., what (sub)functions and / or (sub)use cases are applicable to a particular model, such as CSI / BM / location); (xx) what type is applicable to a particular model (e.g., UE-side model, UE portion of a two-side model, such as encoder at the UE for CSI compression use cases, decoder at the gNB); and / or (xxi) model monitoring methods.
[0164] In some representative embodiments, any parameter may be included as part of the model ID and / or transmitted to WTRU 102 and / or the network as part of the model metadata. In some other embodiments, such information may be inherently known by another endpoint (e.g., due to the association between a particular model and a particular parameter).
[0165] In some representative embodiments, the model switching can be fully visible to the network, such as where WTRU 102 can notify the network of the model ID to which it has switched.
[0166] In some representative embodiments, model switching can be partially visible to the network, such as where WTRU 102 can send an indication to the network that a switching has already occurred without necessarily including the model ID of the target model. This can be applied to function-based LCM. If the network is performing performance monitoring on the UE-side model, the network can send an indication of poor performance to WTRU 102 (e.g., based on low throughput or inaccurate location estimation measured at the network). This indication can trigger WTRU 102 to switch to a different model to improve performance. After the switch, WTRU 102 can indicate a simple switching indication to the network without a model ID. For example, WTRU 102 can provide the network with information only at the function level. For example, WTRU 102 can transmit a function ID to the network and / or indicate a specific function to the network as part of an initial capability exchange (e.g., WTRU 102 can indicate in the RRC that it has AI / ML capabilities for positioning, and / or WTRU 102 can transmit a function ID associated with AI / ML positioning capabilities). WTRU 102 may have multiple models associated with AI / ML positioning capabilities; however, it may not indicate any switching within those models to the network (e.g., model switching within a function is transparent to the NW). If no model at WTRU 102 has sufficiently good performance and WTRU 102 must fall back to a legacy procedure, WTRU 102 may subsequently send an indication to the network that it has fallen back to a legacy procedure for that function (e.g., non-AI / ML, such as GNSS, RAT-dependent positioning methods, WiFi, or sensor-based positioning methods). Such model switching, transparent to the network, may also apply if performance monitoring is at the UE. For example, WTRU 102 may determine that the performance of a particular model is below a pre-configured threshold and switch to another model within the function. For function-based LCM, WTRU 102 may not report any model switching within the function, or WTRU 102 may simply send a switching indication to the network without including the model ID of the target model.
[0167] In some representative embodiments, WTRU 102 may receive configuration information of the AI / ML model (e.g., configurations that associate the model with a PRS configuration). For example, WTRU 102 may receive the association between a PRS configuration (e.g., a TRP ID) and an AI / ML model. WTRU 102 may, for example, receive a CIR, DP, and / or PDP associated with a TRP from the network.
[0168] Availability of the model at WTRU In some representative embodiments, WTRU 102 may be pre-configured or configured with multiple AI / ML models. WTRU 102 may have more than one model associated with different combinations of PRS configurations or general configurations. WTRU 102 may be configured with AI / ML models, wherein each AI / ML model may be associated with a different PRS configuration.
[0169] Figure 5 This is a block diagram illustrating an example association between AI / ML models and PRS configuration information. For example... Figure 5 As shown, a first (e.g., high-level) AI / ML model can be associated with a frequency layer (e.g., an AI / ML model associated with frequency layer ID#1), and / or a second (e.g., low-level) AI / ML model can be associated with a TRP (e.g., an AI / ML model associated with TRP ID#1). WTRU 102 can be configured with N AI / ML models under frequency layer #1, where each model is associated with a TRP (e.g., AI / ML model #1 with TRP #1, AI / ML model #2 with TRP #2, etc.). WTRU 102 can be configured to measure PRS transmitted from the N TRPs.
[0170] For example, WTRU 102 can be configured with criteria (e.g., maximum RSRP) to determine which AI / ML model to use based on measurement results. For example, WTRU 102 can determine which one or more AI / ML models to use, associated with configured PRS parameters. WTRU 102 can determine the reporting of information about the AI / ML model (e.g., model ID) and / or the PRS configuration associated with that model (e.g., TRP ID).
[0171] For example, WTRU 102 can indicate to the network whether more than one AI / ML model is being used. For example, WTRU 102 can report, along with its determined WTRU location (e.g., to the network), information indicating multiple AI / ML models (e.g., more than one AI / ML model ID) and / or multiple PRS configuration parameters, where each parameter is associated with an AI / ML model. In another example, WTRU 102 can indicate to the network, along with its determined WTRU location, whether a single or multiple AI / ML models are being used to determine the WTRU location.
[0172] For example, WTRU capabilities can indicate the type of AI / ML model or the availability of AI / ML models at WTRU 102. In one example, WTRU 102 can send an indication to the network indicating the association of the available AI / ML models at WTRU 102 with the PRS configuration. For example, WTRU 102 can indicate to the network that WTRU 102 has two AI / ML models, each associated with PFL#1 and PFL#2 respectively. In another example, WTRU 102 can indicate to the network that WTRU 102 has two AI / ML models, each associated with cell #1 and cell #2 respectively. WTRU 102 can indicate to the network that it is capable of training AI / ML models that can be associated with TRP and / or cells. In yet another example, WTRU 102 can indicate to the network that the AI / ML models are capable of using localization measurements as their input to generate inference (e.g., WTRU location).
[0173] In some representative embodiments, WTRU 102 may determine to send WTRU capability and / or auxiliary information to the network to indicate the AI / ML model at WTRU 102. For example, WTRU 102 may send an indication to the network of the level of association of the AI / ML model that WTRU 102 can train. For example, WTRU 102 may indicate to the network that WTRU 102 can train an AI / ML model associated with a cell. In this case, WTRU 102 can use measurements obtained from PRS transmitted from the TRP in the serving cell and / or neighboring cells to train the AI / ML model. WTRU 102 may indicate that the AI / ML model is associated with at least one of the PRS configuration parameters (e.g., frequency layer). In this case, WTRU 102 can use measurements obtained from the PRS associated with the configured frequency layer to train the AI / ML model.
[0174] In some representative embodiments, WTRU 102 may have AI / ML models, which are pre-configured and / or configured by the network. For example, WTRU 102 may determine one or more AI / ML models from pre-configured weights. WTRU 102 may send a request to the network to receive weights or parameters associated with the AI / ML models. WTRU 102 may indicate the type of AI / ML model (e.g., an AI / ML model that can be associated with a cell, TRP, etc.).
[0175] For example, WTRU 102 can indicate to the network (e.g., via WTRU capabilities and / or auxiliary information) the location where the AI / ML model at WTRU 102 is trained. The location information associated with the AI / ML model can include any of the following: (i) the absolute or relative location of the WTRU 102 where the AI / ML model is trained, (ii) the location of the TRP for which its PRS is used for training, (iii) the TRP ID, (iv) the cell ID, (v) the region ID, and / or (vi) the region ID. For example, the region ID or area ID can be configured and / or pre-configured by the network. In another example, WTRU 102 can indicate the region information for receiving the PRS used for training.
[0176] For example, WTRU 102 can indicate to the network the training completion level and / or the amount of training data used to train the AI / ML model.
[0177] In some representative embodiments, WTRU 102 may be hard-coded, pre-configured, or programmed with an AI / ML model. For example, WTRU 102 may have weights in its memory, and WTRU 102 may be able to adjust the weights so that after training with ground truth and measurements, the AI / ML model can produce the desired output. In another example, the weights may be hard-coded. In yet another example, WTRU 102 may receive configuration (e.g., weights) from the network. The reception of configuration may only occur at configured events (e.g., when WTRU 102 first connects to the network, and / or when WTRU 102 transitions from IDLE to CONNECTED).
[0178] In some representative embodiments, a single AI / ML model may not be generalized to all possible scenarios, conditions, and / or configurations for which WTRU 102 can be expected to operate. In this case, WTRU 102 may have more than one AI / ML model for any AI / ML-enabled feature and / or function, where each AI / ML model may be associated with applicable conditions. For example, a first set of models may be pre-configured and / or pre-loaded in WTRU 102. For example, a second set of models may be downloaded by WTRU 102 from a network and / or an external server. For example, WTRU 102 may have a third set of models, which may be the result of fine-tuning and / or updating of the first and / or second sets of models.
[0179] General process Figure 6 This is a signaling diagram illustrating an example of communication used for WTRU-based AI / ML positioning. Figure 6 In this context, the WTRU102 can communicate with gNB 180 and LMF 600 (e.g., in CN 106 / 115).
[0180] like Figure 6 As shown, WTRU 102 can determine at 602 to send WTRU capability information to the network, such as information indicating the AI / ML models available at WTRU 102. WTRU 102 can receive one or more PRS configurations from the network at 604. WTRU 102 can receive at 606 a request from the network to perform localization using the AI / ML models at WTRU 102.
[0181] Based on the availability of the AI / ML model at WTRU 102, WTRU 102 can send an accept or reject message (e.g., ACK / NACK) to the network at 608, indicating whether WTRU 102 can or cannot perform AI / ML-based localization. The model can be available at WTRU 102 under either of the following conditions: (i) WTRU 102 receives parameters of the AI / ML model from the network; and / or (ii) WTRU 102 is configured and / or pre-configured with an AI / ML model, and the training of the AI / ML model is completed at a pre-configured level (e.g., 50%, 90%, producing the configured accuracy level).
[0182] For example, if WTRU 102 can determine its location using the AI / ML model at WTRU 102, then WTRU 102 can determine a measurement result (e.g., CIR) at 612 on the PRS received at 610. Based on this measurement result, WTRU 102 can report its location to the network at 616, along with uncertainty information (e.g., indicating the range of fluctuations at the determined location).
[0183] In some representative embodiments, in the first step, WTRU 102 can use an AI / ML model to determine the WTRU location. WTRU 102 may be pre-configured and / or configured with an AI / ML model. WTRU 102 can train and use the AI / ML model at WTRU 102. The AI / ML model may be associated with a configuration (e.g., cell, TA, PFL, and / or PRS configuration). WTRU 102 may determine, based on measurements and / or auxiliary information from the network, that the AI / ML model at WTRU 102 may need to be turned off or deactivated. WTRU 102 may determine, based on conditions (e.g., configuration such as PRS configuration, measurement results), to switch to a model that can be associated with a different configuration. WTRU 102 may send an indication to the network indicating information about the new model (e.g., model ID, association information). If WTRU 102 falls back to a fallback positioning method, WTRU 102 may indicate to the network that a fallback positioning method was used.
[0184] For example, WTRU 102 may be using the first model, "Model-1". One or more conditions may change (e.g., WTRU 102 moves to a new cell, PRS changes, TA changes, PFL changes). WTRU 102 can determine that Model-1 is unsatisfactory (e.g., based on model performance monitoring). WTRU 102 can obtain information for selecting a different model. If some criteria are met (e.g., model switching and / or selection conditions), WTRU 102 can select a model; otherwise, it will roll back. WTRU 102 can send information indicating changes or rollback in the model (e.g., including the model ID, function ID, and / or association with applicable conditions).
[0185] Model ID and its relationship with configuration In some representative embodiments, the model ID may be associated with the PRS configuration.
[0186] For example, a model ID can be used to identify an AI / ML model. The WTRU 102 can be configured and / or indicated by the network (e.g., gNB, LMF) to have a model ID. The model ID can be associated with a logical AI / ML model or a physical AI / ML model. In the first example, the model can be identified based on offline collaboration between the WTRU 102 and the network. In the second example, the model can be identified based on air signaling between the WTRU 102 and the network. For example, the air signaling can be initiated by the WTRU 102 or the network.
[0187] In some representative embodiments, a model ID-based LCM can be used. In some representative embodiments, a function-based LCM can be used. For example, one or more AI / ML models can be abstracted using functions within AI / ML enabling features. For example, a function ID and / or a configuration ID can be used instead of a model ID. In some representative embodiments, any one of a function ID, a configuration ID, and / or a model ID can be used (e.g., to identify an AI / ML model).
[0188] For example, WTRU 102 can be pre-configured and / or configured with a set of PRS configuration parameters associated with the model ID.
[0189] For example, WTRU 102 can determine from association rules that model ID#N is associated with PRS frequency layer ID#1 (e.g., bandwidth of 100MHz). Therefore, WTRU 102 can determine the measurement results made on the PRS associated with PRS frequency layer ID#1, and use the measurement results (e.g., timing, power, carrier phase) as input to the AI / ML model with model ID#N to determine the WTRU location as the output of the AI / ML model.
[0190] For example, WTRU 102 can determine from association rules that model ID#N is associated with TRP IDs#1, #2, and #3 with respect to frequency layer ID#1. WTRU 102 can determine that any PRS transmitted from TRPs with IDs #1, #2, and #3 is measured, and use the measurement results as input to an AI / ML model with model ID#N to determine the WTRU location as the output of the AI / ML model.
[0191] For example, WTRU 102 can determine the association between model ID#N and cell ID#1 from association rules. WTRU 102 can determine to measure any PRS transmitted in the cell with ID#1 and use the measurement results as input to an AI / ML model with model ID#N to determine the WTRU location as the output of the AI / ML model.
[0192] For example, WTRU 102 can receive an instruction to change the model associated with the PRS configuration. For instance, WTRU 102 may be pre-configured with an AI / ML model of ID #1, having a PRS frequency layer ID #1. WTRU 102 can receive configuration messages (e.g., RRC messages, LPP messages) from the network to associate the AI / ML model of ID #2 with frequency layer ID #1. WTRU 102 can also receive AI / ML model parameters (e.g., weights) for the AI / ML model of ID #2 from the network. Alternatively, WTRU 102 can determine that the AI / ML model of ID #2 pre-configured at WTRU 102 will be associated with frequency layer ID #1.
[0193] Figure 7 This is a signaling diagram illustrating an example of communication used to update the association between the AI / ML model and the PRS configuration.
[0194] like Figure 7 As shown, WTRU 102 indicates the availability of the AI / ML model at WTRU 102 to the LMF in the first message at 704. The LMF, in the second message at 704, indicates the association between the AI / ML model and the PRS configuration to WTRU 102. In the third message, the LMF indicates the update of the association between the AI / ML model and the PRS configuration to WTRU 102 at 706. In response to the second message, WTRU 102 can indicate to the network at 708 that the association update is complete. WTRU 102 may need to send a response because it may need to determine whether the newly associated AI / ML model has been trained and is ready for use (e.g., the third message). Afterwards, the network can send the parameters (e.g., weights) of the newly associated AI / ML model to WTRU 102.
[0195] In some representative embodiments, multiple model IDs may be associated with a PRS configuration.
[0196] For example, multiple model IDs can be associated with PRS configurations. WTRU 102 can be pre-configured with such association information. In another example, WTRU 102 can receive configuration messages that associate model IDs with PRS configurations.
[0197] For example, a PRS configuration can be associated with multiple model IDs. For instance, model IDs #N1, #N2, and #N3 can be associated with cell ID #1.
[0198] For example, a PRS configuration set can be associated with multiple model IDs. For instance, TRP IDs #1, #2, and #3 for frequency layer ID #1 can be associated with model IDs #N1, #N2, and #N3.
[0199] After the WTRU 102 is configured with multiple AI / ML models, the WTRU 102 can determine which model to use based on at least one configured criterion.
[0200] In some representative embodiments, WTRU 102 may implicitly and / or determine which AI / ML model to use based on explicit messages.
[0201] For example, WTRU 102 can determine which model to use based on the configured PRS parameters. For example, WTRU 102 can receive a request or configuration message to use the indicated AI / ML model for localization.
[0202] For example, WTRU 102 can indicate to the network the availability of the AI / ML model at WTRU 102. The availability of the AI / ML model at WTRU 102 can be indicated as "yes" or "no". In another example, WTRU 102 can indicate the PRS configuration in a message to indicate that WTRU 102 has an AI / ML model associated with the indicated PRS configuration.
[0203] For example, WTRU 102 can receive a request from the network to perform AI / ML positioning. This request can indicate the PRS configuration that the AI / ML model used for positioning should be associated with. For instance, WTRU 102 can receive a request from the network to use AI / ML positioning, and this request can include frequency layer ID #1. This can instruct WTRU 102 to use the AI / ML model associated with frequency layer ID #1.
[0204] For example, WTRU 102 can receive a request to perform AI / ML localization using a model that WTRU 102 does not possess. For instance, WTRU 102 can receive a request to use AI / ML localization associated with, for example, a serving cell with cell ID #1. WTRU 102 may not have an AI / ML model associated with the serving cell. In this case, WTRU 102 can determine to initiate training of an AI / ML model that can be used for the serving cell by performing a measurement of the PRS.
[0205] Figure 8 This is a signaling diagram illustrating an example of communication used to initiate AI / ML-based location-based communication. (Example:) Figure 8As shown, in the first message, WTRU 102 indicates the availability of the AI / ML model at WTRU 102 to the network at point 802. In the second message, the network sends a request for AI / ML-based localization to WTRU 102 at point 804. Based on this request and the availability of the AI / ML model at WTRU 102, WTRU 102 determines to perform AI / ML-based localization and reports the obtained WTRU location to the network at point 806.
[0206] If the model associated with the PRS configuration is available, WTRU 102 can determine to use AI / ML-based localization. If the model is available, WTRU 102 can determine to use AI / ML-based localization based on the associated PRS configuration.
[0207] Figure 9 This is a signaling diagram illustrating another example of communication used to initiate AI / ML-based location-based communication. (See diagram below.) Figure 9 As shown, in the first message, WTRU 102 indicates the availability of the AI / ML model at WTRU 102 to the network at 902. In the second message, the network sends a request for AI / ML-based localization to WTRU 102 at 904. Based on this request and the availability of the AI / ML model at WTRU 102, WTRU 102 can determine to perform AI / ML model training at 906 and report the training results to the network at 908. The network can send authorization information to WTRU 102 at 910, and WTRU 102 can determine to use AI / ML-based localization (e.g., based on associated PRS configuration) after receiving authorization and send information indicating the WTRU location at 912.
[0208] Model monitoring and switching at the network when WTRU 102 has an AI / ML model In some representative embodiments, WTRU 102 can be configured to report measurement results and / or indicators to the network, allowing the network to monitor or switch any model at WTRU 102. WTRU 102 can receive indicators from the network regarding which AI / ML model is used for localization purposes.
[0209] For example, WTRU 102 may send measurement results and / or indicators to the network to assist the network in monitoring the model at the WTRU 102 side and / or determining whether a model switching at the WTRU 102 side may be necessary. These may include any of the following: (i) performance reports; (ii) raw model output; (iii) processed output data; (iv) requests to switch AI / ML models; and / or (v) measurement reports.
[0210] For example, the report may include performance results for the currently active model. For example, the report may include performance results for another model that WTRU 102 may have tested (e.g., a better-performing model). WTRU 102 may also include the model ID of the better-performing model in the report.
[0211] For example, the report may include the raw model output from the currently active AI / ML model at WTRU 102.
[0212] For example, the report may include processed data of the output from the currently active AI / ML model at WTRU 102.
[0213] For example, a report could include a simple request to switch to another model. WTRU 102 might be able to determine that it needs to switch to another model and the target model to switch to, but it may still need an ACK from the network before activating the new model for inference.
[0214] For example, the report may include measurements reported by WTRU 102, such as WTRU 102 velocity, rotational movement, position, localization and / or changes thereof (e.g., NW may evaluate measurements reported from conventional non-AI / ML localization methods against the raw and / or processed output of the ML model at WTRU 102 and determine that the ML model performance is not good enough and a model switch at WTRU 102 may be necessary).
[0215] For example, the report may include measurement results reported by WTRU 102, such as channel measurements and / or changes thereof. Channel measurements may include any of the following: CSI parameters (e.g., CQI, PMI, RI), Doppler, Doppler spread, channel coherence time, channel coherence bandwidth, cell RSRP, L1-RSRP (e.g., beam RSRP), and / or SINR. For example, if WTRU 102 reports a sudden drop in RSRP, the NW can determine that the drop in channel measurements may be due to a change in location to a poorly covered area. If WTRU 102 also reports the output of the AI / ML positioning model to the NW, the NW can determine that the degradation in channel quality does not match the currently reported location of WTRU 102 and that the AI / ML positioning model at WTRU 102 is performing poorly and may require a model switch at WTRU 102.
[0216] In some representative embodiments, WTRU 102 can be requested by the network to report measurement results and WTRU location for model monitoring purposes.
[0217] For example, WTRU 102 can be configured with a time window during which it performs DL-RS measurements. The network can be configured with time windows so that multiple WTRUs can perform measurements during the configured window duration. WTRU 102 can be configured with a time window having an ID. WTRU 102 can include this ID in its reports to indicate the correlation between measurement results and the time window.
[0218] For example, WTRU 102 can be configured to report measurement results and location information via events. These events can be periodic, trigger-based (e.g., non-periodic reporting), or semi-persistent (e.g., periodic reporting within a configured time window). Events can also be time-based, where WTRU 102 receives from the network an indication of when it should send a report (e.g., a specific time, a relative time relative to a reference). If the configured event occurs, WTRU 102 can send a report.
[0219] Model monitoring or switching at WTRU 102 In some representative embodiments, when WTRU 102 is in a coverage area associated with multiple AI / ML models, WTRU 102 can perform model monitoring and / or switching.
[0220] For example, when WTRU 102 is covered by multiple AI / ML models, WTRU 102 can perform inference from the target or first AI / ML model in addition to performing inference from the source or second AI / ML model.
[0221] In some representative embodiments, WTRU 102 can use a quality indicator offset for model switching. For example, when entering a region corresponding to a target model (e.g., Model 2), WTRU 102 can (e.g., simultaneously) perform a prediction of its location and use a quality indicator (e.g., a range of truth values expressed in meters by a lower and upper bound, such as ±X meters) for both (e.g., source and target) AI / ML models. When the quality indicator of the target model becomes better or smaller than that of the source model due to the quality indicator offset, WTRU 102 can switch to the target model for AI / ML localization.
[0222] In some representative embodiments, WTRU 102 can use monitoring measurement thresholds for model switching: upon entering an area of a target model (e.g., Model 2), WTRU 102 can (e.g., simultaneously) perform location prediction using both (e.g., source and target) AI / ML models. WTRU 102 can receive monitoring measurement thresholds (e.g., RSRP thresholds) associated with the AI / ML model used for localization. If the RSRP of the target TRP (e.g., associated with the target model) becomes the threshold or is better than the RSRP of the source TRP (e.g., associated with the source model), WTRU 102 can switch to the target model for AI / ML localization.
[0223] In some representative embodiments, WTRU 102 may use a network-controlled approach. For example, when entering a region associated with a target model (e.g., Model 2), WTRU 102 may (e.g., simultaneously) perform location predictions using both (e.g., source and target) AI / ML models. WTRU 102 may report both locations and associated quality indicators to the network until the network decides to switch to the target model.
[0224] In some representative embodiments, WTRU 102 can use a quality indicator offset to switch the AI / ML model used for localization. For example, WTRU 102 can determine which AI / ML model to use for localization. WTRU 102 can receive the AI / ML models (e.g., both source and target models) used for localization, which can predict the quality indicator of the model output and the coverage area of each model. WTRU 102 can receive a quality indicator offset (e.g., QI) from the network. offset WTRU 102 can receive the PRS configuration associated with the source model from the network. WTRU 102 can measure any PRS associated with the source model, input the measurement results into the source model, and estimate the WTRU 102 location and its associated quality indicators (e.g., QI). source WTRU 102 can report the estimated location from the source model to the network. By comparing the estimated location from the source model, WTRU 102 can determine whether it has entered the coverage area associated with the target model. WTRU 102 can determine that it is within the coverage area of the target model, and WTRU 102 can request the network to send the PRS associated with the target model. WTRU 102 can measure the PRS associated with the target model, input the measurement results into the target model, and estimate the WTRU 102 location and its associated quality indicator (e.g., QI). target If the quality indicator of the target model is better than that of the source model by a quality indicator offset (e.g., QI...), then... target <(QI source+ QI offset If the target model is switched to the target AI / ML model for localization, then WTRU 102 can switch to the target model. WTRU 102 can notify the network that it has switched to the target AI / ML model for localization.
[0225] In some representative embodiments, WTRU 102 can monitor and / or switch configured AI / ML models based on KPI measurement results and their monitoring. For example, WTRU 102 can receive one or more combinations of KPIs (e.g., Los / NLoS ID, SNR, SINR, RSRP, measurement uncertainty, delay spread, and / or Doppler spread) associated with the configured AI / ML model. In one example, a KPI can be associated with a model ID associated with the AI / ML model. KPIs can be model-specific (e.g., associated with a model ID) or common to more than one model (e.g., the same KPI associated with more than one model ID).
[0226] For example, WTRU 102 can be configured with conditions and / or thresholds associated with each KPI. These conditions and thresholds can indicate the conditions under which WTRU 102 can determine to monitor other configured models for switching or switching to another model.
[0227] For example, the WTRU 102 can be configured by the network with KPI measurement configurations (for example, measurement time, periodicity, frequency, bandwidth, etc.) to indicate measurement parameters. In one example, the measurement configuration can (e.g., also) be associated with KPI measurement conditions and / or thresholds. In one example, the WTRU 102 can measure a KPI (e.g., based on the KPI measurement configuration) and determine, based on this, whether to continue using the current AI / ML model for the same or different KPI measurement results, switch to another AI / ML model, or terminate the process. The WTRU 102 can make this determination based on configured conditions that allow the measured KPI to be higher or lower than a configured threshold.
[0228] For example, a configured AI / ML model with model ID #1 can be associated with a KPI (e.g., associated with one or more TRPs) and RSRP with a LosS / NLoS ID. Model ID #1 can be further associated with a KPI measurement time t, which has: a first measurement period p1, associated with a first threshold (such as threshold1_LoS / NLOS and threshold1_RSRP); and a second measurement period p2, associated with a second measurement threshold (such as threshold2_LoS / NLOS and threshold2_RSRP).
[0229] For example, a configured AI / ML model with ID#2 can be associated with a KPI delay spread having conditions and thresholds (e.g., threshold2_delayspread). Furthermore, the measurement configuration can be associated with a second measurement configuration associated with the same model ID#1. In this example, WTRU 102 can use the first measurement configuration (e.g., period p1) to measure the KPI LoS / NLoS ID and RSRPP, and determine to continue using model #1 but utilize the second measurement configuration (e.g., period p2) based on either of the following conditions: (i) the determined LoS / NLoS ID associated with one or more TRPs is lower than the (pre)configured threshold1_LoS / NLOS; and / or (ii) the measured RSRP associated with one or more received RSs is lower than the (pre)configured threshold1_RSRP.
[0230] WTRU 102 can then decide to switch the AI / ML model to model #2 based on the measurement configuration (e.g., period p2) and any of the following conditions: (i) the determined LoS / NLoS ID associated with one or more TRPs is lower than the configured threshold2_LoS / NLOS; (ii) the measured RSRP associated with one or more received RSs is lower than the configured threshold2_RSRP; and / or (iii) the measured delay spread is lower than the configured threshold2_delayspread, etc.
[0231] If WTRU 102 determines, based on the measured KPIs, that none of the (pre)configured AI / ML models are suitable for localization, then WTRU 102 can determine to terminate the AI / ML-based localization process.
[0232] For example, WTRU 102 can also be configured with (pre-)configured priorities for AI / ML models. The indicated priorities can be categorical (e.g., high, medium, low) or numerical (e.g., 0, 0.1, ..., 1). Priorities associated with AI / ML models (e.g., associated with model IDs) can indicate the order in which WTRU 102 can select models. For example, an AI / ML model associated with time-based measurements as input can have one priority, while an AI / ML model associated with power-based measurements as input can have another priority. Priorities can also be a way of indicating a preference for certain AI / ML models.
[0233] In some embodiments, WTRU 102 may not be configured with model priorities, and WTRU 102 may determine priorities. WTRU 102 may determine model priorities based on any of the following: (i) model inputs; (ii) DL-RS configuration; (iii) WTRU 102 capabilities; (iv) WTRU 102 mobility status; and / or (v) KPI measurement results.
[0234] For example, WTRU 102 can prioritize models based on their inputs. Model inputs can, for example, determine the performance of an AI / ML model. A model with RSRP inputs may have higher error compared to a model with timing measurement results (such as TDoA) as inputs.
[0235] For example, WTRU 102 can assign a priority to a model based on the DL-RS (such as SSB, CSI-RS, DL-PRS, etc.) configuration used for localization. For example, a configuration with a certain bandwidth allocation may favor certain models (such as models that have timing measurement results as input), and WTRU 102 can determine which models to prioritize over others (such as assigning a priority above a threshold).
[0236] For example, WTRU 102 can assign a priority to a model based on WTRU capability information. For instance, WTRU 102 with a large number of antenna elements can determine accurate angle measurements. In such a scenario, WTRU 102 can determine to prioritize models with angle-based measurements (e.g., AoD) as input.
[0237] For example, WTRU 102 can assign a priority based on whether WTRU 102 is moving. If WTRU 102 is moving, it can determine to prioritize AI / ML models trained under dynamic conditions.
[0238] For example, WTRU 102 can assign a priority to a model based on a KPI. For instance, if WTRU 102 measures a DL-RS with a multipath component count exceeding a threshold, WTRU 102 can determine to prioritize models that utilize the multipath measurement results. If WTRU 102 measures a channel SNR exceeding a configured threshold, WTRU 102 can assign a priority to a model. In one example, since one or more KPIs can be associated with an AI / ML model (e.g., having its model ID), WTRU 102 can determine to prioritize models whose measured KPIs exceed a configured threshold.
[0239] In some representative embodiments, WTRU 102 may be configured by the network with different priority updates, or the priority of the AI / ML model may be determined and updated during the localization process. As channel conditions may change throughout the localization process (e.g., due to changes in WTRU 102 mobility or channel conditions), WTRU 102 may receive or determine the priority of the updates and perform subsequent procedures based on them. WTRU 102 may determine the update priority based on any of the following conditions: (i) the measured KPIs (e.g., associated with the AI / ML model) between multiple time instances (e.g., measurement timings) are below a (pre)configured threshold; (ii) the WTRU location difference between multiple time instances is above a (pre)configured threshold; and / or (iii) the input difference of the AI / ML model between multiple time instances is above a (pre)configured threshold.
[0240] In one example, when WTRU 102 wants to switch AI / ML models (e.g., based on KPIs), WTRU 102 can consider the order of suitability of models within a set of models, based on model priority. For example, WTRU 102 can determine the KPIs associated with the AI / ML model to be switched to from the set of available models with the highest priority.
[0241] For example, WTRU 102 may report information indicating any of the following: (i) the measured KPI; (ii) the determined model priority (e.g., associated with the AI / ML model and / or model ID); (iii) the updated model priority (e.g., associated with the AI / ML model and / or model ID); and / or (iv) a timestamp (associated with, for example, the measured KPI, the priority determination).
[0242] For example, WTRU 102 can send measurement reports to networks (e.g., LMF, gNB) via semi-static (e.g., LPP, RRC) and / or dynamic messages (e.g., UCI, UL MAC-CE).
[0243] In some representative embodiments, WTRU 102 may receive a request for AI / ML-based localization from a network. WTRU 102 may be configured with multiple AI / ML models, such that each model is associated with one or more KPIs and thresholds, model priorities, and KPI measurement configurations (e.g., time, periodicity, etc.). WTRU 102 measures the KPIs associated with the configured AI / ML models (e.g., based on the KPI measurement configuration) and determines a subset of suitable AI / ML models for localization based on the measured associated KPIs being higher than a (pre)configured threshold. WTRU 102 may select an AI / ML model from the determined subset of suitable models based on configured model priorities (e.g., the model with the highest priority). WTRU 102 may report the measured KPIs and / or the selected AI / ML model and initiate the localization process using that model.
[0244] In some embodiments, WTRU 102 can measure KPIs (e.g., those associated with a selected AI / ML model) based on a KPI measurement configuration, and determine to switch models if the measured KPIs fall below a (pre)configured threshold. WTRU 102 can measure KPIs associated with different configured AI / ML models based on a KPI measurement configuration, and select a model based on the measurement results and model priority.
[0245] In some embodiments, if WTRU 102 determines that the measured KPIs associated with all configured AI / ML models are below a (pre)configured threshold, WTRU 102 may terminate the localization process.
[0246] Turning accessibility information on and off In some representative embodiments, WTRU 102 may send requests for PRS configuration information. WTRU 102 may send requests to validate AI / ML models, train and / or retune models. WTRU 102 may include a reason for the request, such as training, retuning, and / or validation. In one example, WTRU 102 may indicate the duration of the PRS transmission for the reason indicated in the request. For example, WTRU 102 may send the request via LPP and / or RRC messaging.
[0247] Predict and determine the AI / ML model to use In some representative embodiments, WTRU 102 can predict and / or determine which AI / ML model to use. For example, WTRU 102 can use an AI / ML model that, taking into account the input measurements, is capable of predicting both the location of the WTRU and the optimal model at that location.
[0248] Figure 10 This is a block diagram of an example AI / ML model 1000, showing the best AI / ML model for predicting WTRU location and WTRU position. Figure 10 In this context, the AI / ML model 1000 can receive a set of measurement results as input, and the AI / ML model 1000 can output a prediction (e.g., inference). The prediction may include the location of the WTRU (e.g., x, y, and z coordinates) and information indicating (e.g., the optimal AI / ML model for the predicted WTRU location).
[0249] For example, AI / ML model 1000 can consist of two sub-models trained separately (e.g., a regression model for location and a classification model for the best model). It can also be the same model with two sets of outputs trained end-to-end for both location regression and best model classification.
[0250] In some representative embodiments, each model can be trained for a specific operating region. The operating region of the model can be associated with a cell or TRP set. Figure 11 Figure 1100 shows examples of the operational regions of different AI / ML models. For example... Figure 11 As shown, the operating region of each model can contain smaller internal sub-regions (e.g., Figure 11 In the darker shaded areas, the model performs best in predicting location within that sub-region. Each model can also have external sub-regions (e.g.,...). Figure 11 In a sub-region (with a lighter shade), the model can still predict localization, but it may not be the optimal model (e.g., providing the most accurate prediction). For example, model switching can occur in the outer sub-region. In one example, the sub-model can be associated with a PRS configuration. In another example, the sub-model can be associated with a model ID.
[0251] Figure 12 This is a flowchart illustrating an example of a model switching process based on WTRU trajectories. (Example) Figure 12 As shown, the model switching process from model A to model B can occur as WTRU 102 moves along a specified trajectory. Figure 12 In this scenario, assume WTRU initially uses model A at location 1. The model's output gives the position at that location (e.g., X1, Y1, Z1), and the best model is predicted as "model A". Since this model is currently in use, switching is not necessary.
[0252] As WTRU 102 moves to position 2, the model predicts the position (e.g., X2, Y2, Z2), and model A remains the best choice. Although model B is also effective at this position, model A is still the optimal option.
[0253] Now, WTRU 102 moves to location 3, where model A is still valid, but model B is the optimal model. Using its current model (model A), WTRU 102 predicts its current location (e.g., X3, Y3, Z3), and the optimal model predicted is now model B. WTRU 102 now initiates the model switching process as explained in this paper. Note that if model B were available to WTRU 102 at this location, it could use that model to obtain a more accurate localization prediction.
[0254] By the time WTRU 102 reaches position 4, the model switching process has been completed, the model predicts the position (e.g., X4, Y4, Z4), and the best model has been determined to be model B.
[0255] For example, a more advanced type of AI / ML can be used, which can work with the current best model to predict possible future models. Figure 13 This is a block diagram of an example AI / ML model 1300, showing the best AI / ML model for predicting the location of WTRU 102, the WTRU location, and future AI / ML models. Figure 12 In the example of the WTRU 102 trajectory, WTRU 102 can determine the optimal future model at position 2. At this position, while model A remains the optimal model, there is a high probability that model B will be optimal in the future. Therefore, while still using model A, WTRU 102 can prepare to switch to model B by (e.g., in advance) requesting model B from the gNB. In this way, WTRU 102 can switch models immediately when it detects that model A is no longer the optimal model (e.g., at position 3). This can accelerate model switching and minimize the duration for which WTRU 102 uses a suboptimal model. If model B is already available, WTRU 102 can also use model B at position 3 to obtain a better position prediction.
[0256] Training metrics In some representative embodiments, WTRU 102 can communicate with the network when (e.g., a certain) AI / ML model is not available at WTRU 102.
[0257] Figure 14 This is a signaling diagram illustrating an example of communication used to initiate AI / ML-based localization when an AI / ML model is unavailable. Figure 14In the first message, the WTRU indicates the availability of the AI / ML model at WTRU 102 to the network at 1402. In the second message, the network sends a request for AI / ML-based localization to WTRU 102 at 1404. In this request, the network can indicate any parameters that the AI / ML model should be associated with (e.g., cell ID). WTRU 102 can determine that the AI / ML model is not available at WTRU 102. In this case, WTRU 102 can start training the AI / ML model at 1406. After training is complete, WTRU 102 can send the training results (e.g., performance metrics) to the network at 1408. Examples of training results include any of the following: (i) Mean Absolute Error (MAE); (ii) Mean Squared Error (MSE); (iii) Root Mean Squared Error (RMSE) (e.g., where WTRU can be evaluated based on the dataset used for training and / or the dataset used for validation); (iv) X percentile (%) cumulative distribution function (CDF) error, where X can range from 0 to 100; and / or (v) number of samples (e.g., used for training). In some embodiments, RMSE can be defined as the root mean squared error between the true value and the inference generated by the AIML model.
[0258] For training, MSE can be used as the loss function. Other training metrics can include RMSE, 2D distance, MAE, and CDF(90%) (e.g., the probability that 90% of instances have an MSE performance below a certain value is 90%) or CDF(80%) error value.
[0259] Based on the reported performance metrics, WTRU 102 can receive authorization (e.g., permission) at 1412 to perform AI / ML-based localization using a trained AI / ML model. WTRU 102 can use the AI / ML model to determine its location and report the determined WTRU location at 1412.
[0260] For example, WTRU 102 can receive instructions from the network to use a specific model. In this case, WTRU 102 can receive an explicit indication of the model ID from the network (e.g., via DCI, MAC-CE, RRC, LPP).
[0261] Common benefits In some representative embodiments, WTRU 102 can maintain localization accuracy by seamlessly switching AI / ML models or switching back to different localization methods (e.g., RAT-dependent localization methods, RAT-independent localization methods) using the model switching, verification, and model monitoring techniques described herein.
[0262] Validation based on measurement results from the network In some representative embodiments, the AI / ML model can be associated with a cell. When WTRU 102 moves to a different cell, WTRU 102 can query (e.g., request) PRU measurement results and their location. The network can provide the requested information to WTRU 102, and WTRU 102 can determine whether the current AI / ML model (e.g., the model used in a previous cell) is still usable. After providing auxiliary information, WTRU 102 can report whether it is still using the same model.
[0263] General principles In some representative embodiments, WTRU 102 may use an AI / ML model for localization. In one example, the AI / ML model may be associated with a cell or area (e.g., defined by one or more cells). WTRU 102 may determine the cell or area serving WTRU 102 and use the AI / ML model associated with the serving cell or area. When WTRU 102 moves out of the serving cell or area and into another cell or area, WTRU 102 may determine whether to use the same or a different AI / ML model based on the results of verification. The verification described herein relies on measurements, ground truth, and / or WTRU location determined by other WTRUs or PRUs. WTRU 102 may obtain the measurements, ground truth, and / or WTRU location determined by other WTRUs or PRUs from the network based on a request made by WTRU 102.
[0264] The content of the request and the conditions for sending the request. In some representative embodiments, WTRU 102 may send requests to the network for measurement results and / or inference performed by other WTRUs (e.g., PRUs) in the same cell where WTRU 102 is located. If the PRUs in the cell are configured with AI / ML models, WTRU 102 may want to compare the inference generated by the PRUs with the inference generated by the AI / ML models at WTRU 102 to determine whether to change the AI / ML models, update the AI / ML models, or fall back to the default localization method.
[0265] In some representative embodiments, WTRU 102 may request measurement results and / or inferences performed by other WTRUs and / or PRUs in neighboring cells. WTRU 102 may instruct itself to request the cell ID from which the measurement results are obtained. WTRU 102 may receive the cell ID from the network as part of a positioning method (e.g., to measure PRS transmitted from neighboring cells for DL-TDOA). Based on this request, WTRU 102 may receive measurement results and / or inferences performed by other WTRUs and / or PRUs from the network. WTRU 102 may receive auxiliary information from the network that associates location with measurement results, indicating the measurement results performed at that location. WTRU 102 may use the measurement results sent by the network to determine whether to use the same AI / ML model that WTRU 102 is using, a different AI / ML model, or a backtracking positioning method.
[0266] Figure 15 This is a system diagram illustrating an example of measurement forwarding between PRU 1502 and WTRU 102. Figure 15 In this context, PRU 1502 can be another WTRU 102. For example, WTRU 102 can send a request to the network for measurement results and / or location information performed by another WTRU 102 (e.g., PRU). Location information can be specific (e.g., expressed by geographic coordinates) or region-based (e.g., cell ID, region ID, area ID).
[0267] exist Figure 15 In this document, WTRU 102 can receive positive responses to requests from the network, and WTRU 102 can receive inference performed by PRU 1502. The inference performed by PRU 1502 can be based on an AI / ML model. WTRU 102 can periodically receive inference performed by PRU 1502 from the network. WTRU 102 can receive inference using PRS measured by PRU 1502 from TRP1504. For example, PRU 1502 can measure PRS1, PRS2, and PRS3 at times t=T1, t=T2, and t=T3, respectively. As described herein, "inference" and "measurement result" are used interchangeably. As described herein, "PRU" and "WTRU" are used interchangeably. As described in this article, “WTRU” and “PRU” can be used interchangeably with “target WTRU” and “reference WTRU”, respectively, such as when target WTRU 102 needs to use the measurement results and / or inference generated by reference WTRU 102 to check the performance of its AI / ML model.
[0268] After the network receives a request from WTRU 102, WTRU 102 can receive the requested information from the network as supplementary information (e.g., via LPP, RRC). In the request sent to the network, WTRU 102 may include information indicating any of the following to indicate the desired information from the network, PRU 1502, and / or WTRU 102.
[0269] For example, the request could specify the content of the auxiliary information, where WTRU 102 could indicate whether it requires measurement results, inferences, and / or truth values from another WTRU 102 or PRU in the auxiliary information from the network. In one example, WTRU 102 could indicate that the measurement results provided in the auxiliary information should be associated with the inferences provided in the auxiliary information. In another example, WTRU 102 could indicate that the measurement results provided in the auxiliary information should be associated with the truth values provided in the auxiliary information. If WTRU 102 requests measurement results, it could also request measurement results associated with a specific PRS configuration (e.g., PRS Resource ID, PFL ID, TRP ID) that are relevant to the measurement results.
[0270] For example, a request can indicate the type of measurement result associated with the truth value. WTRU 102 can indicate the PRS configuration associated with the measurement result in the request. For example, WTRU 102 can indicate a request for the Channel Impulse Response (CIR) associated with the truth value. WTRU 102 can indicate the number of TRPs or TRP IDs in the request and request the network to provide a CIR for each TRP.
[0271] For example, a request may indicate desired location information. The requested PRU location may indicate that WTRU 102 expects to obtain inference and measurement results made from a specific location and / or inference made by a PRU located at that specific location. The requested location information (e.g., a specific location expressed in geographic coordinates, or area information such as cell ID, area ID, zone ID), wherein WTRU 102 requests measurement results and / or inference associated with the indicated location information; and / or the desired location of the PRU or WTRU, which may be indicated by the zone ID, location, or area, wherein the area may be a collection of cells indicated by the cell ID. WTRU 102 may include this information to request inference made by a PRU or WTRU in a specific area. For example, an SSB ID, indicating the PRU location or the area covered by an SSB. Another example is an area ID, where the area may be defined by a collection of cell IDs or geographic coordinates.
[0272] For example, the request could indicate the WTRU location (e.g., geographic coordinates expressed as x and y coordinates) to assist the network in selecting a PRU close to WTRU 102. WTRU 102 could indicate the method used to obtain the WTRU location (e.g., GNSS, RAT-dependent positioning method). In another example, WTRU 102 could indicate area information (e.g., area ID, zone ID) in which WTRU 102 is located.
[0273] For example, a request can instruct the AI / ML model at PRU or WTRU 102 to have its associated PRS configuration (e.g., PRS resource ID, frequency layer ID). WTRU 102 can send a request to the network so that the input (e.g., measurement results) to the AI / ML model (e.g., measurement results) at PRU and / or WTRU 102 is made by the requested PRS configuration.
[0274] For example, a request may indicate the desired time instance and / or periodicity of auxiliary information delivery from the network (e.g., every Monday at 9 p.m., every 20 seconds). For example, WTRU 102 may send a request to the network for the periodic provisioning of auxiliary information at the desired periodicity. WTRU 102 may indicate the duration of the auxiliary information provisioning (e.g., up to 4 hours).
[0275] For example, a request could indicate information about the AI / ML model configured on WTRU 102 (e.g., the cell ID associated with the AI / ML model, the PRS configuration associated with the AI / ML model, the model ID). WTRU 102 can include such information, allowing the network to select a PRU or WTRU 102 that generates inference using a similar or identical AI / ML model. For example, WTRU 102 could include a cell ID (e.g., serving cell ID, resident cell ID, ID of a cell previously served by WTRU 102) to indicate that WTRU 102 expects to obtain inference performed by the AI / ML model associated with that cell ID. The cell ID included in WTRU 102 could be the ID of a cell associated with the AI / ML model used by WTRU 102. The network can search for a PRU or WTRU that uses the AI / ML model associated with that cell ID.
[0276] For example, a request may indicate the number of PRUs or WTRUs, the number of locations, the data size (e.g., expressed as the number of samples, bytes, measurement instances, time instances, or timestamps), and the number of samples (e.g., 100 times of provisioning of auxiliary information).
[0277] For example, a request can indicate to WTRU 102 whether it requests a truth value from the same location or a different location. If WTRU 102 indicates that it wants a truth value from the same location or region, then WTRU 102 can receive the truth value and associated measurement result, along with the truth value label quality indicator, from the same location or region (e.g., the same WTRU 102, the same PRU) in the auxiliary information at each time the auxiliary information is prepared. If WTRU 102 requests a different truth value, then WTRU 102 can receive the different truth value in the auxiliary information at each time the auxiliary information is prepared, and associate the truth value and measurement result, along with the truth value label quality indicator, from the same location (e.g., the same WTRU 102, the same PRU). As used herein, the phrases "truth value" and "truth value label" are used interchangeably. As used herein, the phrases "truth value label quality indicator," "quality indicator," and "truth value quality indicator" are used interchangeably.
[0278] For example, a request could indicate a repeatability factor for the provisioning of auxiliary information for each truth value. WTRU 102 might want to request repeated provisioning of auxiliary information so that WTRU 102 can determine consistency in truth values and associated information (e.g., measurement results), or determine how to process or combine them. For example, WTRU 102 could determine to generate a set of truth values and associated information (e.g., measurement results) based on N sets of truth values and associated information by averaging N truth values and N measurement results (e.g., average CIR, average RSRP, average RSTD, average time of flight). For example, WTRU 102 could indicate a repeatability factor N in the request. If, for example, N=2, WTRU 102 could receive truth values and associated information (e.g., measurement results, truth value label quality indicators) from the network at two points in time, where each point in time could have a different timestamp indicating that the truth value measurement result was made at two different time instances (e.g., different days, different hours, different minutes). The repeatability factor can be applied when WTRU 102 receives different truth values. For example, WTRU 102 can receive a first truth value for two occasions and a second truth value for two occasions, followed by two occasions of receiving the first truth value.
[0279] For example, a request can indicate the quality of the truth value, where WTRU 102 can indicate the quality of the WTRU 102 or PRU location estimate. For instance, the uncertainty in the PRU location can be expressed as an uncertainty range (e.g., ±0.5 meters). In another example, WTRU 102 can request the type of WTRU 102 (e.g., WTRU 102, PRU). This is because PRU locations can be verified and known by the network, while WTRU locations (e.g., non-PRU locations) may not be verified by the network. Therefore, in terms of the quality of the truth value, PRU locations may be more reliable than WTRU locations. Another example of truth value quality is the method used to determine the truth value (e.g., GNSS, RAT-dependent positioning methods, RAT-independent positioning methods (e.g., sensors, WIFI) NW-verified locations).
[0280] For example, a request can indicate the type of truth label quality indicator associated with a truth value (e.g., hard indicator, soft indicator).
[0281] For example, a request could indicate a threshold for the truth label quality indicator associated with the truth value, and WTRU 102 could expect the network to provide location information for WTRU 102 or PRU whose associated truth label quality indicator is higher than or equal to that threshold.
[0282] For example, the request may indicate an AI / ML model ID, enabling the network to forward measurements, inferences, and / or truth values associated with the requested AI / ML model ID.
[0283] For example, a request can indicate the reason for the request. Potential reasons include a mobility event, the expiration of a timer associated with the AI / ML model at WTRU 102, the accuracy of the AI / ML model output falling below a configured threshold, variations in the AI / ML model output (e.g., standard deviation, variance, range) exceeding a configured threshold, and / or the failure to meet validity conditions associated with the AI / ML model.
[0284] For example, the WTRU 102 can include more than one parameter (e.g., PRS configuration, cell ID) in a request. The WTRU 102 can have multiple AI / ML models, and each model can be associated with different parameters (e.g., different cell IDs, different PRS configurations). Requests from the WTRU 102 can be made in a single message or in separate messages. For example, the WTRU 102 can indicate a preference for the truth label quality indicator in a first message. The WTRU 102 can send a request for truth in a second message.
[0285] Forwarding measurement results and inferences from the network In some representative embodiments, WTRU 102 may receive (e.g., from the network) measurement results and / or inferences performed by the PRU. In some representative embodiments, WTRU 102 may receive information determined by the network. This information may include any of the following: (i) measurements performed by the PRU (e.g., timing, power, phase, CIR); (ii) the DL-RS resource ID and / or RS configuration associated with the measurements (e.g., PRS resource set ID, frequency layer ID, TRP ID); (iii) training-related information of the AI / ML model used by the PRU (e.g., the ground truth used by the PRU, the quality of the ground truth labels, the amount of data used); (iv) inference performed by the PRU and information about the AI / ML model associated with the generated inference (e.g., model ID, measurements used as input to the AI / ML model, DL-RS configuration associated with the AI / ML model, parameters related to the PRS configuration); (v) the location of the PRU (e.g., the geographic coordinates of the PRU); (vi) location, where the location and associated measurements may suggest what WTRU 102 measured at that location; (vii) PRU ID; (viii) The timestamps associated with the measurement results and / or inference (e.g., absolute time, symbol index, slot index, subframe index, frame index, SFN), where the timestamps can indicate when the inference or measurement results were made; (ix) the number of truth values to be provided; and / or (x) the number of times the truth values and associated information are prepared.
[0286] For example, WTRU 102 can receive measurement results, inference, and / or correlation information performed by PRUs as auxiliary information from the network via LPP and / or RRC messaging. As an example, WTRU 102 can receive measurement results and / or inference performed by multiple PRUs from the network. WTRU 102 can receive measurement results and / or inference from the network in one or more sets of auxiliary information, such as each set containing measurement results and / or inference performed by one PRU. For example, WTRU 102 can determine to train an AIML model at WTRU 102 using any one of the measurement results, inference, and / or correlation information.
[0287] For example, WTRU 102 can receive measurement results and / or inference from multiple PRUs or WTRUs sequentially from the network. For example, WTRU 102 can be configured with a set of PRUs or WTRUs identified by indexes. If periodic delivery of PRUs and / or WTRU 102 inference and / or measurement results is configured, WTRU 102 can receive from the network an indication of the delivery order of auxiliary information indicating the measurement results and / or truth values. WTRU 102 can receive a set of indexes, and each index can correspond to a WTRU 102 or PRU, indicating the delivery order. For example, if there are three PRUs with indices 1, 2, and 3, WTRU 102 can receive an index set [1 2 3] indicating the delivery order of measurement results and inference from the PRUs. In one example, WTRU 102 can be configured to receive measurement results and / or inference in ascending order of the indices.
[0288] Periodicity, termination, or initiation of measurement results and inference forwarding In some representative embodiments, WTRU 102 may periodically receive measurement results and / or inferences performed by the PRU, such as auxiliary information from the network. WTRU 102 may be configured by the network to periodically transmit auxiliary information. For example, WTRU 102 may send a request to the network (e.g., via RRC and / or LPP) to initiate the transmission of auxiliary information. For example, WTRU 102 may send a request to the network to stop the transmission of auxiliary information.
[0289] For example, WTRU 102 may indicate to the network information indicating any of the following: (i) when the forwarding of inference and / or measurement results should begin (e.g., indicated by absolute time, SFN, time slot or frame number, or relative timing in seconds); (ii) whether the inference and / or measurement results should be forwarded periodically or aperiodically; and / or (iii) the duration of the forwarding of inference and / or measurement results, such as where the duration can be expressed in terms of symbols, time slots, frames, time (e.g., seconds, minutes, hours), etc.
[0290] For example, when WTRU 102 determines to terminate the positioning session (e.g., the positioning session begins when WTRU 102 receives the PRS configuration to initiate positioning) or to fall back to a fallback positioning method (e.g., DL-TDOA) that does not require measurement results and / or inference by the PRU or other WTRUs, WTRU 102 may determine to send a request to terminate the transmission of auxiliary information.
[0291] AI / ML model validation In some representative embodiments, WTRU 102 can validate AI / ML models.
[0292] For example, once WTRU 102 receives the inference and / or measurement results made by PRU, WTRU 102 can determine to use the inference and / or measurement results made by PRU to validate or confirm the AI / ML model at WTRU 102.
[0293] For example, WTRU 102 can send a request for authentication to the network. WTRU 102 can determine to send the request based on events. This request can be sent before WTRU 102 queries the network for auxiliary information (such as measurement results and / or inferences made by the PRU). Examples of events include mobility-based events (such as changes in cell or area), periodic events (such as periodic authentication), measurement-based events (such as RSRP of PRS or SSB falling below a threshold), and / or time-based events (such as timers associated with AI / ML models at WTRU 102 expiring). For example, WTRU 102 can be configured to have timings during which WTRU 102 should perform authentication. The timing can be periodic, and WTRU 102 can be configured to have opportunistic periodicity. The timing can be semi-persistent, such as where WTRU 102 can be configured with time windows during which WTRU 102 should perform periodic authentication.
[0294] For example, WTRU 102 can send a request to the network for measurements and / or inference performed by the PRU. WTRU 102 can be configured with an AI / ML model applicable to a region, such as a region consisting of multiple cells. If WTRU 102 moves from its current cell (e.g., cell 1) to another cell (e.g., cell 2), where the region includes both cells 1 and 2, WTRU 102 can determine to send a request to the network for measurements and / or inference performed by the PRU. WTRU 102 may want to check whether the AI / ML model can produce the desired performance compared to the measurements and / or inference performed by the PRU. WTRU 102 can request ground truth and associated measurements to train the AI / ML model at WTRU 102. WTRU 102 can include a reason for the request in the request.
[0295] For example, WTRU 102 can determine the completion of the AI / ML model at WTRU 102 after a mobility event has occurred. An example of a mobility event could be a change of serving cell or when WTRU 102 moves from its current serving cell to a neighboring cell. For instance, WTRU 102 can determine the completion of a mobility event when it receives cell identity information (e.g., cell ID) of a new cell (e.g., a neighboring cell of the current serving cell) from the network.
[0296] Figure 16 This is a signaling diagram illustrating an example of model validation following a mobility event. Figure 16 In this scenario, WTRU 102 moves to a new cell after mobility event 1602 and obtains information about the new serving cell from the network (e.g., gNB 180) at 1604. WTRU 102 may receive a location request from the network at 1606. The location request may include a location method (e.g., AI / ML-based location) that WTRU 102 should use to determine its location. WTRU 102 may send a request to the network at 1608 for auxiliary information (e.g., measurement results and / or inferences performed by the PRU). Once WTRU 102 receives the measurement results and / or inferences performed by the PRU at 1610, WTRU 102 can determine whether to verify or validate the AI / ML model it possesses at 1612. WTRU 102 may report the validation results to the network at 1614. Examples of validation results may include the error between the AI / ML model output at WTRU 102 and the received truth label or inference done by the PRU, or a hard indicator indicating 1 for successful validation or 0 for unsuccessful validation. After reporting the validation, WTRU 102 may receive configuration at 1616 for the localization method (e.g., the AI / ML-based localization method, which AIM model is used, and configuration related to DL-TDOA).
[0297] Figure 17 This is a signaling diagram illustrating another example of model validation following a mobility event. Figure 17 In this context, WTRU 102 can perform AI / ML-based localization without verifying the AI / ML model at WTRU 102 after a mobility event. For example, after a mobility event at 1702 and after obtaining information about a new serving cell from the network (e.g., gNB 180) at 1704, WTRU 102 can determine at 1706 to send WTRU capability information and / or the status of the AI / ML model at WTRU 102 to the network. For example, WTRU 102 can indicate auxiliary information of the AI / ML model (e.g., the AI / ML model can be associated with a cell) to the network as auxiliary information. In another example, WTRU 102 can indicate the capabilities of the AI / ML model at WTRU 102 (e.g., the number or type of input the AI / ML model can accept). After sending auxiliary information and / or WTRU capability information, WTRU 102 may receive a location request at 1708, and / or receive one or more configurations related to the location method (e.g., which AI / ML model(s) to use, model ID, configuration messages related to DL-TDOA) at 1710.
[0298] For example, WTRU 102 can be configured with a timer for the AI / ML model at WTRU 102. Once the AI / ML model is activated, WTRU 102 can determine when to start the timer for the AI / ML model. If the timer associated with the AI / ML model (e.g., duration) expires, WTRU 102 can determine whether to perform verification or send a request for verification or auxiliary information (e.g., PRU measurement results).
[0299] For example, WTRU 102 can receive a request to perform authentication from the network. Requests from the network or WTRU 102 can be received or sent via any of the following messaging methods: DCI, UCI, MAC-CE, RRC, and / or LPP messaging.
[0300] In some representative embodiments, WTRU 102 may use the measurement results made by PRU and make comparative inferences (e.g., based on PRU measurement results).
[0301] Figure 18 This is a system diagram illustrating an example of verifying measurement results using a PRU. Figure 18 In this context, WTRU 102 can determine the measurement results made by PRU 1502 to generate inference (e.g., WTRU location) using the AI / ML model at WTRU 102. WTRU 102 can determine the difference (e.g., absolute error, squared error, mean squared error) between the generated inference and the inference generated by PRU 1502. WTRU 102 and PRU 1502 may or may not use the same AI / ML model. For example, WTRU 102 can determine the difference between the generated inference and the ground truth (e.g., PRU location) associated with PRU 1502. Examples of errors could be... e = | x̂ - x |², where x̂ and x It is the inference and truth value generated at WTRU 102 (e.g., PRU position).
[0302] In some representative embodiments, WTRU 102 may use the measurement results made by WTRU 102 and make comparative inferences (e.g. based on the measurement results of WTRU 102).
[0303] Figure 19 This is a system diagram illustrating an example of validating measurement results using PRS. Figure 19In this context, WTRU 102 can determine to perform measurements on the PRS and compare the inference generated by the AI / ML model at WTRU 102 with the inference generated by the AI / ML model at PRU 1502. For example, WTRU 102 can be configured to perform measurements on the indicated PRS resource during a configured time window, where parameters associated with the window can be start / end time or duration (e.g., number of symbols, number of frames or subframes, number of time slots). WTRU 102 can determine inference (e.g., WTRU location) based on the measurements of the indicated PRS performed by WTRU 102. WTRU 102 can compare the generated inference with the inference generated by PRU 1502. WTRU 102 can determine the comparison of inference generated based on measurements performed on the same PRS resource. For example, WTRU 102 can determine the error between the inference generated by WTRU 102 and PRU 1502 based on measurements performed on PRS1. Figure 19 In this process, WTRU 102 and PRU measure PRS1, PRS2 and PRS3, and generate their respective inferences using AI / ML models #1 and #X at WTRU 102 and PRU 1502, respectively.
[0304] In some representative embodiments, WTRU 102 may use processed measurement results from the network. For example, WTRU 102 may receive CIR from the network based on measurement results (e.g., power, timing, phase, phase measurement results) reported by WTRU 102.
[0305] Figure 20 This is a signaling diagram illustrating an example of model validation using CIR. For example, WTRU 102 can send WTRU capability information to the network at 2002. WTRU 102 can send a request to the network for auxiliary information (such as measurement results performed by the PRU) at 2004. WTRU 102 can receive PRS configuration from the network at 2006. For example, PRU 1502 can receive PRS configuration from the network. WTRU 102 can receive the PRS transmitted by the network at 2008 and measure the received PRS at 2010. PRU 1502 can receive the PRS transmitted by the network at 2008 and measure the received PRS at 2010.
[0306] PRU 1502 can send a measurement report to the network at 2012. WTRU 102 can send a measurement report to the network at 2014. The network can include a timestamp indicating when the reported measurement result was made. WTRU 102 can receive the CIR associated with the measurement result reported by WTRU 102 at 2016. The network can associate information about the measurement results used to determine the CIR. This information can be at least one parameter from the PRS configuration (e.g., PRS Resource ID, TRP ID, PRSID, TRP location). If the CIR is determined based on the measurement results of a PRU or other WTRU, WTRU 102 can receive location information about the PRU or WTRU.
[0307] WTRU 102 can receive the CIR based on PRU measurement results from the network. PRU 1502 can send inference (e.g., based on PRS measurement results) to the network at 2018. At 2020, WTRU 102 can also receive inference from the AI / ML model at PRU 1502 based on the measurement results made by PRU 1502. WTRU 102 can use the CIR determined based on PRU measurement results as input to the AI / ML model at WTRU 102 and obtain the corresponding inference from the AI / ML model. For example, WTRU 102 can determine an error metric based on this inference and the inference obtained from the AI / ML model at PRU. For example, WTRU 120 can determine the error metric based on the inference obtained from the AI / ML model at WTRU 102 and the ground truth associated with PRU (e.g., PRU location). WTRU 102 can verify the inference obtained from the AI / ML model at 2022. At point 2024, WTRU 102 can send information to the network indicating the status of the AI / ML model used at WTRU 102. For example, the status can reflect the results of validation.
[0308] In some representative embodiments, the error metric used to validate the AI / ML model at WTRU 102 can be defined as follows. WTRU 102 can receive an indication from the network regarding which definition to use. If WTRU 102 autonomously determines the definition, then WTRU 102 can report to the network which metric WTRU 102 used. For example, WTRU 102 can report the inference generated by the AI / ML model at WTRU 102 (e.g., Figure 18 Inference 1) and inference generated by the AI / ML model at PRU (e.g. Figure 18The difference between the inference (X) generated by the AI / ML model at WTRU 102 and the ground truth (e.g., the geographic coordinates of the PRU). For example, WTRU 102 can report the difference between the inference generated by the AI / ML model at WTRU 102 and the ground truth (e.g., the geographic coordinates of the PRU).
[0309] For example, error metrics can be used with any combination of inputs described herein, such as Figure 18 and Figure 19 For example, WTRU 102 can use the AI / ML model at WTRU 102 to determine inference based on the measurements made by WTRU 102 and / or PRU.
[0310] In some representative embodiments, WTRU 102 can report error metrics during verification.
[0311] For example, WTRU 102 can report error metrics during validation. WTRU 102 can be configured by the network to report error metrics periodically. WTRU 102 can report aggregated error metric reports (e.g., average error metrics during the validation period) to the network. WTRU 102 can instruct error metrics to be averaged. If WTRU 102 receives a message to terminate error metric reporting, WTRU 102 can determine to stop reporting error metrics.
[0312] In some representative embodiments, WTRU 102 may receive from the network instructions regarding information that WTRU 102 should use to validate the AI / ML model at WTRU 102. For example, WTRU 102 may receive instructions to use PRU measurement results to generate inference using the AI / ML model at WTRU 102 for validation. For example, WTRU 102 may receive from the network instructions to use measurement results made by WTRU 102 to generate inference using the AI / ML model at WTRU 102.
[0313] In some representative embodiments, if the error determined during validation is less than or equal to a configured threshold, WTRU 102 may determine to use the AI / ML model at WTRU 102. If the error is greater than the configured threshold, WTRU 102 may perform either of the following operations: (i) determine to validate another AI / ML model, if available at WTRU 102; (ii) determine to use a pre-configured backtracking localization method (e.g., a RAT-dependent localization method, GNSS); and / or (iii) determine to train or fine-tune the AI / ML model.
[0314] In some representative embodiments, the threshold (e.g., used to determine error) can be configured by the network along with the configuration of the AI / ML model.
[0315] In some representative embodiments, the threshold can be configured by the network when the WTRU 102 is moved to a new cell.
[0316] In some representative embodiments, the duration and / or amount of data used for verification can be configured. For example, WTRU 102 can be configured with a duration and / or amount of data to be used for verification. For example, WTRU 102 can be configured with a start and / or end time for verification. The start and / or end time can be specified as any of absolute time, slot index, frame index, SFN, and / or other TTIs. The start and / or end time can be specified as a relative timing relative to a certain moment (e.g., after WTRU 102 receives auxiliary information from the network, or after WTRU 102 receives an instruction to initiate verification from the network).
[0317] For example, WTRU 102 can be configured with the amount of data to be used for verification. For example, WTRU 102 can be configured with the number N opportunities to receive auxiliary information from the network. Once WTRU 102 has received N auxiliary information opportunities from the network, WTRU 102 can determine that the verification period is complete.
[0318] rollback mechanism In some representative embodiments, WTRU 102 may determine that AI / ML model-based localization cannot be performed. WTRU 102 may determine to send a message to the network indicating that the localization process has been terminated. For example, a fallback localization method may be configured at WTRU 102, and WTRU 102 may determine to execute the fallback localization method. WTRU 102 may send a message indicating that the fallback localization method is being used.
[0319] Validated WTRU behavior In some representative embodiments, WTRU 102 may determine to report to the network the AI / ML model that WTRU 102 determines to use.
[0320] In some representative embodiments, WTRU 102 may determine (e.g., send to the network) after the verification period information (e.g., model ID) instructing WTRU 102 to use any AI / ML model. For example, WTRU 102 may determine any PRS configuration (e.g., TRP ID, PRS resource ID) and / or general configuration (e.g., cell ID) associated with the AI / ML model (e.g., the model WTRU 102 determines to use). WTRU 102 may instruct the network on a positioning method, such as a RAT-dependent and / or RAT-independent positioning method. For example, if there are any changes to the model used by WTRU 102, WTRU 102 may report model-related information (e.g., model ID, PRS resource ID, TRPID) to the network.
[0321] For example, if WTRU 102 determines that the same AI / ML model was used prior to the verification event, WTRU 102 can indicate to the network whether the same or a different model was used at WTRU 102. For instance, if WTRU 102 used an AI / ML model at cell #X and WTRU 102 moves to cell #Y (e.g., a new cell), WTRU 102 can indicate to the network after verification that the same AI / ML model used in cell #X is used in cell #Y.
[0322] For example, if WTRU 102 determines to train or fine-tune an AI / ML model, WTRU 102 can determine the progress of training (e.g., the start or end of training) and WTRU 102 can indicate to the network that training or fine-tuning is complete, along with model information (e.g., model ID, PRS configuration associated with the model).
[0323] Example of a verification and validation report In some representative embodiments, WTRU 102 may receive a request from the network to send a verification report to the network as a result of verification or validation. WTRU 102 may include information in the report indicating any of the following: (i) a (e.g., hard) indicator indicating whether the verification or validation of the AI / ML model at WTRU 102 was successful; (ii) the result of the verification or validation, including an error metric, such as MSE, between the AI / ML output and the truth or inference performed by the PRU; (iii) any PRS configuration and / or inference used by WTRU 102 or the PRU for verification (e.g., to derive the error metric) or validation; (iv) a timestamp associated with the verification or validation; and / or (v) any measurement results (e.g., RSTD, RSRP, ToF) used for verification and associated information (e.g., the PRS resource ID used to obtain the measurement results).
[0324] Obtain truth values from other virtual and real WTRUs on the network. In some representative embodiments, truth label indicators may be used. For example, a truth label quality indicator may be (e.g., a hard) indicator, such as where values "1" and "0" can indicate whether the associated truth value is suitable or unsuitable for training an AI / ML model, respectively. Alternatively, a truth label quality indicator may be (e.g., a soft) indicator, such as where a value "0.8" indicates a relatively high confidence level when training an AI / ML model using the associated truth value. On the other hand, a value "0.2" can indicate a relatively low confidence level when training an AI / ML model using the associated truth value.
[0325] In some representative embodiments, for soft or hard quality indicators, a value "1" may correspond to a truth value generated by the PRU. For soft quality indicators with values less than 1 and greater than 0, they may be used to indicate the quality of a truth value generated by the WTRU or PRU, such as when the truth value is generated using a RAT-dependent positioning method (e.g., DL-TDOA) or a RAT-independent positioning method (e.g., GNSS). For soft or hard quality indicators with a value equal to 0, they may indicate that a truth value label quality indicator cannot be assigned to a truth value.
[0326] For example, truth label indicators can have ranges (e.g., defined by minimum and maximum values). The minimum and maximum values of soft indicators can be 0 and 1, respectively. The WTRU 102 can be pre-configured or configured by the network with ranges for the indicators. The granularity of the soft indicators can be predefined (e.g., 0.1, 0.01, etc.).
[0327] In another example, the truth label indicator can be a range of values that indicate the uncertainty of the truth value. For example, if the truth value is expressed as a set of coordinates (e.g., x, y), the indicator can be the range of uncertainty for each value, such as x ± 0.2 meters.
[0328] For example, a truth label quality indicator can be associated with each coordinate of the truth value. For instance, for a truth value represented by coordinates (x, y), there can be two indicators, where each of "x" and "y" can be associated with a truth label indicator.
[0329] The truth value label quality indicator indicates the confidence level associated with the truth value location. For example, a truth value obtained via DL-TDOA in a severe NLOS environment can be associated with a low truth value label indicator. On the other hand, a truth value obtained via DL-TDOA in a severe LOS environment can be associated with a high truth value label indicator.
[0330] In another example, the WTRU 102 can be configured with an association table that associates a truth value label quality indicator with a range of uncertainty in the truth value. For example, the truth value indicator could be an index to the configured association table, where the index indicates the uncertainty in the truth value (e.g., ±2 meters).
[0331] In some representative embodiments, the truth label quality indicator requirement may be included in the request (e.g., from WTRU102).
[0332] For example, WTRU 102 can include information indicating a threshold for the truth label quality indicator in a truth value request sent to the network. WTRU 102 can send the request, for example, as a semi-static message (e.g., RRC message, LPP message) or a dynamic message (e.g., MAC-CE, UCI). The threshold indicated by WTRU 102 can be used by the network to forward truth values and associated measurement results from other WTRUs or PRUs, such as truth value label quality indicators exceeding that threshold.
[0333] For example, WTRU 102 can instruct WTRU 102 to request a truth value and associated measurement result where the truth label quality indicator (e.g., soft or hard) is greater than 0.7. In this example, the truth label quality indicator can be a soft value between 0 and 1. In response to the request, WTRU 102 can receive the measurement result, the associated truth value, and the associated truth label quality indicator greater than 0.7 from the network. WTRU 102 can receive truth values greater than or equal to the requested threshold. WTRU 102 can specify the threshold in the request, indicating that WTRU 102 does not want any truth values where the associated truth label quality indicator is below that threshold.
[0334] For example, WTRU 102 may instruct in a request that WTRU 102 request any truth value for which the truth value label quality indicator value is 1. In such a request, WTRU 102 provides an indication of preference for truth value quality.
[0335] For example, WTRU 102 may not include a threshold for the truth label quality indicator in the request. In this case, WTRU 102 can receive truth values and associated measurements from the network without any restrictions on the truth label quality indicator (e.g., the network will forward any truth values and associated measurements). In another example, WTRU 102 may be configured with a threshold by the network, where WTRU 102 can use truth values with an associated truth quality indicator higher than that threshold for training or validation. If WTRU 102 is not configured with this threshold, then WTRU 102 may determine to send a request to the network that includes the threshold for the truth quality indicator.
[0336] In some representative embodiments, the threshold can be in meters. For example, if the truth indicator indicates uncertainty (e.g., an uncertainty of ±0.2 meters), the threshold can be expressed in meters. For instance, WTRU 102 could indicate a threshold of 0.1 meters, indicating that WTRU 102 requests to receive any truth value with a maximum uncertainty less than 0.1 meters. In this example, an uncertainty of ±0.01 meters satisfies the request of WTRU 102, but an uncertainty of ±0.3 meters does not.
[0337] In the example, (for instance, the first) WTRU 102 can send requests for any truth values and associated measurement results from the network. WTRU 102 can receive truth values, associated truth label quality indicators, and associated measurement results from the network.
[0338] The measurement result associated with the truth value indicates that the measurement result was made at the location indicated by the truth value. Another WTRU 102 (e.g., a second WTRU 102) or PRU may be located at the truth value, and the measurement result may be made by that other WTRU 102 or PRU. The measurement result and the truth value or location estimate may be reported by that other WTRU 102 or PRU, and the network may forward the truth value and associated measurement result to WTRU 102 (e.g., the first WTRU 102 that made the request for the truth value and associated measurement result).
[0339] In some representative embodiments, the truth value label quality indicator associated with the truth value can be determined by WTRU 102, PRU, and / or the network. For example, the network can determine the truth value and associated measurement results based on the measurements reported by WTRU 102 and / or PRU. In some embodiments, the network can perform interpolation and / or extrapolation using the truth values and measurement results reported by WTRU 102 and / or PRU, and can derive new pairs of virtual truth values and associated measurement results. For example, a virtual truth value can be identified by the network or WTRU 102 as a virtual WTRU or PRU or associated with a virtual WTRU or PRU. WTRU 102 can receive indications from the network that any measurement results and / or truth values indicated in the auxiliary information are actual measurement results and / or truth values derived by WTRU 102 or PRU, or virtual measurement results and / or truth values derived by the network.
[0340] Figure 21 This is a signaling diagram illustrating an example of truth-finding. Figure 21In the example shown, the second WTRU 102b (e.g., PRU) can receive PRS configuration from the network (e.g., LMF) at 2102 and receive PRS from gNB 180 at 2104. The second WTRU 102b measures the received PRS and reports the measurement results to LMF 600 at 2108. The first WTRU 102a can make a request for the truth value at 2106 and can receive the truth value of the second WTRU 102b from LMF 600 as auxiliary information, which may include the measurement results or inferences made by the second WTRU 102b, the truth value of the second WTRU 102b, and / or the truth value determined by the second WTRU 102b.
[0341] When the truth value of the second WTRU 102b is provided by the network, the first WTRU 102a may receive information from the network indicating any of the following: (i) the truth value of the second WTRU 102b; (ii) a truth value label quality indicator (e.g., a hard or soft indicator) associated with the truth value; (iii) a measurement result associated with the truth value; (iv) one or more PRS configurations (e.g., PRS resource IDs) associated with the measurement result, indicating PRS information used to obtain the measurement result at the second WTRU 102b; (v) the identity of the second WTRU 102b (e.g., PRU ID, UE ID, RNTI associated with the WTRU); (vi) the type of truth value or measurement result (e.g., actual truth value or actual measurement result, virtual truth value, virtual measurement result); (vii) a measurement result timestamp indicating when the measurement result was reported by the second WTRU 102b; (viii) a measurement result timestamp indicating when the measurement result was made by the second WTRU 102b; and / or (ix) Any method used to derive the truth (e.g., PRU, RAT-dependent localization methods such as DL-TDOA).
[0342] In some representative embodiments, WTRU 102 may send requests for truth information. For example, a request may include information indicating any of the following: (i) a threshold for the truth label quality indicator (e.g., indicating a request from the WTRU to obtain truth values for which the associated truth indicator is above the threshold); (ii) the periodicity of truth value provisioning; (iii) whether WTRU 102 requests the same or different truth values for each provisioning of the truth value; and / or (iv) the number of repetitions per truth value (e.g., N). WTRU 102 may receive configuration information from the network that may include any of the following: (i) the number of truth values (e.g., M); (ii) the number of repetitions; and / or (iii) the periodicity of truth value provisioning. For each truth value from the network, WTRU 102 may receive the truth value, the associated truth indicator, and the measurement result associated with the truth value (e.g., RSTD, CIR, timing measurement result, power measurement result, or phase measurement result) for N time points (e.g., WTRU 102 receives auxiliary information containing the truth value and associated information at N different time points). For example, WTRU 102 can receive timestamps and information indicating any PRS configuration associated with the measurement results. WTRU 102 can receive auxiliary information for each of M truth values at a repetition factor of N per truth value under the configured periodicity. WTRU 102 can send a message to the network requesting termination of the preparation of truth values and / or associated information.
[0343] Figure 22 This is a signaling diagram illustrating an example of setting up repeated truth information. Figure 22 In this example, assuming M=2 (number of truth values) and N=3 (number of repetitions), the first WTRU 102a can send a request for truth value information to the network. Measurement results reported by the second WTRU 102b and PRU 1502 are used to derive the truth values. Both the second WTRU 102b and PRU 1502 can periodically report measurement results to the network.
[0344] exist Figure 22 In the example, the requested number of truth values and / or the requested number of repetitions may differ from the configured number of truth values and number of repetitions.
[0345] Example Implementation Figure 23 This is a signaling diagram illustrating an example of validating an AI / ML model using PRU measurement results. Figure 23In this configuration, WTRU 102 can send WTRU capability information (such as the model ID carried by WTRU 102 and the cell ID of the previous cell) to the network (e.g., gNB 180) at point 2302. WTRU 102 can send a request for auxiliary information at point 2304. WTRU 102 may include a request for inference and / or measurement results from PRUs in the cell. WTRU 102 may also include the desired location of the PRU in the request.
[0346] WTRU 102 can receive one or more PRS configurations (e.g., PRS resource IDs used for measurement) from the network at 2306. WTRU 102 receives PRS from the network at 2308. WTRU 102 can perform PRS measurements on the received PRS at 2310. PRU 1502 can perform PRS measurements on the PRS at 2310. PRU 1502 can send a measurement report to the network at 2312. WTRU 102 can receive measurement results and / or inferences performed by the PRU from the network at 2314. Measurement information may include the PRS resource ID used by the PRU to perform the measurement. For example, measurement and / or inference information may include information about the AI / ML model used by the PRU to generate inference (e.g., model ID, associated PRS configuration), measurement results used as input to the AI / ML model, and / or PRU location.
[0347] Based on measurements and / or inference received from the network, WTRU 102 may determine to validate the AI / ML model at WTRU 102 at 2316. Once the model is validated, WTRU 102 may determine to perform any of the following at 2318: (i) send an indication that fine-tuning is complete (e.g., fine-tuning was performed using the current AI / ML model); (ii) send an indication of the association information of the new model (e.g., cell ID); (iii) send an indication to reuse the model used in the previous cell; and / or (iv) send an indication for WTRU 102 to fall back to the fallback positioning method.
[0348] The WTRU 102 can determine its location and report that location to the network.
[0349] In the example embodiment, WTRU 102 may move to a new cell, for example, due to mobility. WTRU 102 may indicate WTRU capability information (e.g., support for WTRU-based AI / ML-based positioning) to the network.
[0350] WTRU 102 can send requests for auxiliary information (such as PRU location, PRU inference, timestamps, associated measurements), indicating the required uncertainty (such as standard deviation) of PRU inference, the desired associated information (such as the desired SSB), and / or the cell ID of the previous cell.
[0351] The WTRU 102 can be configured with a backtracking method (e.g., DL-TDOA).
[0352] The WTRU 102 can perform measurements on a configured PRS set within a configured time window (e.g., periodicity, duration).
[0353] WTRU 102 may (e.g., periodically) receive auxiliary information (e.g., PRU measurement results, corresponding PRU inference, PRU location) from the network. For example, the auxiliary information may include information indicating the duration of the provision of auxiliary information, such as the duration being the same as the duration of a time window.
[0354] At the end of the preparation of auxiliary information (or if WTRU 102 receives a request from the network), WTRU 102 may report information indicating any of the following: (i) fine-tuning completed (e.g., fine-tuning performed using the current AI / ML model); (ii) the association information of the new model (e.g., cell ID); (iii) the model used in the previous cell is reused; and / or (iv) WTRU 102 falls back to the fallback positioning method.
[0355] Determined based on reference AI / ML model In some representative embodiments, WTRU 102 can perform reference-based AI / ML model determination.
[0356] In some representative embodiments, the AI / ML model may be associated with a region defined by a PRS set. WTRU 102 may be configured with N PRS sets (e.g., associated with N regions). Based on measurement conditions associated with the PRS sets (e.g., average RSRP above / below a configured threshold), WTRU 102 may determine whether to use a different model. WTRU 102 may transmit AI / ML outputs, back-off localization outputs (e.g., WTRU location estimates), and / or measurement results for AI / ML and / or back-off localization methods to the network for validation purposes (e.g., for a configured duration).
[0357] The Relationship Between AI / ML Models and Regions In some representative embodiments, the AI / ML model may be associated with a region. A region may be defined as a collection of cells (e.g., indicated by a cell ID), PRS resource IDs, PRS resource set IDs, and / or TRP IDs. WTRU 102 may be configured with references (e.g., cell IDs, PRS resource IDs, SSB IDs, TRP IDs) associated with the region and / or the AI / ML model.
[0358] Figure 24 This is a system diagram illustrating an example of WTRU 102 moving from one PRS set to another. Figure 24 In this configuration, WTRU 102 can be configured, for example, to have two PRS resource sets, namely PRS set #1 and PRS set #2.
[0359] WTRU 102 can (e.g., initially) measure PRS set #1. Due to mobility, WTRU 102 can move to another region where PRS set #2 may have a higher RSRP than that measured for PRS set #1. WTRU 102 can be configured with an AI / ML model associated with PRS set #1, and WTRU 102 can determine to use that AI / ML model if the RSRP associated with PRS set #1 is higher than a pre-configured threshold.
[0360] Due to mobility, for example, if WTRU 102 moves to another region and observes that the RSRP corresponding to PRS set #1 is below a threshold, WTRU 102 can determine the AI / ML model used for validation. WTRU 102 can measure PRS set #2, and if WTRU 102 determines that the RSRP of PRS set #2 is above a pre-configured threshold, WTRU 102 can determine to use the AI / ML model associated with PRS set #2 (e.g., if it is available at WTRU 102).
[0361] exist Figure 24 In this context, WTRU 102 can be configured with AI / ML models associated with regions that cover the areas associated with PRS sets #1 and #2. For example, Figure 23 TRP1, TRP2, TRP3, and TRP4 can be located within this region. In the example, WTRU 102 can determine to use measurements taken from any TRP within this region as input to the AI / ML model (e.g., as long as WTRU 102 is located within this region).
[0362] Examples of references and conditions In some representative embodiments, WTRU 102 may be configured with a reference associated with a group (e.g., cell set, PRS resource, PRS resource set, TRP, or SSB ID). WTRU 102 may perform measurements (e.g., RSRP) on the reference and determine whether the measurement results satisfy one or more conditions. If the one or more conditions are satisfied, WTRU 102 may determine to use the AI / ML model associated with the reference (e.g., the group to which the reference is associated).
[0363] In some representative embodiments, WTRU 102 may be configured with multiple (e.g., a set of) references associated with multiple (e.g., a group of) groups. WTRU 102 may be configured with an AI / ML model associated with each group. WTRU 102 may be configured with references such that WTRU 102 can determine the AI / ML model to use based on measurements of the references. The references may be one of the signals or resources from the set, and WTRU 102 may receive an indication from the network indicating which signal or resource to use as the reference.
[0364] For example, if WTRU 102 is configured with a PRS resource set, the reference can be one of the PRS resources. WTRU 102 can be configured with a resource ID by the network, indicating which PRS resource to use as the reference. Another example of a reference can be an SSB from a cell in a cell set. In this case, WTRU 102 can receive the cell ID and SSB ID from the network, indicating which SSB to use as the reference.
[0365] Figure 25 This is a block diagram illustrating an example configuration of the relationship between the AI / ML model, the SSB, and the reference. Figure 25 In this context, each AI / ML model can be associated with a set (e.g., a group) of SSBs or other DL RSs. For example, AI / ML model #1 is associated with SSB IDs #1 and #2, while AI / ML model #2 is associated with SSB IDs #3 and #4. WTRU 102 can receive information indicating that SSB IDs #1 and #4 are used as references to AI / ML model #1 and AI / ML model #2, respectively.
[0366] In some representative embodiments, the reference may be a set or group of signals and / or resources. For example, WTRU 102 may receive a request from the network to use a set of PRS resources as a reference. In this case, WTRU 102 may determine whether to process and / or aggregate measurements from that set of PRS resources (e.g., the average RSRP from that set of PRS resources) and whether the processed and / or aggregated measurements meet a condition. For example, WTRU 102 may be configured with N PRS resources as references. WTRU 102 may determine to average the RSRP measured from the N PRS resources and compare the average RSRP with a condition to determine whether an AI / ML model associated with the N PRS resources or an AI / ML model associated with the group of PRS resources associated with the N PRS resources should be used.
[0367] Model based on measurement condition switching and region association In some representative embodiments, WTRU 102 may be configured with multiple TRP sets (e.g., groups), and the TRPs may be located within a pre-configured area (e.g., composed of multiple cells). WTRU 102 may determine to use measurements from different TRP sets as input to an AI / ML model, such as when conditions are met (e.g., above or below a pre-configured threshold). If the conditions are not met, WTRU 102 may determine to use measurements from another TRP set that meet the conditions as input to the AI / ML model. WTRU 102 may determine to use measurements from a configured TRP set (e.g., a first set) as input to the AI / ML model if reference RSRP measurements associated with that TRP set (e.g., SSBs transmitted from one of the TRPs) meet the conditions. If measurements from that TRP set do not meet the conditions, WTRU 102 may determine to measure a different set (e.g., a second TRP set). If WTRU 102 measures the PRS from the second TRP set and they meet the conditions, then WTRU 102 can determine to use the measurement results from the second TRP set as input to the AI / ML model.
[0368] Figure 26 This is a block diagram illustrating another example configuration of the relationship between the AI / ML model, the SSB, and the reference. Figure 26In this configuration, AI / ML model #1 is associated with SSB IDs #1, #2, #3, and #4. AI / ML model #2 is associated with SSB IDs #5, #6, #7, and #8. WTRU 102 can receive configuration messages from the network indicating that SSB IDs #1 and #2 should be used as references for AI / ML model #1. WTRU 102 can also receive configuration messages indicating that SSB IDs #6, #7, and #8 should be used as references for AI / ML model #2. WTRU 102 can be configured with the same or different number of signals or reference signals for each AI / ML model to be used as a reference.
[0369] For example, WTRU 102 can be configured or requested to monitor reference measurements (e.g., RSRP). WTRU 102 can be configured with a reference signal (e.g., PRS) or signal (e.g., SSB) associated with a region. WTRU 102 can also be configured with thresholds by the network. WTRU 102 can be configured with multiple AI / ML models, each of which can be associated with a region. If, for example, the RSRP of the SSB associated with that region is below a threshold, WTRU 102 can determine whether different models available at WTRU 102, possibly associated with different regions, should be used.
[0370] Example Implementation In some representative embodiments, WTRU 102 may receive a request from the network to perform AI / ML-based positioning. WTRU 102 may report its location to the network. For example, the location may be determined based on a RAT-independent positioning method (e.g., GNSS).
[0371] In some representative embodiments, WTRU 102 may receive from the network more than one set of PRS resource IDs, a fallback positioning method (e.g., GNSS, RAT-dependent positioning method), and a verification duration (e.g., 10 seconds) and / or an RSRP threshold.
[0372] WTRU 102 can identify AI / ML models that use a set of PRS resources (e.g., first measurement results, first PRS set) that produces at least K PRS resources associated with the highest RSRP.
[0373] WTRU 102 can determine AI / ML input values (e.g., channel impulse response) for an AI / ML model based on measurements from an associated PRS set. For example, WTRU 102 can determine the channel impulse response based on a PRS set associated with the AI / ML model. WTRU 102 can determine the PRS resources to report the CIR to the network and the measured quantities to determine the CIR. WTRU 102 can receive indications from the network of the PRS resources to be used to determine the CIR.
[0374] If the RSRP (e.g., average) of the associated PRS set (e.g., the second measurement result) drops below the RSRP threshold (e.g., due to WTRU mobility), then WTRU 102 can determine to use the AI / ML model associated with one of the pre-configured PRS sets (e.g., the second PRS set) that has the highest (e.g., average) RSRP.
[0375] WTRU 102 can indicate to the network which resource set it is using. WTRU 102 can send location estimates and backtracking localization methods (e.g., GNSS, RAT-dependent localization methods) generated by AI / ML models, along with corresponding measurement results (e.g., CIR, raw measurement results such as RSRP, RSTD, and measurement results used for backtracking RAT-dependent localization methods and AI / ML localization methods) for the verification duration (or until WTRU 102 receives a termination message from the network).
[0376] If WTRU 102 does not have an AI / ML model associated with one of the pre-configured PRS sets with the highest (e.g., average) RSRP, then WTRU 102 can determine to use the backtracking approach.
[0377] Figure 27 This is a signaling diagram illustrating an example determined by an AI / ML model based on the PRS region. Figure 27 In this configuration, WTRU 102 can indicate WTRU capability information to the network at point 2702. WTRU 102 can receive N PRS configuration sets from the network at point 2704. WTRU 102 can receive PRS at point 2706 and perform measurements on the received PRS at point 2708. WTRU 102 can perform validation on the AI / ML model at point 2710. After validation, WTRU 102 can indicate to the network at point 2712 which AI / ML model WTRU 102 will use.
[0378] In some representative embodiments, WTRU 102 can indicate WTRU capability information (e.g., support for WTRU-based AI / ML-based positioning) to the network. WTRU 102 can receive requests for AI / ML-based positioning from the network. WTRU 102 can report its location to the network (e.g., based on RAT-dependent positioning, GNSS). WTRU 102 can be configured with N PRS sets, a fallback positioning method (e.g., GNSS, RAT-dependent positioning method), a verification duration (e.g., 10 seconds), and an RSRP threshold. WTRU 102 can determine the AI / ML model (e.g., a first measurement from a first PRS set) associated with a PRS set that produces at least K PRSs associated with the highest RSRP.
[0379] WTRU 102 can determine AI / ML input values (e.g., CIR) for AI / ML models based on measurements from associated PRS sets.
[0380] If the average RSRP of the associated PRS set (e.g., using a second measurement) falls below the RSRP threshold (e.g., due to WTRU mobility), WTRU 102 can determine to use the AI / ML model associated with one of the pre-configured PRS sets (e.g., the second PRS set) with the highest average RSRP.
[0381] WTRU 102 can indicate to the network which set it is using. WTRU 102 can send location estimates and backtracking localization methods (e.g., GNSS, RAT-dependent localization methods) generated by AI / ML models, along with corresponding measurements (e.g., CIR, raw measurements, such as RSRP, RSTD, and measurements used for backtracking RAT-dependent localization methods and / or AI / ML localization methods) for the verification duration (or until WTRU 102 receives a termination message from the network).
[0382] If WTRU 102 does not have an AI / ML model associated with one of the pre-configured PRS sets with the highest average RSRP, then WTRU 102 can fall back to the fallback localization method.
[0383] Determined based on AI / ML models with advance timing In some representative embodiments, the AI / ML model may be associated with a timing advance (TA) value (e.g., for a TRP). WTRU 102 can determine the TA and the associated AI / ML model. If the TA value changes, WTRU 102 can determine the AI / ML model to use that falls within the range of the changed (e.g., the new) TA value. When ranges from different TRPs overlap, WTRU 102 can determine the AI / ML model based on prioritization rules.
[0384] Relationship between AI / ML models and TA scope In some representative embodiments, WTRU 102 may be configured with an AI / ML model associated with a TA value. The AI / ML model may be associated with a TRP or a specific TRP indicated by a TRP ID.
[0385] Figure 28 This is a system diagram illustrating an example of TRP and Timing Advance (TA) associated with an AI / ML model. Figure 28 In this configuration, WTRU 102 can be configured with multiple AI / ML models, such as AI / ML model #1 and AI / ML model #2. WTRU 102 can be configured with a range of TA values, based on which WTRU 102 can determine which AI / ML model to use. AI / ML model #1 is associated with TA values between 0 and TA1 (e.g., in seconds). AI / ML model #2 is associated with TA values between TA1 and TA2. For TA values between TA2 and TA3, WTRU 102 can determine to use a configured RAT-dependent localization method. As another example, for any value exceeding TA2, WTRU 102 can determine to use a configured RAT-dependent localization method.
[0386] In some representative embodiments, WTRU 102 can be configured to allow WTRU 102 to use AI / ML models based on distance from a reference point. WTRU 102 may have an AI / ML model optimized for regions close to the reference point and another AI / ML model far from the reference point. For example, the AI / ML model trained at any location far from the reference point may be trained using measurements with multipaths or many NLOS paths (e.g., CIR). On the other hand, the AI / ML model trained closer to the reference point may be trained based primarily on measurements based on LOS paths.
[0387] exist Figure 28 There are three thresholds, TA1, TA2, and TA3. For example, WTRU 102 can receive the thresholds from the network. For example, WTRU 102 can determine the thresholds based on WTRU capabilities. Figure 28 As can be seen, the AI / ML model is associated with the distance to TRP#1 and #2 (e.g., gNB or reference point in the cell), and TA is defined relative to that distance.
[0388] TA range configuration In some representative embodiments, WTRU 102 may send auxiliary information to the network indicating the capabilities of the AI / ML model at WTRU 102. For example, WTRU 102 may indicate to the network the range of TA values associated with the AI / ML model. In another example, WTRU 102 may indicate the location where the AI / ML model was trained.
[0389] For example, WTRU 102 can receive a request to send its location to the network. Based on the request, WTRU 102 can determine its location, which is determined based on a RAT-dependent positioning method (e.g., DL-TDOA) or a RAT-independent positioning method (e.g., GNSS).
[0390] For example, based on auxiliary information or WTRU capabilities, WTRU 102 can receive a range of TA values from the network, and WTRU 102 determines which AI / ML model(s)(s) to use based on that range. If WTRU 102 does not receive a range of TA values from the network, WTRU 102 can determine which AI / ML model to use based on the TA range associated with the AI / ML model at WTRU 102.
[0391] For example, if WTRU 102 is configured with more than one TA range, where each range is associated with a TRP, WTRU 102 can also receive an index or ID associated with each TA range.
[0392] Resolving ambiguities in configuration In some representative embodiments, the range may be associated with either TRP or gNB. WTRU 102 may need to determine which model to use based on the TA value transmitted from the network. Figure 28 As shown, each of TRP1 and TRP2 is associated with a corresponding range. For example, WTRU 102 can receive two TA values associated with each of TRP1 and TRP2. Based on each TA value, WTRU 102 can determine which TRP and associated AI / ML model to use to determine its location.
[0393] For example, WTRU 102 could be located in a region where the TA ranges associated with the TRP overlap (e.g., TA2 and TA3), such as Figure 28 As shown. In this scenario, WTRU 102 may need to determine which TA range to associate the AI / ML model at WTRU 102 with. In one example, WTRU 102 may determine the TA range based on configured priority rules.
[0394] For example, if WTRU 102 needs to determine between conventional localization methods and AI / ML-based localization due to overlapping TA ranges (e.g., between two TRPs), WTRU 102 can prioritize AI / ML-based localization, and WTRU 102 can indicate the TRP ID or TA range index associated with the AI / ML model.
[0395] In some representative embodiments, the AI / ML model can be associated with any TRP.
[0396] For example, if WTRU 102 needs to determine between two AI / ML models associated with different TA values, WTRU 102 can determine to use the AI / ML model associated with the smaller TA value. For example, in Figure 28 In this example, WTRU 102 can receive TA commands from two TRPs, where the TA value associated with TRP1 falls between TA1 and TA2 (e.g., within a range), and the TA value associated with TRP2 falls between TA2 and TA3 (e.g., within a range). In this example, since the TA value associated with TRP1 is less than the TA value associated with TRP2, WTRU 102 can determine that an AI / ML model (e.g., AI / ML model #2) associated with the range between TA1 and TA2 should be used. WTRU 102 can report to the network that the AI / ML model associated with the range between TA1 and TA2 has been used. WTRU 102 can also report that measurements of the PRS transmitted from TRP1 are used as input to AI / ML model #2.
[0397] In some representative embodiments, the AI / ML model may be associated with a specific TRP.
[0398] For example, if WTRU 102 is configured with AI / ML models associated with a specific TRP, WTRU 102 can determine which AI / ML model to use based on the TA value associated with the TRP. For example, WTRU 102 can be configured with a first set of AI / ML models associated with TRP1, where each AI / ML model can be associated with a different TA value or range of TA values.
[0399] Example Implementation In some representative embodiments, WTRU 102 can indicate WTRU capability information to the network, which may indicate support for and / or availability of AI / ML models associated with the TA. WTRU 102 can receive requests from the network to perform AI / ML-based localization. WTRU 102 can be configured with thresholds for the TA value and the TRP location. The threshold may be associated with the TRP. WTRU 102 can receive configuration of a fallback localization method. If WTRU 102 receives a TA command from the network (e.g., via a MAC message), WTRU 102 can determine which AI / ML model to use based on the TA value indicated in the TA command.
[0400] If a change exists in the TA and the TA is within one of the pre-configured ranges (e.g., defined by a threshold), then WTRU 102 can determine to use the AI / ML model associated with that range. If WTRU 102 is not configured with an AI / ML model corresponding to that TA (e.g., the TA is below or above a minimum or maximum range of the TA value), then WTRU 102 can determine to fall back to the fallback positioning method.
[0401] For example, WTRU 102 can (e.g., if necessary) determine between conventional localization methods and AI / ML-based localization due to overlapping ranges (e.g., between two TRPs). WTRU 102 can prioritize AI / ML-based localization and indicate the TRP ID associated with the AI / ML model.
[0402] For example, WTRU 102 can (e.g., need to) determine between two AI / ML models associated with different TA values, and WTRU 102 can determine which AI / ML model is used, associated with the smaller TA value.
[0403] In some representative embodiments, WTRU 102 can indicate WTRU capability information (e.g., support for TRP-based AI / ML models) to the network. WTRU 102 can be configured by the network with TRP locations, TA value ranges, and AI / ML models associated with each range. The AI / ML models can be configured publicly for the configured TRPs. WTRU 102 can be configured with a fallback positioning method (e.g., DL-TDOA). WTRU 102 can receive requests for AI / ML-based positioning from the network. WTRU 102 can receive TA commands from the network. The TA commands can contain multiple TA values specific to each TRP. Based on the TA values and the pre-configured ranges into which the TA values fall, WTRU 102 can determine the AI / ML model. If there is a change in the TA and the TA is within one of the pre-configured ranges, WTRU 102 determines to use the AI / ML model associated with one of the ranges where the TA is located.
[0404] If WTRU 102 is not configured with an AI / ML model corresponding to the TA (e.g., the TA is below or above the minimum or maximum range of the TA value), then WTRU 102 can determine to fall back to the fallback positioning method.
[0405] For example, if WTRU 102 needs to determine between conventional localization methods and AI / ML-based localization due to overlapping ranges (e.g., between two TRPs), WTRU 102 can prioritize AI / ML-based localization and indicate the TRP ID associated with the AI / ML model.
[0406] For example, if WTRU 102 needs to determine between two AI / ML models associated with different TA values, WTRU 102 can determine to use the AI / ML model associated with the smaller TA value.
[0407] AI / ML model determination based on PFL In some representative embodiments, the AI / ML model may be associated with an initial PFL set. WTRU 102 can determine whether to use the same AI / ML model (e.g., the same as the previous cell) or a backoff positioning method in the new cell based on whether the PFL configuration in the new cell includes the initial PFL set or a subset thereof.
[0408] The Relationship Between AI / ML Models and Frequency Layers In some representative embodiments, WTRU 102 may be configured with an AI / ML model (e.g., a first AI / ML model) associated with a PRS configuration set (e.g., a first PRS configuration set). WTRU 102 can use the AI / ML model to determine the WTRU location. For example, if the PRS configuration is updated and the updated PRS configuration contains that PRS configuration set, WTRU 102 may determine to use the first AI / ML model to determine the WTRU location. For example, if the updated PRS configuration is a subset of the first PRS configuration set, WTRU 102 may determine to use the first AI / ML model to determine the WTRU location.
[0409] For example, WTRU 102 can determine the association between an AI / ML model and a configured frequency layer. When WTRU 102 is configured with another set of frequency layers, WTRU 102 can determine whether the AI / ML model in use can be used.
[0410] For example, if the updated PRS configuration contains at least one of the localization frequency layers associated with the AI / ML model, then WTRU 102 can determine that the same AI / ML model is used.
[0411] Figure 29 This is a system diagram illustrating an example of determining an AI / ML model due to mobility.
[0412] exist Figure 29 In this configuration, WTRU 102 is configured with three frequency layers in the region associated with TRP 1-1 (e.g., cell #1). The three frequency layers are PFL#1, PFL#2, and PFL#3. WTRU 102 has AI / ML models associated with PFL#1, PFL#2, and PFL#3 (e.g., trained using measurements taken using PFL#1, PFL#2, and PFL#3). WTRU 102 moves to another region associated with TRP 2-1 (e.g., cell #2). WTRU 102 can be configured with a new set of frequency layers (e.g., PFL#2 only). Since the AI / ML model is associated with PFL#2, WTRU 102 can determine (e.g., continue) to use the same model. WTRU 102 can then move to the region associated with TRP 3-1 (e.g., cell #3). In this region, WTRU 102 can be configured with (e.g., PFL#3 only). Since the AI / ML model at WTRU 102 is associated with PFL#3, WTRU 102 can determine (e.g., continue) to use the same model.
[0413] Figure 30 This is a system diagram illustrating another example of how AI / ML models are determined due to mobility.
[0414] exist Figure 30 In this configuration, WTRU 102 is configured with PFL#2 in the region associated with TRP 1-1 (e.g., cell #1). WTRU 102 can train an AI / ML model using measurements obtained using PFL#2. When WTRU 102 moves to a (e.g., a new) region associated with TRP 2-1 (e.g., cell #2), WTRU 102 can be configured with three frequency layers (e.g., PFL#1, PFL#2, and PFL#3). Since PFL#2, previously associated with the AI / ML model at WTRU 102, is one of the configured PFLs, WTRU 102 can determine (e.g., continue) to use the same AI / ML model. WTRU 102 can then move to a region associated with TRP 3-1, where WTRU 102 is configured with (e.g., only) PFL#3. Since PFL#3 is not associated with the AI / ML model at WTRU 102, WTRU 102 can determine (e.g., switch) to use a fallback localization method (e.g., DL-TDOA) that can be configured by the network.
[0415] Figure 31This is a system diagram illustrating another example of how AI / ML models are determined due to mobility. Figure 31 In this configuration, WTRU 102 is configured with three frequency layers in the region associated with TRP 1-1 (e.g., cell #1). These three frequency layers are PFL#1, PFL#2, and PFL#3. WTRU 102 has AI / ML models associated with PFL#1, PFL#2, and PFL#3 (e.g., trained using measurements taken using PFL#1, PFL#2, and PFL#3). WTRU 102 can be moved to another (e.g., a new) region associated with TRP 2-1 (e.g., cell #2). WTRU 102 can be configured with a new set of frequency layers, namely PFL#1 and PFL#2. Since the AI / ML models are associated with both PFL#1 and PFL#2, WTRU 102 can determine (e.g., continue) to use the same models. WTRU 102 can be moved to a (e.g., a new) region associated with TRP 3-1 (e.g., cell #3). In this region, WTRU 102 can be configured with PFL#3. Since the AI / ML model at WTRU 102 is associated with PFL#3, WTRU102 can determine (e.g., continue) to use the same model.
[0416] For example, if the input to the AI / ML model is a timing measurement result, the WTRU 102 can determine to use the same AI / ML model.
[0417] For example, if WTRU 102 determines that the same AI / ML model is being used, WTRU 102 can send information to the network indicating that the same AI / ML model configured at the initiation reference point (e.g., the cell where the AI / ML model is associated) is being used.
[0418] For example, WTRU 102 can determine to associate the AI / ML model with the PRS configuration at cell #1, where cell #1 becomes the reference point, based on an instruction from the network (e.g., an instruction to associate the AI / ML model with the indicated PRS configuration). If WTRU 102 moves to another cell and WTRU 102 determines to use the same AI / ML model based on a condition (e.g., the PFLs in the new PRS configuration are a subset of the PFLs associated with the AI / ML model), then WTRU 102 can send information to the network indicating that the same AI / ML model is being used.
[0419] For example, the reference could be a first (e.g., initial) PRS configuration associated with an AI / ML model. WTRU 102 can be configured with a PRS configuration. WTRU 102 can determine to associate the AI / ML model with that PRS configuration (e.g., the initial PRS configuration). WTRU 102 can receive updates to the first PRS configuration or a second (e.g., new) PRS configuration from the network. If one or more PRS parameters that are part of the first PRS configuration associated with the AI / ML model are included in or part of the updated or second PRS configuration, then WTRU 102 can determine to use the same AI / ML model.
[0420] For example, if a fallback positioning method (such as a positioning method that depends on RAT) is used, WTRU 102 can indicate that the fallback positioning method is used.
[0421] In some representative embodiments, WTRU 102 can indicate WTRU capability information (e.g., support for TRP-based AI / ML models) to the network. WTRU 102 can be configured with a fallback positioning method (e.g., DL-TDOA). WTRU 102 can be configured with AI / ML positioning by the network, where the AI / ML model is associated with a set of PFLs (e.g., an initial PFL set). WTRU 102 can receive the current cell's PRS configuration (e.g., supported PFLs) from the network. WTRU 102 can determine the positioning method based on one of the following: If WTRU 102 is configured with aggregated PFLs and the initial PFL set is a subset of the aggregated PFLs, then WTRU 102 can determine to use the same AI / ML model. If WTRU 102 is configured with aggregated PFLs and the aggregated PFLs include the initial PFL set, then WTRU 102 can determine to use the same AI / ML model. If WTRU 102 is configured with PFLs not included in the initial PFL set, then WTRU 102 can determine to use the fallback positioning method.
[0422] Figure 32 This is a flowchart illustrating an example of AI / ML model validation. In some representative embodiments, Figure 32 The process can be implemented by WTRU 102 (e.g., as a method). Figure 32At point 3202, WTRU 102 can send a request for auxiliary information to the network, which includes any one of the following: PRU location, PRU inference of the PRU location, PRS set associated with the PRU inference, first measurement result set associated with the PRU inference, and / or cell identifier associated with the PRU inference. At point 3204, WTRU 102 can receive auxiliary information. At point 3206, WTRU 102 can obtain inference of the WTRU location from the AI / ML model. At point 3208, WTRU 102 can verify the obtained inference based on the auxiliary information. At point 3210, WTRU 102 can send the verification result of the obtained inference to the network.
[0423] For example, WTRU 102 can (e.g., prior to 3202) send WTRU capability information to the network indicating support for AI / ML-based positioning.
[0424] For example, obtaining inference from an AI / ML model at point 3206 may include inputting measurement information associated with a first set of measurement results into the AI / ML model.
[0425] For example, WTRU 102 can perform a second measurement result set on the PRS set. At 3206, obtaining inference from the AI / ML model can include inputting measurement information associated with the second measurement result set into the AI / ML model.
[0426] For example, verifying the obtained inference at 3208 may include determining (i) the difference between the obtained inference and (ii) the PRU location and / or PRU inference. At 3210, the result may include information indicating the tuning of the WTRU to the AI / ML model based on the determined difference being less than a threshold.
[0427] For example, verifying the obtained inference at 3208 may include determining (i) the difference between the obtained inference and (ii) the PRU location and / or PRU inference. At 3210, the result may include information indicating an identifier of the AI / ML model based on the determined difference being less than a threshold.
[0428] For example, validating the obtained inference at 3208 may include determining (i) the difference between the obtained inference and (ii) the PRU location and / or PRU inference. At 3210, the result may include information indicating an identifier of another AI / ML model based on the determined difference being greater than or equal to a threshold.
[0429] For example, verifying the obtained inference at 3208 may include determining (i) the difference between the obtained inference and (ii) the PRU location and / or PRU inference. At 3210, the result may include information indicating an identifier of the backtracking method based on the determined difference being greater than or equal to a threshold.
[0430] Figure 33 This is a flowchart illustrating an example of switching AI / ML models. In some representative embodiments, Figure 33 The process can be implemented by WTRU 102 (e.g., as a method). At 3302, WTRU 102 can receive configuration information indicating multiple PRS sets. At 3304, WTRU 102 can obtain first measurement information associated with the multiple PRS sets. At 3306, WTRU 102 can select a first set among the multiple PRS sets based on the first measurement information associated with the multiple PRS sets. At 3308, WTRU 102 can obtain inference of WTRU location from a first AI / ML model associated with the first set. At 3310, WTRU 102 can obtain second measurement information associated with the multiple PRS sets. At 3312, WTRU 102 can send information indicating a second AI / ML model to the network based on the second measurement information associated with the first set being below a threshold. The second AI / ML model can be associated with a second set among the multiple PRS sets.
[0431] For example, the inference of the WTRU location obtained from the first AI / ML model at 3308 can use the first measurement information associated with the first set as input to the first AI / ML model.
[0432] For example, selecting the first set in the plurality of PRS sets at 3306 can be based on the highest first measurement information associated with the first set among the plurality of PRS sets.
[0433] For example, the second set in the plurality of PRS sets can be selected by WTRU 102 based on the second measurement information associated with the second set being the highest among the plurality of PRS sets (e.g., other than the first set).
[0434] For example, the first measurement information may include the reference signal received power (RSRP) value of each of the plurality of PRS sets during a first time interval.
[0435] For example, the second measurement information may include the reference signal received power (RSRP) value of each of the plurality of PRS sets during the second time interval.
[0436] For example, WTRU 102 can receive information indicating the association between the second AI / ML model and the second set.
[0437] Figure 34 This is a flowchart illustrating another example of switching AI / ML models. In some representative embodiments, Figure 34 The process can be implemented by WTRU 102 (e.g., as a method). At 3402, WTRU 102 can receive configuration information from the network indicating multiple TA ranges associated with multiple AI / ML models. At 3404, WTRU 102 can receive a TA command including information indicating a first TA value. At 3406, WTRU 102 can select a first AI / ML model from the multiple AI / ML models that is associated with a first TA range including the first TA value. At 3408, WTRU 102 can determine a second TA value (e.g., from another TA command updating the first TA value). At 3410, WTRU 102 can select a second AI / ML model from the multiple AI / ML models that is associated with a second TA range including the second TA value.
[0438] For example, WTRU 102 can obtain inference of WTRU location from the first AI / ML model.
[0439] For example, WTRU 102 can obtain inference of WTRU location from a second AI / ML model.
[0440] For example, the TA command may include information indicating a first TA value associated with a first TRP and a third TA value associated with a second TRP. At 3406, selecting a first AI / ML model may be based on the first TA value being less than the third TA value.
[0441] For example, at 3408, WTRU 102 can determine the fourth TA value. The second TA value can be associated with the third TRP, and the fourth TA value can be associated with the fourth TRP. WTRU 102 can select the second AI / ML model at 3410 based on the second TA value being less than the fourth TA value.
[0442] For example, WTRU 102 can send information to the network indicating the selected first AI / ML model.
[0443] For example, WTRU 102 can send information to the network indicating the selected second AI / ML model.
[0444] Figure 35 This is a flowchart illustrating another example of switching AI / ML models. In some representative embodiments, Figure 35The process can be implemented by WTRU 102 (e.g., as a method). At 3502, WTRU 102 can receive first configuration information from the network indicating a first set of PFLs (e.g., PRS configuration) associated with the AI / ML model. At 3504, after moving from the first cell to the second cell, WTRU 102 can receive second configuration information indicating a second set of PFLs. At 3506, WTRU 102 can send information indicating the AI / ML model to the network based on the fact that the first and second sets together have at least one PFL.
[0445] Figure 36 This is a flowchart illustrating another example of switching AI / ML models. In some representative embodiments, Figure 36 The process can be implemented by WTRU 102 (e.g., as a method). At 3602, WTRU 102 can receive first configuration information from the network indicating a first set of PFLs (e.g., PRS configuration) associated with the AI / ML model. At 3604, after moving from the first cell to the second cell, WTRU 102 can receive second configuration information indicating a second set of PFLs. At 3606, WTRU 102 can send information indicating a backoff positioning method to the network based on the fact that each PFL in the first set is different from each PFL in the second set.
[0446] For example, (in Figure 35 (and / or 36) WTRU 102 can infer the WTRU location from the AI / ML model when it is in the first cell.
[0447] For example, (in Figure 35 (In the middle) WTRU 102 can obtain inference of WTRU location from AI / ML model when it is in the second cell.
[0448] For example, (in Figure 35 (And / or 36) AI / ML models can be trained using measurement information associated with the multiple PFLs.
[0449] For example, (in Figure 35 (and / or 36) The first configuration information may be received from the first transmit / receive point (TRP).
[0450] For example, (in Figure 35 (and / or 36) The second configuration information may be received from the second transmit / receive point (TRP).
[0451] Figure 37 This is a flowchart illustrating an example process according to one or more embodiments of this disclosure. Figure 37As shown, at 3702, WTRU 102 can send capability information indicating that the WTRU supports AI / ML positioning to the base station associated with the first cell and based on the WTRU 102 moving from the second cell to the first cell. At 3704, WTRU 102 can send a request to the base station for auxiliary information for PRU inference. At 3706, WTRU 102 can receive configuration information indicating the backoff positioning method and resources associated with the PRS set. At 3708, WTRU 102 can measure the PRS set. At 3710, WTRU 102 can receive auxiliary information indicating the location of the PRU and / or the measurement information set associated with the location PRU inference. At 3712, WTRU102 may send a report to the base station based on a request from the base station, the report indicating any of the following: (i) a first AI / ML model used at the WTRU for positioning in the second cell to be used in the first cell, (ii) fine-tuning of the first AI / ML model used at the WTRU for positioning in the second cell, (iii) a second AI / ML model to be used at the WTRU for positioning in the second cell, (iv) a back-off positioning method to be used at the WTRU for positioning in the second cell, and / or (v) inference generated based on the indicated set of measurement information and other measurement information associated with the measured PRS set.
[0452] For example, a request for auxiliary information may include information indicating any one of the following: PRU location, PRU inference of PRU location, time interval, and / or PRU measurement information type.
[0453] For example, a request for auxiliary information may include information indicating the identifier of the first cell, the identifier of the second cell, and / or any item in the SSB set associated with the first cell.
[0454] For example, the backoff positioning method can be downlink time difference of arrival (DL-TDoA) positioning.
[0455] For example, the received auxiliary information can be associated with the first time interval.
[0456] For example, the PRS set can be measured during the first time interval.
[0457] For example, WTRU 102 can obtain inference from a first AI / ML model based on the indicated set of measurement information and other measurement information associated with the measured PRS set.
[0458] For example, the report information may include information indicating that the first AI / ML model used for positioning in the second cell should be used in the first cell based on the fact that the difference between the inference obtained from the first AI / ML model and the PRU inference is less than a first threshold.
[0459] For example, the report information may include information indicating fine-tuning of the first AI / ML model used for positioning in the second cell at the WTRU based on the difference between the inference obtained from the first AI / ML model and the PRU inference being greater than a first threshold and less than a second threshold.
[0460] For example, the report information may include information indicating that a second AI / ML model or backoff positioning method should be used at the WTRU for positioning in the second cell, based on the fact that the difference between the inference obtained from the first AI / ML model and the PRU inference is greater than a second threshold.
[0461] Figure 38 This is a flowchart illustrating an example process according to one or more embodiments of this disclosure. Figure 38 As shown, WTRU 102 can determine its location at 3802. At 3804, WTRU 102 can, based on the determined location, (i) select an AI / ML model from the AI / ML model set or (ii) select a back-off positioning method. Each AI / ML model in the AI / ML model set can be associated with a corresponding area. At 3806, WTRU 102 can send information to the base station indicating the selected AI / ML model or the selected back-off positioning method (e.g., one of them).
[0462] For example, WTRU 102 can determine the location of WTRU 102 when WTRU moves from the second cell to the first cell.
[0463] For example, WTRU 102 can determine its position when it receives information from a set of indication positioning reference signals (PRS).
[0464] For example, WTRU 102 can determine its location when it receives information indicating the TA value.
[0465] For example, WTRU 102 can determine its location upon receiving information indicating PFL.
[0466] For example, WTRU 102 can select AI / ML models from the AI / ML model set based on the determined (e.g., WTRU 102's) location within the region associated with the AI / ML model.
[0467] For example, WTRU 102 can send information to the base station indicating that WTRU 102 supports AI / ML positioning capabilities.
[0468] For example, WTRU 102 can receive configuration information indicating any one of multiple PRS, multiple cells, and / or multiple TRPs. Each region can be associated with at least a portion of the configuration information. For example, a first region can be associated with one or more of the PRS, cells, and / or TRPs. For example, a second region can be associated with another one or more of the PRS, cells, and / or TRPs. For example, a third region can be associated with yet another one or more of the PRS, cells, and / or TRPs.
[0469] For example, WTRU 102 can determine its location based on configuration information (e.g., WTRU 102's location).
[0470] Figure 39 This is a flowchart illustrating an example process according to one or more embodiments of this disclosure. Figure 39 As shown, WTRU 102 can receive information indicating a first PRS configuration at 3902. For example, the first PRS configuration can be associated with an AI / ML model. At 3904, WTRU 102 can receive information indicating a second PRS configuration associated with the second cell when WTRU 102 moves from the first cell to the second cell. At 3906, WTRU 102 can select (i) an AI / ML model or (ii) a fallback positioning method based on whether the second PRS configuration is included in the first PRS configuration. At 3908, WTRU 102 can send information associated with the selected AI / ML model or the selected fallback positioning method (e.g., one of them).
[0471] For example, the information indicating the first PRS configuration may be received from a base station associated with the first cell.
[0472] For example, the information indicating the second PRS configuration may be received from a base station associated with the second cell.
[0473] For example, information associated with the selected AI / ML model or the selected backoff positioning method can be sent to the base station associated with the second cell.
[0474] For example, the first PRS configuration may include information associated with the first PFL set.
[0475] For example, the second PRS configuration may include information associated with the second PFL set. The first PFL set may include the second PFL set.
[0476] For example, WTRU 102 can select an AI / ML model based on the fact that the second PFL set is a subset of the first PFL set.
[0477] For example, WTRU 102 can select a backtracking method based on the fact that the second PFL set is not a subset of the first PFL set.
[0478] For example, WTRU 102 can receive information indicating the backoff positioning method from the base station associated with the second cell.
[0479] For example, WTRU 102 can receive information indicating the backoff positioning method from the base station associated with the first cell.
[0480] Figure 40 This is a flowchart illustrating an example process according to one or more embodiments of this disclosure. Figure 40 As shown, WTRU 102 can receive information indicating a set of AI / ML models at 4002. For example, each AI / ML model can be associated with a corresponding TA value range. For example, the TA value range can be associated with a specific base station. At 4004, WTRU 102 can receive information indicating the TA value associated with a base station. At 4006, WTRU can select an AI / ML model from the set of AI / ML models based on the indicated TA value falling within the TA value range associated with the AI / ML model. For example, the indicated TA value can fall within the TA value range associated with a first AI / ML model and can fall outside the TA value range associated with a second AI / ML model. At 4008, WTRU 102 can send information associated with the selected AI / ML model. For example, WTRU 102 can send the identifier of the selected AI / ML model and / or information associated with the inference of the selected AI / ML model.
[0481] For example, information indicating an AI / ML model set can be associated with a TRP.
[0482] For example, an AI / ML model set may include a first AI / ML model associated with a first TA value range and a second AI / ML model associated with a second TA value range.
[0483] For example, WTRU 102 can select one of the first AI / ML models or the second AI / ML model based on whether the indicated TA value falls within a first TA value range or a second TA value range.
[0484] For example, WTRU 102 can select a backtracking method based on the fact that the indicated TA value falls outside the range of TA values associated with each AI / ML model in the AI / ML model set.
[0485] For example, WTRU 102 can send information indicating that WTRU supports AI / ML positioning capabilities.
[0486] For example, WTRU 102 can send information indicating the WTRU's ability to support AI / ML model sets.
[0487] For example, each AI / ML model can be associated with a corresponding TA value range and a corresponding base station and / or TRP.
[0488] For example, WTRU 102 can select an AI / ML model from the AI / ML model set based on the fact that the indicated TA value for the base station (or TRP) falls within the range of TA values associated with both the AI / ML model and the base station.
[0489] For example, each AI / ML model can be associated with a corresponding TA value range and a base station (or TRP). In another example, each AI / ML model can be associated with a corresponding TA value range and a specific base station (or TRP).
[0490] Figure 41 This is a flowchart illustrating an example process according to one or more embodiments of this disclosure. Figure 41 As shown, WTRU 102 can receive information at 4102 indicating a first set of measurement results associated with the PRU. At 4104, WTRU 102 can use the first set of measurement results to obtain location inference from a first AI / ML model. At 4106, WTRU 102 can send information associated with the obtained inference to the network.
[0491] For example, WTRU 102 can send WTRU capability information to the network indicating support for AI / ML-based positioning.
[0492] For example, WTRU 102 can use measurement information associated with a first set of measurement results as input to an AI / ML model to obtain inference from the AI / ML model.
[0493] For example, WTRU 102 can perform a second measurement result set on the PRS set. WTRU 102 can obtain location inference from the AI / ML model based on measurement information associated with the first measurement result set and measurement information associated with the second measurement result set as input to the AI / ML model.
[0494] For example, the first set of measurement results and / or the second set of measurement results can be associated with any of the following: timing information, phase information, power information, channel impulse response, and / or delay distribution.
[0495] For example, WTRU 102 can receive information indicating the PRU location. WTRU 102 can determine the difference between the obtained inference and the PRU location. WTRU 102 can include this difference in the information reported to the network.
[0496] For example, based on the determined difference being less than a threshold, the information associated with the obtained inference may include any one of the obtained inference, PRU location, and / or the identifier of the first AI / ML.
[0497] For example, based on the determined difference being less than a threshold, the information associated with the obtained inference may include any one of the obtained inference, PRU location, and / or the identifier of the second AI / ML.
[0498] For example, a PRU can be associated with the cell in which WTRU 102 is located.
[0499] For example, the information associated with the obtained inference may include any one of the cell identifier, the PRU identifier, and / or the identifier of the first AI / ML model.
[0500] In some representative embodiments, WTRU 102 may send a request to the network for auxiliary information, which may include any one of the following: PRU location, PRU inference of the PRU location, PRS set associated with the PRU inference, first measurement result set associated with the PRU inference, and / or cell identifier associated with the PRU inference. WTRU 102 may receive auxiliary information. WTRU 102 may obtain inference of the WTRU location from an AI / ML model. WTRU 102 may verify the obtained inference based on the auxiliary information. WTRU 102 may send the verification result of the obtained inference to the network.
[0501] For example, WTRU 102 can send WTRU capability information to the network indicating support for AI / ML-based positioning.
[0502] For example, WTRU 102 can obtain inference from the AI / ML model after inputting measurement information associated with a first set of measurement results into the AI / ML model.
[0503] For example, WTRU 102 can perform a second measurement result set on the PRS set. WTRU 102 can obtain inference from the AI / ML model by inputting measurement information associated with the second measurement result set into the AI / ML model.
[0504] For example, WTRU 102 can verify the obtained inference, which includes determining (i) the difference between the obtained inference and (ii) the PRU location and / or PRU inference. The results of the verification may include information indicating the tuning of the WTRU to the AI / ML model based on the determined difference being less than a threshold.
[0505] For example, WTRU 102 can verify the obtained inference, which includes determining (i) the difference between the obtained inference and (ii) the PRU location and / or PRU inference. The result of the verification may include information indicating an identifier of the AI / ML model based on the determined difference being less than a threshold.
[0506] For example, WTRU 102 can verify the obtained inference, which includes determining (i) the difference between the obtained inference and (ii) the PRU location and / or PRU inference. The result of the verification may include information indicating an identifier of another AI / ML model based on the determined difference being greater than or equal to a threshold.
[0507] For example, WTRU 102 can verify the obtained inference, which includes determining (i) the difference between the obtained inference and (ii) the PRU location and / or PRU inference. The result of the verification may include information indicating an identifier of the backtracking method based on the determined difference being greater than or equal to a threshold.
[0508] In some representative embodiments, WTRU 102 may receive configuration information indicating multiple PRS sets. WTRU 102 may obtain first measurement information associated with the multiple PRS sets. WTRU 102 may select a first set among the multiple PRS sets based on the first measurement information associated with the multiple PRS sets. WTRU 102 may obtain inference of WTRU location from a first AI / ML model associated with the first set. WTRU 102 may obtain second measurement information associated with the multiple PRS sets. WTRU 102 may send information indicating a second AI / ML model to the network based on the second measurement information associated with the first set being below a threshold. For example, the second AI / ML model may be associated with a second set among the multiple PRS sets.
[0509] For example, WTRU 102 can use the first measurement information associated with the first set as input to the first AI / ML model to obtain inference of the WTRU location from the first AI / ML model.
[0510] For example, WTRU 102 can select the first set of multiple PRS sets based on the highest first measurement information associated with the first set among the multiple PRS sets.
[0511] For example, WTRU 102 can select the second set of multiple PRS sets based on the highest value of the second measurement information associated with the second set among the multiple PRS sets.
[0512] For example, the first measurement information may include the RSRP value of each of the plurality of PRS sets during a first time interval.
[0513] For example, the second measurement information may include the RSRP value of each of the plurality of PRS sets during the second time interval.
[0514] For example, WTRU 102 can receive information indicating the association between the second AI / ML model and the second set.
[0515] In some representative embodiments, WTRU 102 can receive configuration information from the network indicating multiple TA ranges associated with multiple AI / ML models. WTRU 102 can receive a TA command including information indicating a first TA value. WTRU 102 can select a first AI / ML model from the multiple AI / ML models that is associated with a first TA range that includes the first TA value. WTRU 102 can determine a second TA value. WTRU 102 can select a second AI / ML model from the multiple AI / ML models that is associated with a second TA range that includes the second TA value.
[0516] For example, WTRU 102 can obtain inference of WTRU location from the first AI / ML model.
[0517] For example, WTRU 102 can obtain inference of WTRU location from a second AI / ML model.
[0518] For example, a TA command may include information indicating a first TA value associated with a first TRP and a third TA value associated with a second TRP. The selection of a first AI / ML model may be based on the first TA value being less than the third TA value.
[0519] For example, WTRU 102 can determine a second TA value, which includes determining a fourth TA value, and the second TA value is associated with a third TRP, and the fourth TA value is associated with a fourth TRP. The selection of a second AI / ML model can be based on the second TA value being less than the fourth TA value.
[0520] For example, WTRU 102 can send information to the network indicating the selected first AI / ML model.
[0521] For example, WTRU 102 can send information to the network indicating the selected second AI / ML model.
[0522] In some representative embodiments, WTRU 102 can receive first configuration information from the network indicating a first set of PFLs associated with an AI / ML model. After moving from a first cell to a second cell, WTRU 102 can receive second configuration information indicating a second set of PFLs. WTRU 102 can send information indicating the AI / ML model to the network based on the fact that the first and second sets together have at least one PFL.
[0523] In some representative embodiments, WTRU 102 can receive first configuration information from the network indicating a first set of PFLs associated with an AI / ML model. WTRU 102 can receive second configuration information indicating a second set of PFLs after moving from a first cell to a second cell. WTRU 102 can send information indicating a backoff positioning method to the network based on the fact that each PFL in the first set is different from each PFL in the second set.
[0524] For example, WTRU 102 can infer the location of WTRU from an AI / ML model when it is in the first cell.
[0525] For example, WTRU 102 can obtain inference of WTRU location from an AI / ML model when it is in a second cell.
[0526] For example, an AI / ML model can be trained using measurement information associated with the multiple PFLs.
[0527] For example, the first configuration information may be received from the first TRP.
[0528] For example, the second configuration information may be received from the second TRP.
[0529] One or more embodiments provide a computer program including instructions that, when executed by one or more processors, cause such processors to perform encoding and / or decoding methods according to any of the above embodiments. One or more embodiments also provide a computer-readable storage medium storing instructions thereon for encoding or decoding video data according to the methods described above.
[0530] One or more embodiments provide a computer-readable storage medium storing video data generated according to the method described above. One or more embodiments also provide a method and apparatus for transmitting or receiving video data generated according to the method described above.
[0531] The embodiments described herein can be implemented, for example, as methods or processes, apparatuses, software programs, data streams, or signals. Even if discussed only in the context of a single implementation (e.g., as a method), implementations of such features can be implemented in other forms. Apparatuses can be implemented, for example, in suitable hardware, software, and firmware. Corresponding methods can be implemented, for example, in a processor.
[0532] Various numerical values are used in this application. Such specific values are for illustrative purposes, and the described embodiments are not limited to these specific values.
[0533] This document describes various methods, and such methods include one or more steps or actions for implementing the described methods. Unless a specific order of steps or actions is required for the proper operation of the method, the order and / or use of specific steps and / or actions may be modified or combined. Furthermore, terms such as "first," "second," etc., may be used in various embodiments to modify elements, components, steps, operations, etc., e.g., "first decoding" and "second decoding." The use of such terms does not imply an ordering of operations unless specifically required.
[0534] This disclosure may refer to the term "determine" for each piece of information. Determining information may include one or more of information such as estimates, calculations, predictions, or retrievals (e.g., from memory).
[0535] This disclosure may refer to the term "access" for various pieces of information. Accessing information may include, for example, receiving, retrieving (e.g., from memory), storing, moving, copying, calculating, determining, predicting, or estimating information. Similarly, this disclosure may refer to the term "receiving" for various pieces of information. Receiving information may include, for example, accessing or retrieving (e.g., from memory) information.
[0536] It should be understood that the use of any one of the following “ / ”, “and / or” and “…at least one of…” is intended to cover all possible options for the listed items, whether taken individually or in any combination thereof.
[0537] While specific embodiments have been described in conjunction with the accompanying drawings in the foregoing description, it should be understood that the embodiments described herein are merely examples and should not be construed as limiting the scope of this disclosure or the appended claims. Although features and elements have been described herein in specific combinations, those skilled in the art will appreciate that such features or elements can be used alone or in any combination with other features and elements. Therefore, it should be understood that the general teachings of this disclosure are not limited to the specific embodiments, implementations, and examples disclosed herein, but are intended to cover variations, modifications, and alternatives as defined by the appended claims and any and all their equivalents.
Claims
1. A wireless transmit / receive unit (WTRU), comprising: The transceiver, memory, and processor are configured as follows: The system sends information indicating the WTRU's ability to support artificial intelligence / machine learning (AI / ML) positioning to the base station associated with the first cell and to the WTRU that the system is moving from the second cell to the first cell. Send a request to the base station for auxiliary information for Positioning Reference Unit (PRU) inference. Receive instructions on the backtracking positioning method and configuration information for resources associated with the Positioning Reference Signal (PRS) set. Measuring the PRS set, Receive auxiliary information indicating the location of the PRU and / or a set of measurement information associated with the PRU inference of the location, and Based on a request from the base station, a report message is sent indicating any of the following: (i) a first AI / ML model used for positioning in the second cell at the WTRU is to be used in the first cell; (ii) fine-tuning of the first AI / ML model used for positioning in the second cell at the WTRU; (iii) a second AI / ML model to be used for positioning in the second cell at the WTRU; (iv) a back-off positioning method to be used for positioning in the second cell at the WTRU; and / or (v) inference generated based on the indicated set of measurement information and other measurement information associated with the measured PRS set.
2. The WTRU according to claim 1, wherein, Requests for auxiliary information include information indicating PRU location, PRU inference of PRU location, time interval, and / or any of the following PRU measurement information types.
3. The WTRU according to any one of claims 1-2, wherein, Requests for auxiliary information include information indicating the identifier of the first cell, the identifier of the second cell, and / or any item in the set of synchronization signal blocks (SSBs) associated with the first cell.
4. The WTRU according to any one of claims 1-3, wherein, The backoff positioning method is downlink time difference of arrival (DL-TDoA) positioning.
5. The WTRU according to any one of claims 1-4, wherein, The received auxiliary information is associated with the first time interval.
6. The WTRU according to claim 5, wherein, The PRS set was measured during the first time interval.
7. The WTRU according to any one of claims 1-6, wherein, The transceiver, memory, and processor are configured to obtain inference from the first AI / ML model based on the indicated set of measurement information and other measurement information associated with the measured PRS set.
8. The WTRU according to claim 7, wherein, The report information includes information indicating that the first AI / ML model used for positioning in the second cell should be used in the first cell, based on the fact that the difference between the inference obtained from the first AI / ML model and the PRU inference is less than a first threshold.
9. The WTRU according to claim 7, wherein, The report information includes information indicating fine-tuning of the first AI / ML model used for positioning in the second cell at the WTRU, based on the fact that the difference between the inference obtained from the first AI / ML model and the PRU inference is greater than a first threshold and less than a second threshold.
10. The WTRU according to claim 7, wherein, The report information includes information indicating that a second AI / ML model or backoff positioning method should be used at the WTRU for positioning in the second cell, based on the fact that the difference between the inference obtained from the first AI / ML model and the PRU inference is greater than a second threshold.
11. A method implemented by a wireless transmit / receive unit (WTRU), the method comprising: Send information indicating the WTRU's ability to support artificial intelligence / machine learning (AI / ML) positioning to the base station associated with the first cell and to the WTRU moving from the second cell to the first cell; Send a request to the base station for auxiliary information for positioning reference unit (PRU) inference; Receive instructions on the backtracking positioning method and configuration information of resources associated with the Positioning Reference Signal (PRS) set; Measure the PRS set; Receive auxiliary information indicating the location of the PRU and / or a set of measurement information associated with the PRU inference of the location; as well as Based on a request from the base station, a report message is sent indicating any of the following: (i) a first AI / ML model used for positioning in the second cell at the WTRU is to be used in the first cell; (ii) fine-tuning of the first AI / ML model used for positioning in the second cell at the WTRU; (iii) a second AI / ML model to be used for positioning in the second cell at the WTRU; (iv) a back-off positioning method to be used for positioning in the second cell at the WTRU; and / or (v) inference generated based on the indicated set of measurement information and other measurement information associated with the measured PRS set.
12. The method according to claim 11, wherein, Requests for auxiliary information include information indicating PRU location, PRU inference of PRU location, time interval, and / or any of the following PRU measurement information types.
13. The method according to any one of claims 11-12, wherein, Requests for auxiliary information include information indicating the identifier of the first cell, the identifier of the second cell, and / or any item in the set of synchronization signal blocks (SSBs) associated with the first cell.
14. The method according to any one of claims 11-13, wherein, The backoff positioning method is downlink time difference of arrival (DL-TDoA) positioning.
15. The method according to any one of claims 11-14, wherein, The received auxiliary information is associated with the first time interval.
16. The method according to claim 15, wherein, The PRS set was measured during the first time interval.
17. The method according to any one of claims 11-16, further comprising: Inference is derived from the first AI / ML model based on the indicated set of measurement information and other measurement information associated with the measured PRS set.
18. The method according to claim 17, wherein, The report information includes information indicating that the first AI / ML model used for positioning in the second cell should be used in the first cell, based on the fact that the difference between the inference obtained from the first AI / ML model and the PRU inference is less than a first threshold.
19. The method of claim 17, wherein, The report information includes information indicating fine-tuning of the first AI / ML model used for positioning in the second cell at the WTRU, based on the fact that the difference between the inference obtained from the first AI / ML model and the PRU inference is greater than a first threshold and less than a second threshold.
20. The method of claim 17, wherein, The report information includes information indicating that a second AI / ML model or backoff positioning method should be used at the WTRU for positioning in the second cell, based on the fact that the difference between the inference obtained from the first AI / ML model and the PRU inference is greater than a second threshold.
