Updating machine learning models for localization

Abstract

A method is disclosed, comprising: obtaining a machine learning model for positioning, where the machine learning model is trained based on first training data associated with one or more first network nodes; transmitting information including at least one of: a machine learning model, a request to update the machine learning model based on the second training data, or a request to provide the second training data for updating the machine learning model at the device; receiving at least one of: a message indicating the updated machine learning model, or second training data; and sending the updated machine learning model to at least one or more second network nodes.

Description

Technical Field

[0001] The following example embodiments relate to wireless communications and to updating a machine learning model for positioning. Background Art

[0002] Positioning technology can be used to estimate the physical location of a device. It is desirable to improve positioning accuracy so as to more accurately estimate the location of a device. Summary of the invention

[0003] The scope of protection sought by various example embodiments is set out by the independent claims. Example embodiments and features described in this specification that do not fall within the scope of the independent claims, if any, are to be construed as useful examples for understanding the various embodiments.

[0004] According to one aspect, a device is provided, comprising at least one processor and at least one memory, wherein the at least one memory stores instructions that, when executed by the at least one processor, cause the device to at least: obtain a machine learning model for positioning, wherein the machine learning model is trained based on first training data associated with one or more first network nodes; send information comprising at least one of the following: a machine learning model, a request for updating the machine learning model based on second training data, or a request for providing second training data for updating the machine learning model at the device; receive at least one of the following: a message indicating an updated machine learning model, or second training data; and send the updated machine learning model to at least one or more second network nodes.

[0005] According to another aspect, a device is provided, comprising: a component for obtaining a machine learning model for positioning, wherein the machine learning model is trained based on first training data related to one or more first network nodes; a component for sending information including at least one of the following: the machine learning model, a request for updating the machine learning model based on second training data, or a request for providing second training data for updating the machine learning model at the device; a component for receiving at least one of the following: a message indicating an updated machine learning model, or the second training data; and a component for sending the updated machine learning model at least to one or more second network nodes.

[0006] According to another aspect, a method is provided, comprising: obtaining, by a device, a machine learning model for positioning, wherein the machine learning model is trained based on first training data related to one or more first network nodes; sending, by the device, information comprising at least one of the following: the machine learning model, a request for updating the machine learning model based on second training data, or a request for providing second training data for updating the machine learning model at the device; receiving, by the device, at least one of the following: a message indicating an updated machine learning model, or the second training data; and sending, by the device, the updated machine learning model at least to one or more second network nodes.

[0007] According to another aspect, a computer program is provided that includes instructions that, when executed by an apparatus, cause the apparatus to perform at least the following: obtain a machine learning model for positioning, wherein the machine learning model is trained based on first training data associated with one or more first network nodes; send information including at least one of the following: a machine learning model, a request for updating the machine learning model based on second training data, or a request for providing second training data for updating the machine learning model at the apparatus; receive at least one of the following: a message indicating an updated machine learning model, or second training data; and send the updated machine learning model at least to one or more second network nodes.

[0008] According to another aspect, a computer-readable medium comprising program instructions is provided, which, when executed by an apparatus, causes the apparatus to perform at least the following: obtain a machine learning model for positioning, wherein the machine learning model is trained based on first training data related to one or more first network nodes; send information comprising at least one of the following: a machine learning model, a request for updating the machine learning model based on second training data, or a request for providing second training data for updating the machine learning model at the apparatus; receive at least one of the following: a message indicating an updated machine learning model, or second training data; and send the updated machine learning model to at least one or more second network nodes.

[0009] According to another aspect, a non-transitory computer-readable medium comprising program instructions is provided, which, when executed by an apparatus, causes the apparatus to perform at least the following: obtain a machine learning model for positioning, wherein the machine learning model is trained based on first training data associated with one or more first network nodes; send information comprising at least one of the following: a machine learning model, a request for updating the machine learning model based on second training data, or a request for providing second training data for updating the machine learning model at the apparatus; receive at least one of the following: a message indicating an updated machine learning model, or second training data; and send the updated machine learning model at least to one or more second network nodes.

[0010] According to another aspect, a device is provided, comprising at least one processor and at least one memory, wherein the at least one memory stores instructions which, when executed by the at least one processor, cause the device to at least: receive information comprising at least one of: a machine learning model for positioning, a request for updating the machine learning model at the device based on second training data, or a request for providing second training data for updating the machine learning model, wherein the machine learning model has been trained based on first training data associated with one or more first network nodes; obtain the second training data; and send at least one of: a message indicating an updated machine learning model or the second training data.

[0011] According to another aspect, a device is provided, comprising: a component for receiving information including at least one of the following: a machine learning model for positioning, a request for updating the machine learning model at the device based on second training data, or a request for providing second training data for updating the machine learning model, wherein the machine learning model has been trained based on first training data related to one or more first network nodes; a component for obtaining the second training data; and a component for sending at least one of the following: a message indicating an updated machine learning model or the second training data.

[0012] According to another aspect, a method is provided, comprising: receiving, by a device, information comprising at least one of: a machine learning model for positioning, a request for updating the machine learning model at the device based on second training data, or a request for providing second training data for updating the machine learning model, wherein the machine learning model has been trained based on first training data associated with one or more first network nodes; acquiring, by the device, the second training data; and sending, by the device, at least one of: a message indicating the updated machine learning model, or the second training data.

[0013] According to another aspect, a computer program comprising instructions is provided, which, when executed by a device, causes the device to perform at least the following: receive information comprising at least one of the following: a machine learning model for positioning, a request for updating the machine learning model at the device based on second training data, or a request for providing second training data for updating the machine learning model, wherein the machine learning model has been trained based on first training data associated with one or more first network nodes; obtain the second training data; and send at least one of the following: a message indicating the updated machine learning model, or the second training data.

[0014] According to another aspect, a computer-readable medium comprising program instructions is provided, which, when executed by an apparatus, causes the apparatus to perform at least the following: receive information comprising at least one of: a machine learning model for positioning, a request for updating the machine learning model at the apparatus based on second training data, or a request for providing second training data for updating the machine learning model, wherein the machine learning model has been trained based on first training data associated with one or more first network nodes; obtain the second training data; and send at least one of: a message indicating the updated machine learning model, or the second training data.

[0015] According to another aspect, a non-transitory computer-readable medium comprising program instructions is provided, which, when executed by a device, causes the device to perform at least the following: receive information comprising at least one of the following: a machine learning model for positioning, a request for updating the machine learning model at the device based on second training data, or a request for providing second training data for updating the machine learning model, wherein the machine learning model has been trained based on first training data associated with one or more first network nodes; obtain the second training data; and send at least one of the following: a message indicating the updated machine learning model, or the second training data. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Hereinafter, various example embodiments will be described in more detail with reference to the accompanying drawings, in which Figure 1 An example of a cellular communication network is shown; Figure 2 An example of a positioning scenario is shown; Figure 3 shows a signaling diagram according to an example embodiment; Figure 4 shows a signaling diagram according to an example embodiment; Figure 5 shows a flow chart according to an example embodiment; Figure 6 shows a flow chart according to an example embodiment; Figure 7 An example of an apparatus is shown; Figure 8 An example of an apparatus is shown; Fig. 9 An example of an apparatus is shown; Fig.10 An example of an artificial neural network is shown; and Fig.11 An example of a compute node is shown. DETAILED DESCRIPTION

[0017] The following embodiments are exemplary. Although the specification may refer to "one", "an" or "some" embodiments in several places in the text, this does not necessarily mean that the same (multiple) embodiments are referenced each time, or that a particular feature applies only to a single embodiment. Individual features of different embodiments may also be combined to provide other embodiments.

[0018] In the following, different example embodiments will be described using radio access architectures based on Advanced Long Term Evolution (LTE-Advanced, LTE-A), New Radio (NR, 5G), Beyond 5G or Sixth Generation (6G) as examples of access architectures to which the example embodiments may be applied, however, the example embodiments are not limited to such architectures. It will be apparent to a person skilled in the art that the example embodiments may also be applied to other types of communication networks with suitable components by appropriately adjusting the parameters and procedures. Some examples of other options for suitable systems may be Universal Mobile Telecommunications System (UMTS) Radio Access Network (UTRAN or E-UTRAN), Long Term Evolution (LTE, essentially the same as E-UTRA), Wireless Local Area Network (WLAN or Wi-Fi), Worldwide Interoperability for Microwave Access (WiMAX), Personal Communications Service (PCS), Wideband Code Division Multiple Access (WCDMA), systems using Ultra-Wideband (UWB) technology, sensor networks, Mobile Ad Hoc Networks (MANETs), and Internet Protocol Multimedia Subsystem (IMS), or any combination thereof.

[0019] Figure 1 An example of a simplified system architecture showing some elements and functional entities is depicted, all of which are logical units, the implementation of which may differ from what is shown. Figure 1 The connections shown are logical connections; the actual physical connections may be different. It will be apparent to those skilled in the art that the system may also include in addition to Figure 1 Other functions and structures than those shown.

[0020] However, the exemplary embodiments are not limited to the systems given as examples, but a person skilled in the art may apply the solutions to other communication systems having the necessary properties.

[0021] Figure 1 The example of FIG. 1 shows a portion of an exemplary radio access network.

[0022] Figure 1A user device 100 and a user device 102 are shown, which are configured to be wirelessly connected to an access node (AN) 104 (such as an evolved Node B (abbreviated as eNB or eNodeB) or a next generation Node B (abbreviated as gNB or gNodeB)) on one or more communication channels in a radio cell, thereby providing a radio cell. The physical link from the user equipment to the access node can be referred to as an uplink (UL) or reverse link, and the physical link from the access node to the user equipment can be referred to as a downlink (DL) or forward link. The user equipment can also communicate directly with another user equipment via a side link (SL) communication. It should be understood that the access nodes or their functionality can be implemented by using any node, host, server or access point entity suitable for such purpose.

[0023] The communication system may include more than one access node, in which case the access nodes may also be configured to communicate with each other via wired or wireless links designed for this purpose. These links may be used for signaling purposes and also for routing data from one access node to another access node. An access node may be a computing device configured to control the radio resources of the communication system to which it is coupled. An access node may also be referred to as a base station, a base transceiver station (BTS), an access point, or any other type of interface device including a relay station capable of operating in a wireless environment. An access node may include or be coupled to a transceiver. From the transceiver of the access node, a connection may be provided to an antenna unit that establishes a bidirectional radio link to a user device. The antenna unit may include multiple antennas or antenna elements. The access node may also be connected to a core network 110 (CN or next generation core NGC). Depending on the deployed technology, the counterpart to which the access node can be connected on the CN side can be a serving gateway (S-GW, routing and forwarding user data packets), a packet data network gateway (P-GW) for providing user equipment with connectivity to external packet data networks, a user plane function (UPF), a mobility management entity (MME) or an access and mobility management function (AMF), etc.

[0024] Regarding positioning, the service-based architecture (core network) may include an AMF 111 and a location management function (LMF) 112. The AMF may provide location information for call processing, policy, and charging to other network functions in the core network and to other entities requesting for positioning of terminal devices. The AMF may receive and manage location requests from several sources: mobile originated location requests (MO-LR) from user equipment and mobile terminated location requests (MT-LR) from other functions of the core network or from other network elements. The AMF may select an LMF for a given request and use its positioning service to trigger a positioning session. The LMF may then perform positioning when such a request is received from the AMF. The LMF may manage the resources and timing of positioning activities. The LMF may request the positioning of the user equipment from one or more access nodes using the Namf_Communication service on the NL1 interface, or the LMF may communicate with the user equipment over N1 for UE-based or UE-assisted positioning. Positioning may include an estimate of the location, and additionally, the LMF may also estimate the movement or accuracy of the location information upon request. In terms of connection, the AMF may be between the access node and the LMF, and is therefore closer to the access node than the LMF.

[0025] A user equipment illustrates one type of apparatus to which resources on the air interface may be allocated and assigned, and thus any features described herein with a user equipment may be implemented with a corresponding apparatus, such as a relay node.

[0026] An example of such a relay node may be a layer 3 relay towards an access node (self-backhaul relay). A self-backhaul relay node may also be referred to as an integrated access and backhaul (IAB) node. An IAB node may include two logical parts: a mobile terminal (MT) part, which is responsible for the backhaul link(s) (i.e., the link(s) between the IAB node and the donor node, also referred to as the parent node); and a distributed unit (DU) part, which is responsible for the access link(s) (i.e., the sub-link(s) between the IAB node and the user equipment(s)) and / or the sub-link(s) between the IAB node and other IAB nodes (multi-hop scenario).

[0027] Another example of such a relay node may be a layer 1 relay referred to as a repeater. A repeater may amplify a signal received from an access node and forward it to a user equipment and / or amplify a signal received from a user equipment and forward it to an access node.

[0028] User equipment may also be referred to as a subscriber unit, mobile station, remote terminal, access terminal, user terminal, terminal device, or user equipment (UE), just to mention some names or devices. User equipment may refer to a portable computing device including a wireless mobile communication device operating with or without a subscriber identity module (SIM), including but not limited to the following types of devices: mobile station (mobile phone), smart phone, personal digital assistant (PDA), mobile phone, device using a wireless modem (alarm or measurement device, etc.), laptop and / or touch screen computer, tablet computer, game console, notebook, multimedia device, reduced capability (RedCap) device, wireless sensor device or any device integrated in a vehicle.

[0029] It should be understood that the user device may also be an almost exclusive uplink-only device, an example of which may be a camera or video camera that loads images or video clips to a network. The user device may also be a device with the ability to operate in an Internet of Things (IoT) network, where an IoT network is a scenario in which objects can be provided with the ability to migrate data over a network without requiring human-to-human or human-to-computer interaction. The user device may also utilize the cloud. In some applications, the user device may include a small portable or wearable device (such as a watch, headset, or glasses) with a radio component, and the calculation may be performed in the cloud or in another user device. The user device (or in some example embodiments, a layer 3 relay node) may be configured to perform one or more of the user device functions.

[0030] The various techniques described herein may also be applied to cyber-physical systems (CPS) (systems of cooperating computing elements that control physical entities). CPS may enable the implementation and exploitation of a large number of interconnected ICT devices (sensors, actuators, processor microcontrollers, etc.) embedded in physical objects at different locations. Mobile cyber-physical systems, in which the physical systems in question may have inherent mobility, are a subcategory of cyber-physical systems. Examples of mobile physical systems include mobile robots and electronic devices transported by humans or animals.

[0031] Additionally, although the device has been depicted as a single entity, different units, processors and / or memory units may be implemented. Figure 1 Not all are shown).

[0032] 5G enables the use of multiple input-multiple output (MIMO) antennas, which are much more base stations or nodes than LTE (the so-called small cell concept), including macro sites that operate in collaboration with smaller stations and adopt various radio technologies depending on service requirements, use cases and / or available spectrum. 5G mobile communications can support a wide range of use cases and related applications, including video streaming, augmented reality, different data sharing methods, and various forms of machine type applications (such as (massive) machine type communications (mMTC), including vehicle safety, different sensors and real-time control). 5G can have multiple radio interfaces. That is, below 6GHz, cmWave and mmWave, and also integrated with existing traditional radio access technologies (such as LTE). At least in the early stages, integration with LTE can be implemented as a system in which macro coverage can be provided by LTE, and 5G radio interface access can come from a small cell by aggregation to LTE. In other words, 5G can support both inter-RAT operability (such as LTE-5G) and inter-RI operability (inter-radio interface operability, such as below 6GHz-cmWave-mmWave). One of the concepts considered for use in 5G networks may be network slicing, in which multiple independent and dedicated virtual subnetworks (network instances) can be created within essentially the same infrastructure to run services with different requirements for latency, reliability, throughput and mobility.

[0033] The current architecture in LTE networks can be completely distributed in the radio and completely concentrated in the core network. Low-latency applications and services in 5G may require content to be close to the radio, which leads to local interruptions and multi-access edge computing (MEC). 5G may enable analysis and knowledge generation to occur at the data source. This approach may require the use of resources that may not be continuously connected to the network, such as laptops, smart phones, tablets, and sensors. MEC can provide a distributed computing environment for application and service hosting. It can also have the ability to store and process content very close to cellular subscribers to obtain faster response times. Edge computing can cover a wide variety of technologies, such as wireless sensor networks, mobile data acquisition, mobile signature analysis, collaborative distributed peer-to-peer self-organizing networking and processing, and can also be classified as local cloud / fog computing and grid / grid computing, dew computing, mobile edge computing, micro-cloud, distributed data storage and retrieval, autonomous self-healing networks, remote cloud services, augmented and virtual reality, data caching, Internet of Things (large-scale connectivity and / or latency critical), critical communications (autonomous vehicles, traffic safety, real-time analysis, time-critical control, healthcare applications).

[0034] The communication system can also communicate with or utilize services provided by one or more other networks 113, such as a public switched telephone network or the Internet. The communication network can also support the use of cloud services, for example, at least a portion of the core network operations can be performed as a cloud service (this is in the Figure 1 114). The communication system may also include a central control entity or the like which provides facilities for networks of different operators to collaborate, for example in spectrum sharing.

[0035] The access node can also be divided into: a radio unit (RU), which includes a radio transceiver (TRX), i.e., a transmitter (Tx) and a receiver (Rx); one or more distributed units (DU) 105, which can be used for so-called layer 1 (L1) processing and real-time layer 2 (L2) processing; and a central unit (CU) 108 (also referred to as a centralized unit), which can be used for non-real-time L2 and layer 3 (L3) processing. The CU 108 can be connected to one or more DUs 105, for example, via an F1 interface. Such a division can achieve the centralization of the CU relative to the cell site and the DU, while the DU can be more distributed and can even be retained at the cell site. The CU and DU together can also be referred to as a baseband or baseband unit (BBU). The CU and DU can also be included in a radio access point (RAP).

[0036] The CU 108 may be defined as a logical node that hosts the higher layer protocols of the access node, such as radio resource control (RRC), service data adaptation protocol (SDAP), and / or packet data convergence protocol (PDCP). The DU 105 may be defined as a logical node that hosts the radio link control (RLC), media access control (MAC), and / or physical (PHY) layers of the access node. The operation of the DU may be at least partially controlled by the CU. The CU may include a control plane (CU-CP), which may be defined as a logical node that hosts the control plane portion of the RRC and the PDCP protocol of the CU for the access node. The CU may also include a user plane (CU-UP), which may be defined as a logical node that hosts the user plane portion of the PDCP protocol and the SDAP protocol of the CU for the access node.

[0037] The cloud computing platform can also be used to run CU 108 and / or DU 105. The CU can run in the cloud computing platform, which can be referred to as a virtualized CU (vCU). In addition to the vCU, there can also be a virtualized DU (vDU) running in the cloud computing platform. In addition, there can also be a combination in which the DU can use a so-called bare metal solution, such as an application specific integrated circuit (ASIC) or a customer-specific standard product (CSSP) system-on-chip (SoC) solution. It should also be understood that the functional distribution between the above-mentioned access node units or different core network operations and access node operations may be different.

[0038] The edge cloud can be brought to the radio access network (RAN) by leveraging network function virtualization (NFV) and software defined networking (SDN). Using the edge cloud can mean access node operations to be performed at least in part in a server, host or node operably coupled to a remote radio head (RRH) or a radio unit (RU), or an access node including a radio portion. Node operations can also be distributed among multiple servers, nodes or hosts. The application of the cloud RAN architecture enables RAN real-time functions to be performed on the RAN side (e.g., in DU 105), and non-real-time functions to be performed in a centralized manner (e.g., in CU 108).

[0039] It should also be understood that the functional distribution between core network operations and access node operations may be different from that of LTE, or even non-existent. Some other technological advances that may be used include big data and all-IP, which may change the way networks are built and managed. 5G (or new radio, NR) networks may be designed to support multiple layers, where MEC servers may be placed between the core and access nodes. It should be understood that MEC may also be applied to 4G networks.

[0040] 5G may also make use of non-terrestrial communications (e.g., satellite communications) to enhance or supplement the coverage of 5G services, for example by providing backhaul. Possible use cases may be to provide service continuity for machine-to-machine (M2M) or Internet of Things (IoT) devices or for passengers in a vehicle, or to ensure service availability for critical communications and future rail / maritime / aeronautical communications. Satellite communications may make use of geosynchronous orbit (GEO) satellite systems as well as low earth orbit (LEO) satellite systems, in particular mega-constellations (systems in which hundreds of (nano) satellites are deployed). A given satellite 106 in a mega-constellation may cover several satellite-enabled network entities creating a terrestrial cell. A terrestrial cell may be created by a terrestrial relay node or by an access node 104 located on the ground or in a satellite.

[0041] 6G networks are expected to adopt flexible decentralized and / or distributed computing systems and architectures and ubiquitous computing, where local spectrum licensing, spectrum sharing, infrastructure sharing and intelligent automation management are determined by mobile edge computing, artificial intelligence, short packet communication and blockchain technology. Key features of 6G may include management and control functions of intelligent connections, programmability, integrated sensing and communication, reduced energy usage, trusted infrastructure, scalability and affordability. In addition to these, 6G also targets new use cases that cover the integration of positioning and sensing capabilities into system definitions to unify user experience across the physical and digital worlds.

[0042] It is obvious to a person skilled in the art that the depicted system is only an example of a part of a radio access system, and in practice, the system may include multiple access nodes, a user equipment may access multiple radio cells, and the system may also include other devices, such as physical layer relay nodes or other network elements, etc. At least one of the access nodes may be a Home eNodeB or a Home gNodeB.

[0043] Additionally, in a geographical area of ​​a radio communication system, a plurality of different kinds of radio cells and a plurality of radio cells may be provided. A radio cell may be a macro cell (or umbrella cell), which may be a large cell with a diameter of up to tens of kilometers, or a smaller cell, such as a micro cell, a femto cell or a pico cell. Figure 1 The access node(s) may provide any kind of these cells. A cellular radio system may be implemented as a multi-layer network comprising several radio cells. In a multi-layer network, one access node may provide one or more radio cells, and therefore multiple access nodes may be required to provide such a network structure.

[0044] In order to meet the need for improving the deployment and performance of communication systems, the concept of "plug-and-play" access nodes can be introduced. In addition to the home eNodeB or home gNodeB, a network capable of using "plug-and-play" access nodes can also include a home nodeB gateway or HNB-GW ( Figure 1 (not shown). The HNB-GW, which may be installed within an operator's network, may aggregate services from a large number of home eNodeBs or home gNodeBs back to the core network.

[0045] Positioning techniques can be used to estimate the physical location of a user equipment. In this document, the user equipment to be positioned is referred to as a target UE or target user equipment. For example, the positioning techniques used in NR can be based on at least one of the following: time difference of arrival (TDoA), time of arrival (TOA), time of departure (TOD), round trip time (RTT), angle of departure (AoD), angle of arrival (AoA), and / or carrier phase.

[0046] In Uu positioning (UL / DL positioning), multiple transmit and receive points (TRPs) in known locations can send and / or receive one or more positioning reference signals (PRS) to / from the target UE. In the uplink, a sounding reference signal (SRS) can be used as a positioning reference signal. For example, multipoint positioning techniques can then be used to locate (i.e., position) the target UE relative to the TRP. One of these TRPs can be used as a positioning anchor, and the difference in TDoA can be calculated relative to the positioning anchor. The positioning anchor may also be referred to herein as an anchor, anchor node, multipoint positioning anchor, or reference point.

[0047] Sidelink (SL) positioning refers to a positioning method in which the target UE utilizes a sidelink (i.e., a direct device-to-device link) to locate itself in an absolute manner (in the case of absolute positioning) or in a relative manner (in the case of relative positioning). For UE-assisted positioning (in SL positioning and Uu positioning), the target UE can utilize a sidelink to obtain positioning measurements and report the measurements to a network entity, such as a location management function (LMF). Sidelink positioning can also be used to obtain ranging information.

[0048] Ranging means determining the distance between two UEs and / or the direction of one UE relative to another UE via a direct device connection.

[0049] Absolute positioning means estimating the position of the target UE in two-dimensional or three-dimensional geographic coordinates (eg, latitude, longitude and / or altitude) within a coordinate system.

[0050] Relative positioning means estimating the position of the target UE relative to other network elements or relative to other UEs.

[0051] SL positioning may be based on the transmission of a Sidelink Positioning Reference Signal (SL PRS) by multiple anchor UEs (anchor user equipments), wherein the SL PRS is received and measured by the target UE to enable positioning of the target UE within the precise delay and accuracy requirements of the corresponding SL positioning session. Alternatively or additionally, the target UE may send an SLPRS to be received and measured by the anchor UE.

[0052] An anchor UE may be defined as a UE that supports positioning of a target UE, for example by sending and / or receiving reference signals for positioning (e.g., SL PRS) via the SL interface. This may be similar to UL / DL based positioning, where the gNB may act as an anchor to send and / or receive reference signals to / from the target UE for positioning.

[0053] SL PRS refers to a reference signal sent through SL for positioning purposes.

[0054] In addition, a positioning reference unit (PRU) may be used in a positioning session to increase the positioning accuracy for positioning a target UE. A PRU is a reference device at a known location that performs measurements for generating correction data that may be used to refine the position estimate of a target UE in an area, thereby improving the positioning accuracy for positioning the target UE. For example, a UE with a known location may be used as a PRU.

[0055] A PRU at a known location may perform positioning measurements, such as reference signal time difference (RSTD), reference signal received power (RSRP), UE receive-transmit time difference measurements, etc., and report these measurements to a location server, such as an LMF. In addition, the PRU may send a UL SRS for positioning to enable the TRP to measure and report UL positioning measurements (e.g., relative time of arrival, UL-AoA, gNB receive-transmit time difference, etc.) from the PRU at a known location. The location server may compare the PRU measurements with the expected measurements at the known PRU location to determine correction data for (multiple) other nearby target UEs. The DL and / or UL location measurements for (multiple) other target UEs may then be corrected based on the previously determined correction data.

[0056] The PRUs may also serve as positioning anchors for the target UE, or they may simply provide correction data (eg, to the LMF) to help locate the target UE.

[0057] In other words, a PRU located at a known position can act as a reference target UE so that its calculated position is compared with their known position. The comparison of the known position and the estimated position can generate correction data that can be used in the position estimation process of other target UEs in the vicinity, assuming that the same or similar accuracy determination effect applies to both the position of the PRU and the positions of the other target UEs. The correction data can then be used to fine-tune the position estimate of the target UE, thereby improving positioning accuracy.

[0058] The position of the target UE may be calculated at the network, for example at the LMF (in the case of LMF-based positioning), or at the target UE itself (in the case of UE-based positioning). Measurements for positioning may be performed at the UE side (for example, in the case of DL or SL positioning) or at the network side (for example, in the case of UL positioning).

[0059] Figure 2 An example is shown in which one or more PRUs 202, 202A, 202B are used to locate a target UE 200. The PRUs may be configured to transmit reference signals measured for the purpose of locating the target UE 200. The target UE 200 may also transmit reference signals for the purpose of locating the target UE 200. One or more access nodes 204, 204A, 204B may measure reference signals received from (multiple) PRUs 202, 202A, 202B and from the target UE 200. In the case of sidelink positioning, the target UE 200 may measure reference signals received from (multiple) PRUs 202, 202A, 202B, and / or (multiple) PRUs 202, 202A, 202B may measure reference signals received from the target UE and / or from other UEs or PRUs. The measurement parameters (measurement data) derived from the received reference signals may include, for example, reference signal reception time, reference signal time difference (RSTD), reference signal arrival angle, and / or RSRP. The measurement data may be reported to a network element acting as a location management function (LMF) 212, which is configured to perform positioning based on the measurement data. The LMF 212 may estimate the position of the target UE 200 based on the received measurement data and the known positions of the (multiple) PRUs measured by the reporting (multiple) access nodes. For example, a position estimation function used in a real-time dynamic positioning (RTK) application in a global navigation satellite system may be employed. As an example, if the measurement indicates that the signal received from one of the (multiple) PRUs 202 and the target UE 200 has a high correlation, the position of the target UE 200 may be estimated to be close to the PRU 202 and further away from the other PRUs 202A, 202B. A correction to the position from a given PRU 202A, 202B may be calculated based on the measurement data, for example, by using the difference between the measurement data associated with the target UE 200 and the measurement data associated with the closest PRU 202. For example, multi-lateration measurements (multiple measurements of RSRP, RSTD, and / or other parameters) may indicate that the target UE 200 is in a certain direction from the nearest PRU 202, and a correction may be made to that direction.

[0060] The NR air interface may be enhanced with features that enable support for artificial intelligence (AI) and / or machine learning (ML) based algorithms for enhanced performance and / or reduced complexity and overhead. Some use cases for such AI / ML techniques may include (but are not limited to) channel state information (CSI) feedback enhancement (e.g., overhead reduction, improved accuracy, prediction), beam management (e.g., beam prediction in the time and / or spatial domain for overhead and latency reduction, beam selection accuracy improvement), and positioning accuracy enhancement (e.g., in scenarios with severe non-line of sight, positioning reference signaling and measurement reporting overhead reduction, positioning accuracy with limited availability of tagged data, scenarios with devices having significant RF impairments / defects that affect positioning measurements, etc.).

[0061] Using AI / ML techniques for positioning accuracy enhancement can involve performing training for positioning purposes at a central ML unit such as a location management function (LMF) or a 5G network data analysis function (NWDAF). The NWDAF can run data analytics to generate insights and take actions to enhance the user experience, including positioning use cases.

[0062] Training the ML model at the central ML unit can be done, for example, according to the following process (described in steps 1-4 below): 1) A set of data collection devices may be deployed at selected locations. Alternatively, data collection devices may be randomly selected from a given geographic area. For example, these data collection devices may be PRUs. In the following, for simplicity, the data collection devices are referred to as PRUs, but any other type of data collection device may be used instead of PRUs. 2) The PRU performs on-site positioning measurements and reports the measurements to the central ML unit. 3) The central ML unit uses the simulation tool to generate (simulated) positioning measurements. It should be noted that step 3 can be performed instead of step 2, or can also be performed in addition to step 2. 4) The central ML unit uses the above positioning measurements (reported measurements and / or simulated measurements) to train a general ML-based positioning framework. In this paper, this general ML-based localization framework is denoted as GLoc.

[0063] The trained GLoc can then be deployed at a network node running the ML process and / or algorithm. Such a network node is referred to herein as a host. The host can be of different types, where the type can be defined with respect to the host's radio frequency (RF) and / or baseband capabilities, form factor, or target positioning key performance indicators (KPIs). The host executing the ML process can be, for example, a target UE, a PRU, and a radio access network (e.g., a gNB, a TRP, and / or a location management component LMC) to enhance positioning accuracy.

[0064] However, the problem with the generic ML-based positioning framework (GLoc) is that it does not take into account the specific RF limitations (also known as RF imperfections) of the deployed host type (e.g., handheld UE, roadside unit, or gNB). RF limitations / imperfections may depend on hardware limitations of different antenna configurations and form factors, analog-to-digital conversion (ADC) resolution, crystal oscillators, etc. Various RF imperfections may introduce a combination of: carrier frequency offset, sampling time offset, transmit / receive beam offset, clock offset and drift, phase noise, etc.

[0065] These RF imperfections may translate into additional phase rotations and delays of the positioning signal by the RF chain as observed at the baseband receiver. As a result, the positioning entity (e.g., UE, TRP, etc.) hosting the GLoc framework may experience certain RF-based signal delays / rotations that are not accounted for in GLoc and are incorrectly absorbed into the positioning measurements. Such imperfections may be different for different host types. For example, a PRU, UE, or gNB hosting GLoc will require GLoc to be adapted to its own RF specific characteristics.

[0066] Therefore, there may not currently be a one-size-fits-all GLoc framework that can meet the high positioning accuracy requirements at the cm level for, for example, NR Release 18. Thus, a device of a given host type may need to adapt GLoc to its own RF characteristics, such as antenna form factor and configuration.

[0067] However, some example embodiments are described below using the principles and terminology of 5G NR technology without limiting the example embodiments to 5G NR communication systems.

[0068] Some example embodiments may solve the above problems by providing a method that customizes a general GLoc framework for host type specific ML positioning. In other words, some example embodiments may be used to adapt a general machine learning model (e.g., trained using training data collected from different types of NR elements) to a machine learning model suitable for compensating for inherent errors specific to a given host type. Therefore, some example embodiments may use AI / ML techniques to provide positioning accuracy enhancement.

[0069] Meta-learning in the context of AI / ML refers to tailoring a generic model to a specific type of entity and / or task (e.g., trained using features extracted from heterogeneous sources). A sub-branch of meta-learning is transfer learning (TL). The goal of TL is to adapt an already trained model to perform the same task but on a different entity type.

[0070] Some example embodiments may provide a TL framework for positioning, by which a generic GLoc framework for positioning may be customized for a specific NR element host type. More specifically, before the GLoc framework is deployed on a large scale, the GLoc may be refined based on at least the inherent characteristics (e.g., RF limitations) of the NR element type (e.g., target UE, PRU, or gNB).

[0071] Some example embodiments allow a central ML unit (e.g., LMF) to select a NR head unit (NR-HU) as a representative for a given NR element type, and therefore as a representative for a given expected inherent distortion range. The GLoc model can then be refined by or with the help of the NR-HU so that it is tailored to compensate for the distortion specific to that NR element type.

[0072] To perform model adaptation, the central ML unit may provide the NR-HU with a generic GLoc framework and the parameters considered to acquire such a framework (such as the capabilities and RF deficiencies of the device used to generate the GLoc) and the corresponding training process.

[0073] The generic GLoc can then be refined based on the individual types of NR-HUs (e.g., capabilities and RF deficiencies). Finally, the adapted model can be reported to the central ML unit along with the reasoning behind such refinement process. Based on such reasoning, the central ML unit can iteratively further refine or validate the GLoc framework and provide the next refined version to the host entity.

[0074] As an example, a GLoc trained with UL SRS collected by TRP can be customized for static UEs by using the DL PRS observed at the static UE baseband as input. The refined model, called static-UE-GLoc, can be migrated back to the LMF, which then distributes it to other static UEs.

[0075] As another example, GLoc trained using UL SRS collected by TRP can be customized for high-speed UEs.

[0076] As another example, GLoc trained in an outdoor environment may be customized for an indoor environment to provide a refined model referred to as Indoor-GLoc.

[0077] As another example, GLoc trained on samples collected from urban scenes may be customized for suburban scenes to provide a refined model referred to as Suburban-GLoc.

[0078] Training at the NR-HU may be beneficial because the host (NR-HU) collects signals that are similarly distorted as other NR elements of the same type. The TL of the NR-HU can be based on the fact that the inherent signal distortion is inherent to the localization signal collected by the NR-HU, and the NR-HU uses a type-specific cost function to refine the GLoc, as well as type-specific model constraints (e.g., depth of artificial or simulated neural networks, available activation functions, etc.).

[0079] Alternatively or additionally, the central ML unit may request the NR-HU to collect, timestamp and migrate its training data to the central ML unit, and specify its model constraints (if any), in order to refine the GLoc at the central ML unit.

[0080] Figure 3 A signaling diagram according to an example embodiment is shown. Figure 3 , two types of network nodes (type x and type y) are shown, but it should be noted that the number of types may also be different from two. In other words, there may be one or more types of network nodes. In addition, Figure 3 The signaling process shown can be extended and applied according to the actual number of types. The central ML unit (eg, LMF) can determine the actual number of types according to the ability to group various network nodes based on their RF characteristics.

[0081] refer to Figure 3 In block 301, a central ML unit (such as, LMF) acquires a machine learning model for positioning, wherein the machine learning model is trained based on first training data associated with one or more first network nodes. The machine learning model may be trained for positioning, or for similar tasks that may also be used for positioning. The LMF may acquire the machine learning model by training the model at the LMF. Alternatively, the machine learning model may be trained at another entity from which the LMF may receive the machine learning model. The machine learning model may be referred to herein as GLoc. For example, the machine learning model may include an artificial neural network (ANN). Fig.10 An example of an artificial neural network is illustrated in .

[0082] The first training data may include at least one of the following: reference signal measurement information measured from one or more received positioning reference signals (e.g., DL PRS, UL SRS, and / or SL PRS) at one or more first network nodes, simulated reference signal measurement information, or simulated reference signal measurement information related to one or more first network nodes. For example, the reference signal measurement information may include at least a channel impulse response (CIR) measurement, which may be simulated or measured from one or more received positioning reference signals at one or more first network nodes. For example, the simulated reference signal measurement information may be obtained by using simulation tools such as ray tracing, digital twins, and the like.

[0083] The one or more first network nodes may include one or more types of network nodes. In one example, the one or more first network nodes may include a plurality of network nodes of different types.

[0084] In block 302, the LMF defines error ranges for type x and type y. The error ranges may indicate internal timing errors that occur during measurement collection due to RF defects or impairments. For example, a transmit timing error may indicate a time delay from the time a digital signal is generated at the baseband to the time an RF signal is transmitted from a transmit antenna. A receive timing error may indicate a time delay from the time an RF signal arrives at a receive antenna to the time the signal is digitized and time stamped at the baseband. Error ranges may be defined by the LMF in order to classify network nodes into type x and type y. For example, a type x host may be associated with an error range -x of + / - a ns (e.g., a=5 ns).

[0085] In block 303, the LMF selects a third network node of type x (denoted as type x NR-HU) that will perform transfer learning, i.e., update / refine the machine learning model (generic GLoc) to take into account RF defects specific to type x. The type of the third network node may be different from the type of one or more first network nodes.

[0086] Therefore, a single type x NR-HU can be selected as a representative for all type x network nodes, and this single type x NR-HU can perform model adaptation. This limits the computational complexity and signaling overhead of transfer learning by avoiding a scenario where each type x network node would perform model adaptation independently.

[0087] The LMF may also select a fourth network node of type y (denoted as type y NR-HU) which will update / refine the generic GLoc to account for RF deficiencies specific to type y. In this document, type y may refer to a type different from type x.

[0088] In this document, the term "network node" may mean, for example, a target user equipment, a positioning reference unit, an anchor user equipment, a TRP, or an access node (e.g., a gNB) of a radio access network.

[0089] In this document, the term "type" may mean, for example, a vendor-specific user equipment, a vendor-specific access node (e.g., gNB), a TRP with certain RF characteristics, a user equipment with a certain number of receive antennas, an Industrial Internet of Things (IIoT) device, a low-power high-precision positioning (LPHAP) device, a reduced-capability (RedCap) device, a handheld user equipment, or a roadside unit (RSU). For example, type N may be a UE with N receive antennas, where, for example, type 1 may be a UE with 1 receive antenna, type 2 may be a UE with 2 receive antennas, etc. The type may be defined with respect to both the target positioning accuracy and the inherent error range that a given network node of that type is expected to introduce.

[0090] In block 304, the LMF sends a request to the third network node (type x NR-HU) for updating the machine learning model at the third network node based on the second training data (i.e., a request for assisted transfer learning). For example, the request may be sent in an information element of an LTE Positioning Protocol (LPP) request message. The third network node may accept or reject the request from the LMF based on the load conditions and / or hardware limitations of the third network node.

[0091] The request may include information about the configuration of the machine learning model to be updated. For example, such information may define the following: the output and cost function of the machine learning model (i.e., the model function), the type, size, and shape of the input to the machine learning model, and the architecture of the machine learning model.

[0092] As a non-limiting example, the output of the machine learning model may be time of arrival (TOA) information to be used for positioning, and the cost function may be mean squared error (MSE).

[0093] As a non-limiting example, the input to the machine learning model may be a number of the strongest channel impulse response (CIR) complex gains.

[0094] As a non-limiting example, the architecture of the machine learning model may refer to a deep neural network (DNN) with N=10 hidden layers, and the activation function is a rectified linear unit (ReLU).

[0095] In block 305, the LMF sends a request to the fourth network node (type y NR-HU) for updating the machine learning model at the fourth network node based on the third training data. For example, the request may be sent in an information element of an LPP request message. The fourth network node may accept or reject the request from the LMF based on the load conditions and / or hardware limitations of the fourth network node.

[0096] The request may include information about the configuration of the machine learning model to be updated. For example, such information may define the following: the output and cost function of the machine learning model (i.e., the model function), the type, size, and shape of the input to the machine learning model, and the architecture of the machine learning model.

[0097] In other words, in blocks 304 and 305, the LMF requests configuration of different types of NR-HUs associated with processing a specific positioning request.

[0098] In block 306, the third network node (type x NR-HU) sends a response message to the LMF to accept the request. For example, acceptance may be indicated in an information element of the LPP reply message including a "yes or no" flag, or in a condition "yes" in which a model constraint is described. For example, the constraint may mean that the type x NR-HU may support a maximum architecture that is different from the architecture configured for GLoc. For example, the reply may include: "model_constraint = Model architecture: DNN with N = 4 hidden", to indicate that the type x NR-HU supports DNNs with a maximum of 4 hidden layers, while the GLoc architecture may include, for example, a DNN with 10 hidden layers.

[0099] In block 307, the LMF sends or transfers the machine learning model to the third network node in response to the third network node accepting the request (type x NR-HU). The sending may indicate at least one of the following: the structure of the machine learning model, one or more activation functions of the machine learning model, a set of weights for each layer of the machine learning model, a set of biases for each layer of the machine learning model, a cost function for training the machine learning model, an input type and format of the machine learning model (i.e., how the input is arranged and what it corresponds to), and / or an output type and format of the machine learning model (e.g., a probability vector or a binary vector, a vector length, etc.).

[0100] The input to the machine learning model may include at least one of the following: received signal samples for each receive beam for the total number of beams (some of the entries may be zero-filled if they are not available), reference signal received power (RSRP) for each positioning source and / or each beam, line of sight (LOS) indication or probability for each positioning source, etc. For example, the input may be a vector of CIRs with a certain length, and the entries are arranged in descending order of magnitude.

[0101] For example, the output may be time of arrival (TOA) information represented as a real-valued scalar.

[0102] The LMF may also send information indicating a set of constraints for updating the machine learning model at the third network node. In this way, the LMF may at least partially parameterize the transfer learning process at the third network node. As an example, the constraint set may indicate at least one of the following: using a reference signal from a selected pool of physical resource blocks (PRBs) to update the machine learning model for a given duration, and / or whether certain conditions are met. For example, the condition may be met if the third network node considers itself to be static, undisturbed, etc. Alternatively or additionally, the constraint set may indicate freezing a portion of the machine learning model and updating the remaining architecture, such as updating weights from layer L forward.

[0103] The LMF may also send information about a reference training process for updating the machine learning model at the third network node. As an example, the reference training process may correspond to using training data collected from multiple network nodes. Therefore, the information about the reference learning process may include at least a set of parameters (e.g., capabilities and RF defects) for constructing the reference training process, and the third network node may consider the parameter set when refining the reference training process based on its own capabilities and defects.

[0104] In block 308, the fourth network node (type y NR-HU) sends a response message to the LMF to accept the request. For example, acceptance may be indicated in an information element of the LPP reply message including a "yes or no" flag, or in a condition "yes" in which a model constraint is described.

[0105] In block 309, the LMF sends or transfers the machine learning model to the fourth network node in response to the fourth network node accepting the request (type y NR-HU). The LMF may also send information indicating a set of constraints for updating the machine learning model at the fourth network node. The LMF may also send information about a reference training process for updating the machine learning model at the fourth network node, wherein the information about the reference training process includes at least a set of parameters for constructing the reference training process.

[0106] The content of the transmission of block 309 may be similar to that described above for block 307 .

[0107] In block 310, the third network node (type x NR-HU) obtains a first updated machine learning model by updating the machine learning model (i.e., the original GLoc) based on the second training data. Updating may mean adjusting, refining, or retraining the machine learning model based on the second training data specific to the third network node of type x. The third network node may also verify the first updated machine learning model. The first updated machine learning model obtained by the third network node may be referred to herein as type-x-GLoc.

[0108] The second training data may include reference signal measurement information measured at a third network node from one or more received positioning reference signals (e.g., DL PRS, UL SRS, and / or SL PRS), wherein the third network node may be different from the one or more first network nodes. For example, the reference signal measurement information may include at least a channel impulse response (CIR) measurement measured at the third network node from one or more received positioning reference signals.

[0109] The update of the machine learning model can include the following steps 1-6: 1) Initialize the machine learning model using the machine learning model received from the LMF. Initialization may mean pruning or otherwise simplifying the machine learning model according to the capabilities of the third network node. 2) Acquire (eg, collect or access) second training data. 3) Clean the second training data and prepare the input to match the format defined by LMF. 4) Selecting a cost function specific to the third network node, for example by using a variation of a generic cost function indicated by the LMF, or a cost function from a list of cost functions selected by the LMF. 5) Select the activation function(s) provided by LMF to produce the output format defined by LMF. 6) Perform update / training under the set of constraints indicated by the LMF.

[0110] In block 311, the third network node (type x NR-HU) sends a message to the LMF indicating a first updated machine learning model (type-x-GLoc) acquired by the third network node. The message may include the first updated machine learning model. Alternatively, the message may include an updated set of weights and biases associated with the first updated machine learning model. In other words, the refined process may be reported as a "delta" to the provided reference process, such that only the weights and biases that have been updated may be reported (i.e., the full model is not reported).

[0111] A message may be sent based on the estimated performance improvement of the updated machine learning model being above a threshold. In other words, if when tested, it produces a performance improvement (compared to the original GLoc) greater than a given threshold, then the refined model may be reported. The threshold may be defined based on positioning accuracy and measurement granularity. This may reduce unnecessary reporting and therefore reduce network signaling.

[0112] In block 312, the fourth network node (type y NR-HU) obtains a second updated machine learning model by updating the machine learning model (i.e., the original GLoc) based on the third training data. The fourth network node may also verify the second updated machine learning model. The second updated machine learning model obtained by the fourth network node may be referred to as type-y-GLoc. The fourth network node may perform updates similar to those described above for block 310.

[0113] The third training data may include reference signal measurement information measured from one or more received positioning reference signals (e.g., DL PRS, UL SRS, and / or SL PRS) at a fourth network node, wherein the fourth network node may be different from the one or more first network nodes and the third network node. For example, the reference signal measurement information may include at least a channel impulse response (CIR) measurement measured from one or more received positioning reference signals at the fourth network node.

[0114] In block 313, the fourth network node (type y NR-HU) sends a message to the LMF indicating a second updated machine learning model (type-y-GLoc) acquired by the fourth network node. The message may include the second updated machine learning model. Alternatively, the message may include an updated set of weights and biases associated with the second updated machine learning model.

[0115] In block 314, the LMF may validate or modify the first updated machine learning model and the second updated machine learning model. For example, before large-scale distribution, the LMF may cross-validate the updated model to ensure that it remains robust and performs within target key performance indicators (KPIs). For example, the LMF may use stored test data to check whether the updated model meets a given KPI target.

[0116] In box 315, the LMF sends or distributes the first updated machine learning model (type-x-GLoc) to one or more second network nodes of type x. In other words, the LMF distributes the first updated machine learning model to other entities of the same type (type x NR-HU) as the third network node. One or more second network nodes may be referred to herein as type x units. The difference between type x NR-HU and type x unit is that type x NR-HU has the ability to train an updated machine learning model, i.e., has the ability and computing resources to collect and label second training data, and is therefore designated to propose an updated machine learning model that works well on all type x units.

[0117] A given second network node (Type-x unit) may be configured to use a first updated machine learning model (Type-x-GLoc) for positioning a target UE. For example, if the Type-x unit is a gNB or an anchor UE, the Type-x unit may measure a reference signal received from the target UE to obtain, for example, a CIR measurement. Alternatively, if the Type-x unit is a target UE, the Type-x unit may measure a reference signal received from the gNB or the anchor UE to obtain, for example, a CIR measurement. The Type-x unit may then provide these measurements as input to the Type-x-GLoc. The output of the Type-x-GLoc may be a position estimate of the target UE. Alternatively, the output of the Type-x-GLoc may be some other useful positioning related information or intermediate features, such as the time of arrival (TOA) and / or angle of arrival (AOA) of a (possible) line-of-sight (LOS) path and / or a strong non-line-of-sight (NLOS) path.

[0118] Type-x-GLoc can be fed with the same input information and produces the same output type as the original GLoc. The difference lies in the accuracy of these models, where the type-x-GLoc model can provide higher accuracy for a specific type of device (i.e., for type x units) compared to the original GLoc because the type-x-GLoc model is refined based on the specific RF imperfections of that device type.

[0119] It should be noted that LMF can also send the first updated machine learning model to other types of network nodes, for example, to one or more first network nodes.

[0120] In block 316, the LMF sends or distributes the type y-GLoc to one or more fifth network nodes of type y. In other words, the LMF distributes the second updated machine learning model to other entities of the same type (type y NR-HU) as the fourth network node.

[0121] It should be noted that the LMF may also send the second updated machine learning model to other types of network nodes, such as to one or more first network nodes.

[0122] In this document, the terms "first network node", "second network node" and the like are used to distinguish network nodes, and they do not necessarily mean specific identifiers of the network nodes.

[0123] Figure 4 A signaling diagram according to another example embodiment is shown. In this example embodiment, the selected NR-HU collects training data and transfers it back to the LMF. The LMF can then aggregate data from multiple network nodes of the same type, store it in memory and generate an updated machine learning model. It should be noted that the LMF can also use the collected training data and combine it across different network node types for updating the general GLoc model.

[0124] Despite Figure 4 , two types of network nodes (type x and type y) are shown, but it should be noted that the number of types may also be different from two. In other words, there may be one or more types of network nodes. In addition, Figure 4 The signaling process shown can be extended and applied depending on the actual number of types. The central ML unit (eg, LMF) can determine the actual number of types depending on the ability to group the various network nodes based on their RF characteristics.

[0125] refer to Figure 4 In block 401, a central ML unit (e.g., LMF) obtains a machine learning model for positioning, wherein the machine learning model is trained based on first training data associated with one or more first network nodes. The machine learning model may be trained for positioning, or for similar tasks that may also be used for positioning. The LMF may acquire the machine learning model by training the model at the LMF. Alternatively, the machine learning model may be trained at another entity from which the LMF may receive the machine learning model. The machine learning model may be referred to herein as GLoc. For example, the machine learning model may include an artificial neural network (ANN). Fig.10 An example of an artificial neural network is shown in .

[0126] The first training data may include at least one of the following: reference signal measurement information measured from one or more received positioning reference signals (e.g., DL PRS, UL SRS, and / or SL PRS) at one or more first network nodes, simulated reference signal measurement information, or simulated reference signal measurement information related to one or more first network nodes. For example, the reference signal measurement information may include at least a channel impulse response (CIR) measurement, which may be simulated or measured from one or more received positioning reference signals at one or more first network nodes. For example, the simulated reference signal measurement information may be obtained by using simulation tools such as ray tracing, digital twins, and the like.

[0127] The one or more first network nodes may include one or more types of network nodes. In one example, the one or more first network nodes may include a plurality of network nodes of different types.

[0128] In block 402, the LMF defines error ranges for type x and type y. The error ranges may indicate internal timing errors that occur during measurement collection due to RF defects or impairments. For example, a transmit timing error may indicate a time delay from the time a digital signal is generated at the baseband to the time an RF signal is transmitted from a transmit antenna. A receive timing error may indicate a time delay from the time an RF signal arrives at a receive antenna to the time the signal is digitized and time stamped at the baseband. Error ranges may be defined by the LMF in order to classify network nodes into type x and type y. For example, a type x host may be associated with an error range x of + / - a ns (e.g., a=5ns).

[0129] In block 403, the LMF selects one or more third network nodes of type x (denoted as type x NR-HU) from which the LMF will request training data to update / refine the machine learning model (generic GLoc) to take into account RF defects specific to type x. The type of the one or more third network nodes may be different from the type of the one or more first network nodes.

[0130] The LMF may also select one or more fourth network nodes of type y (denoted as type y NR-HU), from which the LMF will request training data to update / refine the machine learning model (generic GLoc) to take into account RF defects specific to type y. In this document, type y may refer to a type different from type x.

[0131] In this document, the term “network node” may mean, for example, a target user equipment, a positioning reference unit, an anchor UE, a TRP, or an access node (e.g., a gNB) of a radio access network.

[0132] In this document, the term "type" may mean, for example, a vendor-specific user equipment, a vendor-specific access node (e.g., gNB), a TRP with certain RF characteristics, a user equipment with a certain number of receive antennas, an Industrial Internet of Things (IIoT) device, a Low Power High Accuracy Positioning (LPHAP) device, a Reduced Capability (RedCap) device, a handheld user device, or a roadside unit (RSU). The type may be defined with respect to both the target positioning accuracy and the inherent error range that a given network node of that type is expected to introduce.

[0133] In block 404, the LMF sends a request to one or more third network nodes (type x NR-HU) to provide second training data from one or more third network nodes to the LMF for updating the machine learning model at the LMF. For example, the request may be sent in an information element of an LPP request message. A given third network node may accept or reject the request from the LMF based on the load conditions and / or hardware limitations of the third network node.

[0134] In block 405, the LMF sends a request to one or more fourth network nodes (type y NR-HU) to provide third training data from one or more fourth network nodes to the LMF for updating the machine learning model at the LMF. For example, the request may be sent in an information element of an LPP request message. A given fourth network node may accept or reject the request from the LMF based on the load conditions and / or hardware limitations of the fourth network node.

[0135] In block 406, one or more third network nodes (type x NR-HU) send a response message to the LMF to accept the request. For example, the acceptance may be indicated in an information element of the LPP reply message including a "yes or no" flag.

[0136] In block 407, one or more fourth network nodes (type y NR-HU) send a response message to the LMF to accept the request. For example, acceptance may be indicated in an information element of the LPP reply message including a "yes or no" flag.

[0137] In block 408, one or more third network nodes (type x NR-HU) obtain second training data. For example, the one or more third network nodes may obtain the second training data by performing measurements (such as CIR measurements) on received positioning reference signals (such as DL PRS, UL SRS, or SL PRS). Alternatively, the one or more third network nodes may obtain the second training data by retrieving stored measurements from a memory. The one or more third network nodes may also clear the second training data before sending the second training data to the LMF.

[0138] In block 409, one or more third network nodes (type x NR-HU) send the second training data to the LMF. The LMF may store the received second training data in its memory. The one or more third network nodes may also send information indicating a first set of constraints for updating the machine learning model at the LMF. For example, the first set of constraints may include a maximum supported depth (i.e., number of layers) of the artificial neural network, an available activation function, etc.

[0139] In block 410, one or more fourth network nodes (type y NR-HU) obtain third training data. For example, the one or more fourth network nodes may obtain the third training data by performing measurements (such as CIR measurements) on received positioning reference signals (such as DL PRS, UL SRS, or SL PRS). Alternatively, the one or more fourth network nodes may obtain the third training data by retrieving stored measurements from a memory. The one or more fourth network nodes may also clear the third training data before sending the third training data to the LMF.

[0140] In block 411, one or more fourth network nodes (type y NR-HU) send the third training data to the LMF. The LMF may store the received third training data in its memory. The one or more fourth network nodes may also send information indicating a second set of constraints for updating the machine learning model at the LMF. For example, the second set of constraints may include a maximum supported depth (i.e., number of layers) of the artificial neural network, available activation functions, etc.

[0141] In box 412, the LMF obtains a first updated machine learning model (type-x-GLoc) by updating the machine learning model (original GLoc) based at least in part on second training data received from one or more third network nodes.

[0142] The update may include one or more of the following steps 1-6: 1) Initialize the machine learning model using the original GLoc. 2) Access the second training data. 3) Clean the second training data and prepare the input to match the input format of GLoc. 4) Selecting a cost function specific to one or more third network nodes, for example by using a variation of the generic cost function of GLoc. 5) Select activation function(s) to produce the output format of GLoc. 6) Performing updating / training under a set of constraints indicated by one or more third network nodes.

[0143] The LMF may also obtain a second updated machine learning model (type-y-GLoc) by updating the machine learning model (original GLoc) based at least in part on third training data received from one or more fourth network nodes.

[0144] In box 413, the LMF sends or distributes the first updated machine learning model (type-x-GLoc) to one or more second network nodes of type x. In other words, the LMF distributes the first updated machine learning model to other entities of the same type (type x NR-HU) as one or more third network nodes. The one or more second network nodes may be referred to herein as type x units. The difference between type x NR-HU and type x unit is that type x NR-HU has the ability to train an updated machine learning model, i.e., has the ability and computing resources to collect and label second training data, and is therefore designated to propose an updated machine learning model that works well on all type x units.

[0145] A given second network node (type-x unit) may be configured to use the first updated machine learning model (type-x-GLoc) for positioning a target UE. For example, if the type-x unit is a gNB or an anchor UE, the type-x unit may measure a reference signal received from the target UE to obtain, for example, a CIR measurement. Alternatively, if the type-x unit is a target UE, the type-x unit may measure a reference signal received from the gNB or the anchor UE to obtain, for example, a CIR measurement. The type-x unit may then provide these measurements as input to the type-x-GLoc. The output of the type-x-GLoc may be a position estimate of the target UE. Alternatively, the output of the type-x-GLoc may be some other useful positioning-related information or intermediate features, such as the time of arrival (TOA) and / or angle of arrival (AOA) of a (possible) line-of-sight (LOS) path and / or a strong non-line-of-sight (NLOS) path.

[0146] Type-x-GLoc can be fed with the same input information and produces the same output type as the original GLoc. The difference lies in the accuracy of these models, where the type-x-GLoc model provides higher accuracy for a specific type of device (i.e., for type x devices) compared to the original GLoc because the type-x-GLoc model is refined based on the specific RF imperfections of that device type.

[0147] It should be noted that LMF can also send the first updated machine learning model to other types of network nodes, for example, to one or more first network nodes.

[0148] In block 414, the LMF sends or distributes the type-y-GLoc to one or more fifth network nodes of type y. In other words, the LMF distributes the second updated machine learning model to other entities of the same type (type y NR-HU) as the fourth network node. The one or more fifth network nodes may be referred to herein as type y units.

[0149] It should be noted that the LMF may also send the second updated machine learning model to other types of network nodes, such as to one or more first network nodes.

[0150] Figure 5 A flow chart according to an example embodiment of a method performed by an apparatus, such as a location management function (LMF) or a network data analysis function (NWDAF), or including or included in an LMF or NWDAF, is shown. For example, the apparatus may correspond to Figure 1 LMF 112 or Figure 2 LMF 212.

[0151] refer to Figure 5 In box 501, a machine learning model for positioning is obtained, wherein the machine learning model is trained based on first training data related to one or more first network nodes.

[0152] In box 502, information including at least one of the following is sent: a machine learning model, a request for updating the machine learning model based on second training data, or a request for providing second training data for updating the machine learning model at a device.

[0153] In box 503, at least one of the following is received: a message indicating an updated machine learning model or second training data.

[0154] In block 504, the updated machine learning model is sent to at least one or more second network nodes.

[0155] Figure 6 A flow chart according to an example embodiment of a method performed by an apparatus is shown, the apparatus being, for example, a network node, or comprising, or being comprised in, a network node. A network node may refer to, for example, a user equipment, a positioning reference unit, or an access node of a radio access network. For example, a network node may correspond to Figure 1 UE 100, UE 102 or access node 104, or corresponding to Figure 2 UE 200, any one of PRUs 202, 202A, 202B or any one of access nodes 204, 204A, 204B.

[0156] refer to Figure 6, in box 601, information is received including at least one of the following: a machine learning model for positioning, a request for updating the machine learning model at a device based on second training data, or a request for providing second training data for updating the machine learning model, wherein the machine learning model has been trained based on first training data associated with one or more first network nodes.

[0157] Step 602: Obtain second training data.

[0158] In box 603, at least one of the following is sent: a message indicating an updated machine learning model or second training data.

[0159] As used herein, “at least one of: ” and “at least one of ” and similar expressions, where a list of two or more elements is combined by “and” or “or”, mean at least any one of the elements, or at least any two or more of the elements, or at least all of the elements.

[0160] Above through Figures 3 to 6 The described boxes, related functions and information exchanges (messages) are not in absolute time order, and some of them can be executed simultaneously or in an order different from the described order. Other functions can also be performed between them or within them, and other information can be sent and / or other rules can be applied. Some boxes or parts of boxes or one or more pieces of information can also be omitted or replaced by corresponding boxes or parts of boxes or one or more pieces of information.

[0161] Figure 7 Shown includes for execution Figures 3 to 6 An example of an apparatus 700 of any of the methods of or components of any other example embodiments described above. For example, the apparatus 700 may be an apparatus such as a user device or comprising a user device or included in a user device. The user device may correspond to Figure 1 Any user device 100, 102, or corresponding to Figure 2 A user device 200, or a user device corresponding to Figure 2 A user equipment may also be referred to as a subscriber unit, a mobile station, a remote terminal, an access terminal, a user terminal, a terminal device, a user equipment (UE), a target UE, a target user equipment, an anchor UE, a positioning reference unit (PRU), a NR-HU, a host, or a network node.

[0162] The apparatus 700 includes at least one processor 710. The at least one processor 710 interprets instructions (e.g., computer program instructions) and processes data. The at least one processor 710 may include one or more programmable processors. The at least one processor 710 may include programmable hardware with embedded firmware, and may alternatively or additionally include one or more application specific integrated circuits (ASICs).

[0163] At least one processor 710 is coupled to at least one memory 720. At least one processor is configured to read data from at least one memory 720 and write data to at least one memory 720. At least one memory 720 may include one or more memory cells. The memory cell may be volatile or non-volatile. It should be noted that there may be one or more non-volatile memory cells and one or more volatile memory cells, or alternatively, one or more non-volatile memory cells, or alternatively, one or more volatile memory cells. The volatile memory may be, for example, a random access memory (RAM), a dynamic random access memory (DRAM), or a synchronous dynamic random access memory (SDRAM). The non-volatile memory may be, for example, a read-only memory (ROM), a programmable read-only memory (PROM), an electrically erasable programmable read-only memory (EEPROM), a flash memory, an optical storage device, or a magnetic storage device. In general, the memory may be referred to as a non-transitory computer-readable medium. The term "non-transitory" as used herein is a limitation of the medium itself (i.e., tangible, not a signal), rather than a limitation of data storage persistence (e.g., RAM to ROM). At least one memory 720 stores one or more computer readable instructions executed by at least one processor 710 to perform one or more of the above-described example embodiments. For example, non-volatile memory stores computer readable instructions, and at least one processor 710 executes instructions using volatile memory to temporarily store data and / or instructions. Computer readable instructions may refer to computer program code.

[0164] The computer readable instructions may be pre-stored to at least one memory 720, or alternatively or additionally, they may be received by the device via an electromagnetic carrier signal and / or may be copied from a physical entity such as a computer program product. The execution of the computer readable instructions by at least one processor 710 causes the device 700 to perform one or more of the above-described example embodiments. That is, at least one processor and at least one memory storing instructions may provide a means for providing or causing the execution of any of the above-described methods and / or blocks.

[0165] In the context of this document, "memory" or "computer-readable medium" or "computer-readable medium" can be any non-transitory medium or media or component that can contain, store, communicate, propagate or transport instructions for use by or in connection with an instruction execution system, apparatus or device (such as a computer). The term "non-transitory" as used herein is a limitation on the medium itself (i.e., tangible, not a signal), not on data storage persistence (e.g., RAM versus ROM).

[0166] The device 700 may also include or be connected to an input unit 730. The input unit 730 may include one or more interfaces for receiving input. The one or more interfaces may include, for example, one or more temperature, motion and / or orientation sensors, one or more cameras, one or more accelerometers, one or more microphones, one or more buttons and / or one or more touch detection units. In addition, the input unit 730 may include an interface to which an external device may be connected.

[0167] The device 700 may also include an output unit 740. The output unit may include or be connected to one or more displays capable of presenting visual content, such as a light emitting diode (LED) display, a liquid crystal display (LCD), and / or a liquid crystal on silicon (LCoS) display. The output unit 740 may also include one or more audio outputs. The one or more audio outputs may be, for example, a speaker.

[0168] The device 700 also includes a connection unit 750. The connection unit 750 implements a wireless connection to one or more external devices. The connection unit 750 includes at least one transmitter and at least one receiver that can be integrated into the device 700 or the device 700 can be connected to. At least one transmitter includes at least one transmitting antenna, and at least one receiver includes at least one receiving antenna. The connection unit 750 may include an integrated circuit or a group of integrated circuits that provide wireless communication capabilities for the device 700. Alternatively, the wireless connection can be a hardwired application-specific integrated circuit (ASIC). The connection unit 750 can also provide components for performing at least some of the blocks of one or more of the example embodiments described above. The connection unit 750 may include one or more components controlled by a corresponding control unit, such as: a power amplifier, a digital front end (DFE), an analog-to-digital converter (ADC), a digital-to-analog converter (DAC), a frequency converter, a (demodulation) modulator, and / or an encoder / decoder circuit.

[0169] It should be noted that the device 700 may also include Figure 7 Various components not described in the specification. The various components may be hardware components and / or software components.

[0170] Figure 8 Shown includes for execution Figures 3 to 6An example of an apparatus 800 of any of the methods of or components of any other example embodiments described above. For example, the apparatus 800 may be an access node such as a radio access network, or an access node included in a radio access network, or an apparatus included in an access node of a radio access network. The access node may correspond to Figure 1 The access node 104, or corresponding to Figure 2 The apparatus 800 may also be referred to as, for example, a network node, a radio access network (RAN) node, a next generation radio access network (NG-RAN) node, a NodeB, an eNB, a gNB, a base transceiver station (BTS), a base station, an NR base station, a 5G base station, an access point (AP), a relay node, a repeater, an integrated access and backhaul (IAB) node, an IAB donor node, a distributed unit (DU), a central unit (CU), a baseband unit (BBU), a radio unit (RU), a radio head, a remote radio head (RRH), or a transceiver point (TRP).

[0171] The device 800 may include, for example, a circuit or chipset suitable for implementing one or more of the above-described example embodiments. The device 800 may be an electronic device including one or more electronic circuits. The device 800 may include a communication control circuit 810, such as at least one processor, and at least one memory 820 storing instructions, which, when executed by at least one processor, causes the device 800 to perform one or more of the above-described example embodiments. Such instructions 822 may, for example, include computer program code (software), wherein at least one memory and the computer program code (software) are configured to cause the device 800 to perform one or more of the above-described example embodiments together with at least one processor. In this article, computer program code may also refer to instructions, which, when executed by at least one processor, causes the device 800 to perform one or more of the above-described example embodiments. That is, at least one processor and at least one memory storing instructions may provide a component for providing or causing the execution of any of the above-described methods and / or boxes.

[0172] The processor is coupled to the memory 820. The processor is configured to read data from the memory 820 and write data to the memory 820. The memory 820 may include one or more memory cells. The memory cell may be volatile or non-volatile. It should be noted that there may be one or more non-volatile memory cells and one or more volatile memory cells, or alternatively, one or more non-volatile memory cells, or alternatively, one or more volatile memory cells. The volatile memory may be, for example, a random access memory (RAM), a dynamic random access memory (DRAM), or a synchronous dynamic random access memory (SDRAM). The non-volatile memory may be, for example, a read-only memory (ROM), a programmable read-only memory (PROM), an electrically erasable programmable read-only memory (EEPROM), a flash memory, an optical storage device, or a magnetic storage device. In general, the memory may be referred to as a non-transitory computer-readable medium. The term "non-transitory" as used herein is a limitation of the medium itself (i.e., tangible, not a signal), rather than a limitation of data storage persistence (e.g., RAM to ROM). The memory 820 stores computer-readable instructions executed by the processor. For example, non-volatile memory stores computer readable instructions, and the processor executes the instructions using volatile memory to temporarily store data and / or instructions.

[0173] The computer readable instructions may be pre-stored to the memory 820, or alternatively or additionally, they may be received by the device via an electromagnetic carrier signal and / or may be copied from a physical entity such as a computer program product. Execution of the computer readable instructions causes the device 800 to perform one or more of the above functions.

[0174] The memory 820 may be implemented using any suitable data storage technology, such as semiconductor-based memory devices, flash memory, magnetic memory devices and systems, optical memory devices and systems, fixed memory, and / or removable memory. The memory may include a configuration database for storing configuration data. For example, the configuration database may store a current neighbor cell list and, in some example embodiments, a structure of frames used in detected neighbor cells.

[0175] The device 800 may also include a communication interface 830, which includes hardware and / or software for implementing a communication connection according to one or more communication protocols. The communication interface 830 includes at least one transmitter (Tx) and at least one receiver (Rx) that can be integrated into the device 800 or to which the device 800 can be connected. The communication interface 830 may provide components for performing some of the blocks of one or more of the example embodiments described above. The communication interface 830 may include one or more components controlled by a corresponding control unit, such as: a power amplifier, a digital front end (DFE), an analog-to-digital converter (ADC), a digital-to-analog converter (DAC), a frequency converter, a (demodulation) modulator, and / or an encoder / decoder circuit.

[0176] The communication interface 830 provides the device with radio communication capabilities to communicate in a cellular communication system. The communication interface may, for example, provide a radio interface to one or more user equipments. The device 800 may also include another interface towards a core network (such as a network coordinator device or AMF) and / or an access node of a cellular communication system.

[0177] The apparatus 800 may further include a scheduler 840 configured to allocate radio resources. The scheduler 840 may be configured together with the communication control circuit 810, or it may be configured separately.

[0178] It should be noted that the device 800 may also include Figure 8 Various components are not shown in the figure. The various components can be hardware components and / or software components.

[0179] Fig. 9 Shown includes for execution Figures 3 to 6 An example of an apparatus 900 that may include any of the methods of or components of any other example embodiments described above. For example, the apparatus 900 may be an apparatus such as a central ML unit, or include a central ML unit, or be included in a central ML unit. The central ML unit may also be referred to as, for example, a location management function (LMF), a location server, or a network data analysis function (NWDAF). For example, the central ML unit may correspond to Figure 1 LMF 112, or the equivalent Figure 2 LMF 212.

[0180] The device 900 may include, for example, a circuit or chipset suitable for implementing one or more of the above-described example embodiments. The device 900 may be an electronic device including one or more electronic circuits. The device 900 may include a communication control circuit 910 (such as at least one processor) and at least one memory 920 storing instructions, which, when executed by at least one processor, causes the device 900 to perform one or more of the above-described example embodiments. Such instructions 922 may, for example, include computer program code (software), wherein at least one memory and computer program code (software) are configured to cause the device 900 to perform one or more of the above-described example embodiments together with at least one processor. In this article, computer program code may also refer to instructions, which, when executed by at least one processor, causes the device 900 to perform one or more of the above-described example embodiments. That is, at least one processor and at least one memory storing instructions may provide a component for providing or causing the execution of any of the above-described methods and / or boxes.

[0181] The processor is coupled to the memory 920. The processor is configured to read data from the memory 920 and write data to the memory 920. The memory 920 may include one or more memory cells. The memory cell may be volatile or non-volatile. It should be noted that there may be one or more non-volatile memory cells and one or more volatile memory cells, or alternatively, one or more non-volatile memory cells, or alternatively, one or more volatile memory cells. The volatile memory may be, for example, a random access memory (RAM), a dynamic random access memory (DRAM), or a synchronous dynamic random access memory (SDRAM). The non-volatile memory may be, for example, a read-only memory (ROM), a programmable read-only memory (PROM), an electrically erasable programmable read-only memory (EEPROM), a flash memory, an optical storage device, or a magnetic storage device. In general, the memory may be referred to as a non-transitory computer-readable medium. The term "non-transitory" as used herein is a limitation of the medium itself (i.e., tangible, not a signal), rather than a limitation of data storage persistence (e.g., RAM to ROM). The memory 920 stores computer-readable instructions executed by the processor. For example, non-volatile memory stores computer readable instructions, and the processor executes the instructions using volatile memory to temporarily store data and / or instructions.

[0182] The computer readable instructions may be pre-stored in the memory 920, or alternatively or additionally, they may be received by the device via an electromagnetic carrier signal and / or may be copied from a physical entity such as a computer program product. Execution of the computer readable instructions causes the device 900 to perform one or more of the above functions.

[0183] The memory 920 may be implemented using any suitable data storage technology, such as semiconductor-based memory devices, flash memory, magnetic memory devices and systems, optical memory devices and systems, fixed memory, and / or removable memory. The memory may include a configuration database for storing configuration data. For example, the configuration database may store a current neighbor cell list and, in some example embodiments, a structure of frames used in detected neighbor cells.

[0184] The device 900 may also include a communication interface 930, which includes hardware and / or software for implementing a communication connection according to one or more communication protocols. The communication interface 930 includes at least one transmitter (Tx) and at least one receiver (Rx) that can be integrated into the device 900 or the device 900 can be connected to. The communication interface 930 may provide components for performing some of the blocks of one or more of the example embodiments described above. The communication interface 930 may include one or more components controlled by a corresponding control unit, such as: a power amplifier, a digital front end (DFE), an analog-to-digital converter (ADC), a digital-to-analog converter (DAC), a frequency converter, a (demodulation) modulator, and / or an encoder / decoder circuit.

[0185] The communication interface 930 provides the device with radio communication capabilities to communicate in a cellular communication system. The communication interface may, for example, provide a radio interface to one or more user equipments. The device 900 may also include another interface towards a core network (such as a network coordinator device or AMF) and / or an access node of a cellular communication system.

[0186] It should be noted that the device 900 may also include Fig. 9 Various components are not shown in the figure. The various components can be hardware components and / or software components.

[0187] As used in this application, the term "circuitry" may refer to one or more or all of the following: a) a hardware circuit implementation only (such as an implementation in analog and / or digital circuitry only); and b) a combination of hardware circuitry and software, such as (where applicable): i) a combination of (multiple) analog and / or digital hardware circuits and software / firmware, and (ii) any portion of a (multiple) hardware processor with software (including (multiple) digital signal processors, software and (multiple) memories that work together to enable a device (such as a mobile phone) to perform various functions) and (c) (multiple) hardware circuits and / or (multiple) processors, such as (multiple) microprocessors or portions of (multiple) microprocessors, which require software (e.g. firmware) to operate, but which may not be present when the software is not required to operate.

[0188] This definition of circuitry applies to all uses of this term in this application, including in any claims. As a further example, as used in this application, the term circuitry also covers an implementation of only a hardware circuit or processor (or multiple processors) or a portion of a hardware circuit or processor and its (or their) accompanying software and / or firmware. The term circuitry also covers, for example and if applicable to a particular claim element, a baseband integrated circuit or processor integrated circuit for a mobile device or a similar integrated circuit in a server, cellular network device, or other computing or network device.

[0189] Fig.10 An example of an artificial neural network 1030 having one hidden layer 1002 is shown, and Fig.11 An example of a computational node 1004 is shown. However, it should be noted that the artificial neural network 1030 may also include more than one hidden layer 1002.

[0190] Artificial neural network (ANN) 1030 includes a set of rules designed to perform tasks such as regression, classification, clustering, and pattern recognition. ANN can utilize a learning / training process to achieve such a goal, in which various examples of input data and desired outputs are shown. In this way, ANN learns to identify the appropriate output for any input within the training data manifold. Learning / training by using labels is called supervised learning, and learning without labels is called unsupervised learning. Deep learning may require a large amount of input data.

[0191] Deep learning (also known as deep structured learning or hierarchical learning) is part of a broader family of machine learning methods based on layers used in artificial neural networks. A deep neural network (DNN) 1030 is an artificial neural network that includes multiple hidden layers 1002 between an input layer 1000 and an output layer 1014. Training of a DNN allows it to find the correct mathematical manipulations to transform inputs into appropriate outputs, even when the relationship is highly nonlinear and / or complex.

[0192] A given hidden layer 1002 includes nodes 1004, 1006, 1008, 1010, 1012 where computation occurs. Fig.11 As shown, a given node 1004 combines input data 1000 with a set of coefficients or weights 1100 that amplify or suppress that input 1000, thereby assigning importance to the input 1000 with respect to the task the algorithm is trying to learn. The input weight products are summed 1102, and the sum is passed through an activation function 1104 to determine whether and to what extent the signal should proceed further through the neural network 1030 to affect a final outcome, such as a classification action. In this process, the neural network learns to identify the correlation between certain relevant features and optimal outcomes.

[0193] In the case of classification, the output of the DNN 1030 can be considered as the probability of a particular result. In this case, the number of layers 1002 can vary in proportion to the number of input data 1000 used. However, when the number of input data 1000 is high, the accuracy of the result 1014 is more reliable. On the other hand, when there are fewer layers 1002, the calculation may take less time, thereby reducing latency. However, this is highly dependent on the specific DNN architecture and / or available computing resources.

[0194] The initial weights 1100 of the model can be set in various alternative ways. During the training phase, they can be adapted to improve the accuracy of the process based on the errors in the analytical decisions. Training the model is essentially a trial and error activity. In principle, a given node 1004, 1006, 1008, 1010, 1012 of the neural network 1030 makes a decision (input * weight), and then compares the decision with the collected data to find the difference with the collected data. In other words, it determines the error, based on which the weights 1100 are adjusted. Therefore, the training of the model can be considered as a corrective feedback loop.

[0195] For example, a stochastic gradient descent optimization algorithm may be used to train a neural network model, where the gradient is calculated using a back-propagation algorithm. The gradient descent algorithm attempts to change the weights 1100 so that the next evaluation reduces the error, which means that the optimization algorithm navigates down the gradient (or slope) of the error. Any other suitable optimization algorithm may also be used if it provides sufficiently accurate weights 1100. Thus, the trained parameters of the neural network 1030 may include the weights 1100.

[0196] In the context of an optimization algorithm, the function used to evaluate a candidate solution (i.e., a set of weights) is called an objective function. For a neural network where the goal is to minimize error, the objective function may be called a cost function or a loss function. When adjusting weights 1100, any suitable method may be used as a loss function. Some examples of loss functions are mean square error (MSE), maximum likelihood estimation (MLE), and cross entropy.

[0197] An activation function 1104 for a node 1004 defines the output 1014 of the node 1004 given an input or set of inputs 1000. The node 1004 computes a weighted sum of the inputs, possibly with a bias, and then makes a decision as "activate" or "do not activate" based on a decision threshold as a binary activation or using an activation function 1104 that gives a nonlinear decision function. Any suitable activation function 1104 can be used, such as sigmoid, rectified linear unit (ReLU), normalized exponential function (softmax), sloffplus, tanh, etc. In deep learning, the activation function 1104 can be set at the layer level and applied to all neurons (nodes) in the layer. The output 1014 is then used as the input for the next node, and so on, until the desired solution to the original problem is found.

[0198] The techniques and methods described herein can be implemented in various ways. For example, these techniques can be implemented in hardware (one or more devices), firmware (one or more devices), software (one or more modules), or a combination thereof. For hardware implementations, the (multiple) devices of the example embodiments can be implemented in one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), graphics processing units (GPUs), processors, controllers, microcontrollers, microprocessors, other electronic units designed to perform the functions described herein, or a combination thereof. For firmware or software, the implementation can be performed by a module of at least one chipset (e.g., a process, a function, etc.) that performs the functions described herein. The software code can be stored in a memory unit and executed by a processor. The memory unit can be implemented within a processor or outside a processor. In the latter case, it can be communicatively coupled to the processor via various components known in the art. Additionally, the components of the system described herein can be rearranged and / or supplemented by additional components to facilitate the implementation of various aspects of its description, etc., and they are not limited to the precise configurations set forth in the given figures, as will be understood by those skilled in the art.

[0199] It is obvious to those skilled in the art that, as technology advances, the inventive concept can be implemented in various ways. The embodiments are not limited to the above-described example embodiments, but can be varied within the scope of the claims. Therefore, all words and expressions should be interpreted broadly, and they are intended to illustrate rather than limit the example embodiments.

Claims

1. An apparatus comprising at least one processor and at least one memory, the at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus to at least: Acquire a machine learning model for positioning, wherein the machine learning model is trained based on first training data related to one or more first network nodes; sending information including at least one of: the machine learning model, a request to update the machine learning model based on second training data, or a request to provide the second training data for updating the machine learning model at the device; receiving at least one of: a message indicating an updated machine learning model, or the second training data; as well as The updated machine learning model is sent to at least one or more second network nodes.

2. The apparatus according to claim 1, further caused to: selecting a third network node, wherein the information is sent to the third network node, and The type of the third network node is different from the type of the one or more first network nodes, and the type of the one or more second network nodes corresponds to the type of the third network node.

3. The apparatus according to any preceding claim, further caused to: Before sending the updated machine learning model to the one or more second network nodes, verify or modify the updated machine learning model.

4. The apparatus according to any preceding claim, further caused to: Sending a set of constraints for updating the machine learning model.

5. The apparatus according to any preceding claim, further caused to: Send information about a reference training process for updating the machine learning model, wherein the information about the reference training process includes at least a set of parameters used to construct the reference training process.

6. An apparatus according to any preceding claim, wherein the message comprises the updated machine learning model.

7. An apparatus according to any one of claims 1 to 5, wherein the message includes an updated set of weights and biases associated with the updated machine learning model.

8. The apparatus according to any preceding claim, further caused to: An updated machine learning model is obtained by updating the machine learning model based at least in part on the second training data.

9. The apparatus according to any preceding claim, further caused to: Receive information indicating a set of constraints for updating the machine learning model at the device.

10. An apparatus according to any preceding claim, further caused to: The updated machine learning model is sent to the one or more first network nodes.

11. An apparatus according to any preceding claim, wherein the one or more first network nodes comprises a plurality of network nodes of different types.

12. An apparatus according to any preceding claim, wherein the first training data comprises at least one of the following: reference signal measurement information measured at the one or more first network nodes, simulated reference signal measurement information, or simulated reference signal measurement information related to the one or more first network nodes.

13. The apparatus of any preceding claim, wherein the second training data comprises reference signal measurement information measured at a third network node different from the one or more first network nodes.

14. The apparatus according to any one of claims 12 to 13, wherein the reference signal measurement information comprises at least a channel impulse response measurement measured from one or more received positioning reference signals.

15. An apparatus comprising at least one processor and at least one memory, the at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus to at least: receiving information comprising at least one of: a machine learning model for positioning, a request to update the machine learning model at the device based on second training data, or a request to provide the second training data for updating the machine learning model, wherein the machine learning model has been trained based on first training data associated with one or more first network nodes; Acquire the second training data; as well as Send at least one of: a message indicating an updated machine learning model or the second training data.

16. The apparatus of claim 15, further caused to: The updated machine learning model is obtained by updating the machine learning model based on the second training data.

17. An apparatus according to any one of claims 15 to 16, wherein the message is sent based on the estimated performance improvement of the updated machine learning model being above a threshold.

18. The apparatus according to any one of claims 15 to 17, further caused to: Receive information indicating a set of constraints for updating the machine learning model at the device.

19. The apparatus according to any one of claims 15 to 18, further caused to: Receive information about a reference training process for updating the machine learning model at the device, wherein the information about the reference training process includes at least a set of parameters used to construct the reference training process.

20. An apparatus according to any one of claims 15 to 19, wherein the message includes the updated machine learning model.

21. An apparatus according to any one of claims 15 to 19, wherein the message includes an updated set of weights and biases associated with the updated machine learning model.

22. The apparatus according to any one of claims 15 to 21, further caused to: Sending a set of constraints for updating the machine learning model.

23. An apparatus according to any one of claims 15 to 22, wherein the first training data comprises at least one of the following: reference signal measurement information measured at the one or more first network nodes, simulated reference signal measurement information, or simulated reference signal measurement information related to the one or more first network nodes.

24. The apparatus according to any one of claims 15 to 23, wherein the second training data comprises reference signal measurement information measured at the apparatus.

25. The apparatus of any one of claims 23 to 24, wherein the reference signal measurement information comprises at least a channel impulse response measurement measured from one or more received positioning reference signals.

26. An apparatus comprising: means for acquiring a machine learning model for positioning, wherein the machine learning model is trained based on first training data associated with one or more first network nodes; means for sending information comprising at least one of: the machine learning model, a request to update the machine learning model based on second training data, or a request to provide the second training data for updating the machine learning model at the device; means for receiving at least one of: a message indicating an updated machine learning model, or said second training data; as well as Means for sending the updated machine learning model to at least one or more second network nodes.

27. An apparatus comprising: means for receiving information comprising at least one of: a machine learning model for positioning, a request for updating the machine learning model at the device based on second training data, or a request for providing the second training data for updating the machine learning model, wherein the machine learning model has been trained based on first training data associated with one or more first network nodes; A component for acquiring the second training data; as well as A component for sending at least one of the following: a message indicating an updated machine learning model or the second training data.

28. A method comprising: Acquiring, by the device, a machine learning model for positioning, wherein the machine learning model is trained based on first training data associated with one or more first network nodes; sending, by the device, information comprising at least one of: the machine learning model, a request for updating the machine learning model based on second training data, or a request for providing the second training data for updating the machine learning model at the device; Receiving, by the device, at least one of: a message indicating an updated machine learning model, or the second training data; as well as The device sends the updated machine learning model to at least one or more second network nodes.

29. A method comprising: receiving, by a device, information comprising at least one of: a machine learning model for positioning, a request to update the machine learning model at the device based on second training data, or a request to provide the second training data for updating the machine learning model, wherein the machine learning model has been trained based on first training data associated with one or more first network nodes; acquiring, by the device, the second training data; as well as At least one of the following is sent by the device: a message indicating an updated machine learning model, or the second training data.

30. A non-transitory computer readable medium comprising program instructions which, when executed by an apparatus, cause the apparatus to at least perform the following: Acquire a machine learning model for positioning, wherein the machine learning model is trained based on first training data related to one or more first network nodes; sending information including at least one of: the machine learning model, a request to update the machine learning model based on second training data, or a request to provide the second training data for updating the machine learning model at the device; receiving at least one of: a message indicating an updated machine learning model, or the second training data; as well as The updated machine learning model is sent to at least one or more second network nodes.

31. A non-transitory computer readable medium comprising program instructions, which when executed by an apparatus cause the apparatus to at least perform the following: receiving information comprising at least one of: a machine learning model for positioning, a request to update the machine learning model at the device based on second training data, or a request to provide the second training data for updating the machine learning model, wherein the machine learning model has been trained based on first training data associated with one or more first network nodes; Acquire the second training data; as well as Send at least one of: a message indicating the updated machine learning model, or the second training data.

Images