Systems, methods, and apparatus for model validity for AI-based user equipment (UE) positioning

Abstract

One or more of the techniques described herein may include determining a location of a user equipment (UE) using a neural network (NN) model. The output of the NN may be the location of the UE, or may be used in a subsequent process to determine the location of the UE. Techniques described herein may enable an NN model to be configured, tagged, and verified for use in one or more conditions or scenarios. The processes described herein may involve one or more of a UE, a base station, an Artificial Intelligence (AI)-Management Function (MF) (AI-MF) server, and a Location Management Function (LMF), and may include a model configuration function, a model selection function, and a model inference function.

Description

[0001] Citation of Related Applications

[0002] This application claims the benefit of U.S. Provisional Application No. 63 / 443,073, filed February 3, 2023, the contents of which are hereby incorporated by reference in their entirety. Technical Field

[0003] The present disclosure relates to wireless communication networks and mobile device capabilities. Background Art

[0004] Wireless communication networks and wireless communication services are becoming increasingly dynamic, complex, and ubiquitous. For example, some wireless communication networks may be developed to implement fifth-generation (5G) or new radio (NR) technologies, sixth-generation (6G) technologies, and the like. Such technologies may include solutions for enabling user equipment (UE) and network devices (such as base stations) to communicate with each other. Features of such networks and devices may include attempts to determine the geographic location of the UE. BRIEF DESCRIPTION OF THE DRAWINGS

[0005] The present disclosure will be easily understood and implemented through the detailed description and accompanying drawings. The same reference numerals may designate the same features and structural elements. The drawings and corresponding descriptions are provided as non-limiting examples of aspects, implementations, etc. of the present disclosure, and references to "one" or "an" aspect, implementation, etc. may not necessarily refer to the same aspect, implementation, etc., and may mean at least one, one or more, etc.

[0006] Figure 1 is a diagram that provides an example overview of one or more of the techniques described herein.

[0007] Figure 2 is a diagram of an example network according to one or more implementations described herein.

[0008] Figure 3 is a diagram of an example of a user equipment (UE), a base station, a core network, and an artificial intelligence (AI)-management function (MF) (AI-MF) server according to one or more implementations described herein.

[0009] Figure 4 is a diagram of examples of AI / machine learning (ML) (AI / ML) functionality and models according to one or more implementations described herein.

[0010] Figure 5 is a diagram of an example of a neural network (NN) 500 according to one or more implementations described herein.

[0011] Figure 6is a diagram of examples of functionality and corresponding entities and devices according to one or more implementations described herein.

[0012] Figure 7 is a diagram of an example table of characteristics of AI-based UE positioning according to one or more implementations described herein.

[0013] Figure 8 is a diagram of an example process for obtaining a valid NN model based on model validity data according to one or more implementations described herein.

[0014] Figure 9 is a diagram of an example process for determining a NN model based on auxiliary information and a model validity label according to one or more implementations described herein.

[0015] Figure 10 is a diagram of an example table of effectiveness labels for a NN model according to one or more implementations described herein.

[0016] Figure 11 is a diagram of an example process for selecting a NN model based on model validity testing according to one or more implementations described herein.

[0017] Figure 12 is a diagram of an example of a process for model selection according to one or more implementations described herein.

[0018] Figure 13 is a diagram of an example process for selecting a NN model based on a model class according to one or more implementations described herein.

[0019] Figure 14 is a diagram of an example data structure of NN model classes and validity labels according to one or more implementations described herein.

[0020] Figure 15 is a diagram of an example process for selecting a NN model based on validity data received for a NN model class according to one or more implementations described herein.

[0021] Figure 16 is a diagram of an example process for selecting multiple NN models based on validity data received for a class of NN models according to one or more implementations described herein.

[0022] Figure 17 is a diagram of an example process for model validation according to one or more implementations described herein.

[0023] Figure 18is a diagram of an example process for model validation according to one or more implementations described herein.

[0024] Figure 19 is a diagram of an example of components of a device according to one or more implementations described herein.

[0025] Figure 20 is a block diagram illustrating components capable of reading instructions from a machine-readable medium or computer-readable medium (e.g., a non-transitory machine-readable storage medium) and performing any one or more of the methodologies discussed herein in accordance with one or more implementations described herein. DETAILED DESCRIPTION

[0026] The following detailed description refers to the accompanying drawings. The same reference numerals in different drawings may identify the same or similar features, elements, operations, etc. Additionally, the present disclosure is not limited to the following description, as other specific implementations may be utilized and structural or logical changes may be made without departing from the scope of the present disclosure.

[0027] A wireless communication network may include user equipment (UE) capable of communicating with a base station and / or other network access nodes. The base station may provide the UE with access to a core network (CN) and additional external networks such as the Internet. The wireless communication network may implement various technologies and standards that enable services to be provided to the UE in a consistent and high-quality manner. Examples of such services may include those related to the geographic location of the UE. The value of such services generally depends on the level of accuracy with which the geographic location of the UE can be determined, and although currently available technologies can attempt to determine the geographic location of the UE, there is still room for improvement.

[0028] The techniques described herein enable the determination of a UE's location with greater accuracy by applying artificial intelligence (AI), machine learning (ML), and neural networks (NN) to the UE positioning process. These techniques may include: labeling different neural network (NN) models for use in determining the UE's location in different conditions or scenarios; validating and selecting an NN model given the current conditions or scenario; and using the selected NN model to determine the UE's location or position. These and many other processes, operations, and features are described below with reference to the accompanying figures.

[0029] Figure 11 is a diagram of an example overview 100 of one or more of the implementations described herein. Example overview 100 may include a UE 110 and / or a base station 120 and a LMF and / or AI-MF server 130. One or more of the techniques described herein may include processes or operations performed by different devices. For example, in some implementations, certain operations may be performed by the UE 110, while in other implementations, some or all of those operations may be performed by the base station 120 or the LMF and / or AI-MF server 130. Detailed examples and explanations of these variations are described below with reference to the accompanying figures. However, to simplify the explanation of example overview 100, "UE 110 and / or base station 120" may be referred to as "UE 110," and "LMF and / or AI-MF server 130" may be referred to as LMF 130.

[0030] As shown, the UE 110 and the LMF 130 can communicate to configure an NN model validity tag (at 1.1) for the NN model designed to locate the UE 110. The NN model validity tag may be referred to herein as a "validity tag," "tag," or the like. Different NN models may be configured and trained to operate under different conditions and in different situations. Tagging the NN model may include specifying appropriate conditions for using the NN model to determine the location of the UE. Such conditions may be defined using one or more of a variety of factors.

[0031] Examples of NN model labels may include capabilities of the NN model, capabilities of the UE, base station or other communication device, one or more dates, days of the week, times, geographic regions, countries, networks, cells or network access devices. Additional examples of NN model labels may include associating the NN model with positioning accuracy quality (e.g., the accuracy with which the NN model can determine or help determine the geolocation or position of the UE) and model inference latency (e.g., the amount of time typically involved in using the NN model to determine the position or location of the UE). Further examples of NN model labels may include UEs that support one or more types of auxiliary signaling (e.g., location assistance signaling) or reference signal configurations. In some implementations, the NN model label may correspond to input layer information of the corresponding NN model.

[0032] The UE 110 and the LMF 130 may also operate to determine current conditions for locating the UE 110 using the NN models (at 1.2). For example, the UE 110 or the LMF 130 may initiate a UE positioning procedure that may involve determining the geographic location of the UE 110. As part of such a procedure (and / or another type of procedure), the UE 110 and / or the LMF 130 may determine the current conditions or circumstances of the UE 110 and match the current conditions or circumstances with the validity tags of the NN models. Doing so may enable the UE 110 and / or the LMF 130 to determine which of the NN models is valid (e.g., appropriate, functional, etc.) for determining the location of the UE 110. Similar to the validity tags described above, the relevant conditions or circumstances of the UE 110 may be any number of multiple factors or combinations thereof, such as the capabilities of the UE, the capabilities of the NN models, supported reference signal configurations, date, day of the week, time, etc.

[0033] When the current conditions satisfy the validity tag of the NN model, the UE 110 and / or LMF 130 may select the NN model (from NN models that are more suitable for other conditions) (at 1.3), and the UE 110 and / or LMF 130 may begin using the NN model to determine the geographic location or position of the UE 110 (at 1.4). Some NN models may be configured to directly determine the location of the UE 110, meaning that the output of the NN model may be the geographic location of the UE 110. Other NN models may be configured to output information that is configured to assist or aid in the positioning of the UE 110 (e.g., by being applied to another NN model, a subsequent positioning algorithm, etc.). Thus, one or more of the techniques described herein may enable AI-based positioning of a UE by ensuring that different NN models are applied to appropriate conditions and scenarios.

[0034] In some implementations, the techniques described herein may include NN model selection based on validity test data to confirm whether the selected NN model matches the requirements or labels of the NN model. An NN model selection entity (e.g., UE 110 or LMF 130) may receive validity test data that may include information about the current conditions, capabilities, or scenarios in which the NN model is to be used, and the selection entity may determine whether the validity test data satisfies the NN model labels of one or more NN models. In some implementations, the NN models may be organized into classes or groups, where each class corresponds to a different category of NN models, and each NN model within an NN model class is associated with a different set of labels. In such scenarios, the NN model selection entity may receive validity test data and multiple model classes, and the NN model selection entity may use the validity test data to validate one or more NN models from one or more of the model classes.

[0035] In other implementations, the UE 110 may transmit information describing the NN models supported by the UE 110 and assistance information, if available (e.g., reference signal measurements, supported reference signal configurations, etc.) to the LMF 130. Based on this information, the LMF 130 may provide a set of NN models and their corresponding validity tags to the UE 110. The UE 110 may test the NN models based on the current conditions and the validity tags, select an appropriate NN model, and notify the LMF 130 of the selection. The LMF 130 may respond by providing model configuration information and support (e.g., assistance information) to the UE 110, which the UE 110 may use to apply the selected NN model during the AI-based positioning process. Additional examples of these and many other techniques, features, and implementations are described below with reference to the following figures.

[0036] Figure 2 2 is an example network 200 according to one or more implementations described herein. Example network 200 may include UE 210-1, UE 210-2, etc. (collectively referred to as "UE 210" and individually as "UE 210"), a radio access network (RAN) 220, a core network (CN) 230, an application server 240, and an external network 250.

[0037] The systems and devices of example network 200 may operate in accordance with one or more communication standards, such as the 2nd Generation (2G) communication standard of the 3rd Generation Partnership Project (3GPP), the 3rd Generation (3G) communication standard, the 4th Generation (4G) (e.g., Long Term Evolution (LTE)) communication standard, and / or the 5th Generation (5G) (e.g., New Radio (NR)) communication standard. Additionally or alternatively, one or more of the systems and devices of example network 200 may operate in accordance with other communication standards and protocols discussed herein, including future versions or generations of 3GPP standards (e.g., the 6th Generation (6G) standard, the 7th Generation (7G) standard, etc.), Institute of Electrical and Electronics Engineers (IEEE) standards (e.g., Wireless Metropolitan Area Network (WMAN), Worldwide Interoperability for Microwave Access (WiMAX), etc.), and more.

[0038] As shown, UE 210 may include a smartphone (e.g., a handheld touchscreen mobile computing device capable of connecting to one or more wireless communication networks). Additionally or alternatively, UE 210 may include other types of mobile computing devices or non-mobile computing devices capable of wireless communication, such as a personal data assistant (PDA), a pager, a laptop computer, a desktop computer, a wireless handheld phone, a watch, etc. In some implementations, UE 210 may include an Internet of Things (IoT) device (or IoT UE), which may include a network access layer designed for low-power IoT applications that utilize short UE connections. Additionally or alternatively, the IoT UE may utilize one or more types of technologies such as machine-to-machine (M2M) communication or machine-type communication (MTC) (e.g., to exchange data with an MTC server or other device via a public land mobile network (PLMN), proximity services (ProSe) or device-to-device (D2D) communication, sensor networks, IoT networks, and the like. Depending on the scenario, the M2M or MTC exchange of data may be machine-initiated, and the IoT network may include IoT UEs (which may include uniquely identifiable embedded computing devices within the Internet infrastructure) interconnected via ephemeral connections. In some scenarios, the IoT UEs may execute background applications (e.g., keep-alive messages, status updates, etc.) to facilitate connectivity to the IoT network.

[0039] A UE 210 can communicate with and establish connections with one or more other UEs 210 via one or more radio channels 212, each of which can include a physical communication interface / layer. Connections can include M2M connections, MTC connections, D2D connections, SL connections, and the like. Connections can involve a PC5 interface. In some implementations, the UEs 210 can be configured to discover each other, negotiate radio resources between each other, and establish connections between each other without intervention or communication involving a RAN node 222 or another type of network node. In some implementations, discovery, authentication, resource negotiation, registration, and the like can involve communication with a RAN node 222 or another type of network node.

[0040] UEs 210 may communicate with each other using one or more wireless channels 212. As described herein, UE 210-1 may communicate with RAN node 222 to request SL resources. RAN node 222 may respond to the request by providing UE 210 with a dynamic grant (DG) or a configured grant (CG) regarding SL resources. A DG may involve a grant based on a grant request from UE 210. A CG may involve a resource grant without a grant request and may be based on the type of service being provided (e.g., a service with strict timing or latency requirements). UE 210 may perform a clear channel assessment (CCA) procedure based on the DG or CG, select SL resources based on the CCA procedure and the DG or CG, and communicate with another UE 210 based on the SL resources. UE 210 may communicate with RAN node 222 using a licensed band and communicate with another UE 210 using an unlicensed band.

[0041] UE 210 may communicate with and establish a connection (e.g., be communicatively coupled) to RAN 220, which may involve one or more radio channels 214-1 and 214-2, each of which may include a physical communication interface / layer. In some implementations, the UE may be configured with dual connectivity (DC) as a multi-radio access technology (multi-RAT) or multi-radio dual connectivity (MR-DC), wherein a UE capable of multiple reception and transmission (Rx / Tx) may use resources provided by different network nodes (e.g., 222-1 and 222-2), which may be connected via non-ideal backhaul (e.g., one network node provides NR access and the other network node provides E-UTRA for LTE or NR access for 5G). In such a scenario, one network node may act as a master node (MN) and the other node may act as a secondary node (SN). The MN and SN may be connected via a network interface, and at least the MN may be connected to CN 230. Additionally, at least one of the MN or the SN may operate via shared spectrum channel access, and the functionality specified for the UE 210 may be used for an integrated access and backhaul mobile terminal (IAB-MT). Similar to the UE 210, the IAB-MT may access the network using one network node or using two different nodes with an enhanced dual connectivity (EN-DC) architecture or a new radio dual connectivity (NR-DC) architecture. In some implementations, a base station (as described herein) may be an example of the network node 222.

[0042] As described herein, the UE 210 may receive and store one or more configurations, instructions, and / or other information for implementing SL-U communications with quality and priority criteria. A PQI may be determined and used to indicate the QoS associated with an SL-U communication (e.g., a channel, a data flow, etc.). Similarly, an L1 priority value may be determined and used to indicate the priority of an SL-U transmission, an SL-U channel, SL-U data, etc. The PQI and / or L1 priority value may be mapped to a CAPC value, and the PQI, L1 priority, and / or CAPC may indicate SL channel occupancy time (COT) sharing, maximum (MCOT), a timing interval for COT sharing, LBT configuration, traffic and channel priority, etc.

[0043] As shown, UE 210 may also or alternatively be connected to an access point (AP) 216 via a connection interface 218, which may include an air interface that enables UE 210 to communicatively couple with AP 216. AP 216 may include a wireless local area network (WLAN), a WLAN node, a WLAN termination point, etc. Connection 218 may include a local wireless connection, such as a connection consistent with any IEEE 702.11 protocol, and AP 216 may include a wireless fidelity protocol. Router or other AP. Figure 2 218. Although not explicitly depicted in the figure, AP 216 may be connected to another network (e.g., the Internet) without being connected to RAN 220 or CN 230. In some scenarios, UE 210, RAN 220, and AP 216 may be configured to utilize LTE-WLAN aggregation (LWA) technology or LTE / WLAN radio level technology integrated with IPsec tunneling (LWIP). LWA may involve RAN 220 configuring UE 210 in an RRC_CONNECTED state to utilize radio resources of LTE and WLAN. LWIP may involve UE 210 using WLAN radio resources (e.g., connection interface 218) to authenticate and encrypt packets (e.g., Internet Protocol (IP) packets) communicated via connection interface 218 via IPsec protocol tunneling. IPsec tunneling may include encapsulating the entire original IP packet and adding a new packet header, thereby protecting the original header of the IP packet.

[0044] The RAN 220 may include one or more RAN nodes 222-1 and 222-2 (collectively, and individually, RAN nodes 222) that enable establishment of a channel 214-1 and a channel 214-2 between the UE 210 and the RAN 220. The RAN node 222 may include a network access point configured to provide radio baseband functionality for data and / or voice connectivity between a user and a network based on one or more of the communication technologies described herein (e.g., 2G, 3G, 4G, 5G, WiFi, etc.). Thus, as an example, the RAN node may be an E-UTRAN Node B (e.g., an enhanced Node B, eNodeB, eNB, 4G base station, etc.), a next-generation base station (e.g., a 5G base station, an NR base station, a next-generation eNB (gNB), etc.). The RAN node 222 may include a roadside unit (RSU), a transmit / receive point (TRxP or TRP), and one or more other types of ground stations (e.g., a terrestrial access point). In some scenarios, the RAN node 222 may be a dedicated physical device such as a macrocell base station and / or a low power (LP) base station for providing a femtocell, picocell, etc. having a smaller coverage area, smaller user capacity, or higher bandwidth than a macrocell.

[0045] Some or all of the RAN nodes 222, or portions thereof, may be implemented as one or more software entities running on a server computer as part of a virtual network that may be referred to as a centralized RAN (CRAN) and / or a virtual baseband unit pool (vBBUP). In these implementations, the CRAN or vBBUP may implement: RAN functional splits, such as a packet data convergence protocol (PDCP) split, where the radio resource control (RRC) and PDCP layers may be operated by the CRAN / vBBUP, and other layer 2 (L2) protocol entities may be operated by separate RAN nodes 222; a medium access control (MAC) / physical (PHY) layer split, where the RRC, PDCP, radio link control (RLC), and MAC layers may be operated by the CRAN / vBBUP, and the PHY layer may be operated by individual RAN nodes 222; or a "lower PHY" split, where the RRC, PDCP, RLC, MAC layer, and upper portions of the PHY layer may be operated by the CRAN / vBBUP, and the lower portions of the PHY layer may be operated by individual RAN nodes 222. The virtualization framework may allow idle processor cores of the RAN node 222 to perform or execute other virtualized applications.

[0046] In some implementations, the separate RAN nodes 222 may represent separate gNB distributed units (DUs) connected to a gNB control unit (CU) via separate F1 or other interfaces. In such implementations, the gNB-DU may include one or more remote radio heads or radio frequency (RF) front-end modules (RFEMs), and the gNB-CU may be operated by a server (not shown) located in the RAN 220 or by a server pool (e.g., a group of servers configured to share resources) in a manner similar to a CRAN / vBBUP. Additionally or alternatively, one or more of the RAN nodes 222 may be next-generation eNBs (i.e., gNBs) that may provide Evolved Universal Terrestrial Radio Access (E-UTRA) user and control plane protocol terminations to the UE 210 and may be connected to the 5G core network (5GC) 230 via an NG interface.

[0047] Any of the RAN nodes 222 can serve as an endpoint for an air interface protocol and can be the first point of contact for the UE 210. In some implementations, any of the RAN nodes 222 can perform various logical functions of the RAN 220, including, but not limited to, functions of a radio network controller (RNC), such as radio bearer management, uplink and downlink dynamic radio resource management, data packet scheduling, and mobility management. The UE 210 can be configured to communicate with each other or with any of the RAN nodes 222 using orthogonal frequency division multiplexing (OFDM) communication signals over multi-carrier communication channels according to various communication technologies, such as, but not limited to, OFDMA communication technologies (e.g., for downlink communications) or single-carrier frequency division multiple access (SC-FDMA) communication technologies (e.g., for uplink and ProSe or sidelink (SL) communications), although the scope of such implementations may not be limited in this respect. OFDM signals may include multiple orthogonal subcarriers.

[0048] In some implementations, a downlink resource grid can be used for downlink transmissions from any of the RAN nodes 222 to the UE 210, and similar techniques can be used for uplink transmissions. The grid can be a time-frequency grid (e.g., a resource grid or a time-frequency resource grid) that represents the physical resources for the downlink in each time slot. This type of time-frequency representation is common for OFDM systems and makes radio resource allocation intuitive. Each column and row of the resource grid corresponds to an OFDM symbol and an OFDM subcarrier, respectively. The duration of the resource grid in the time domain corresponds to a time slot in a radio frame. The smallest time-frequency unit in the resource grid is represented as a resource element. Each resource grid includes resource blocks, which describe the mapping of certain physical channels to resource elements. Each resource block can include a set of resource elements (REs); in the frequency domain, this can represent the minimum amount of resources currently available for allocation. Such resource blocks are used to transport several different physical downlink channels.

[0049] Further, the RAN node 222 may be configured to wirelessly communicate with the UE 210 and / or with each other over a licensed medium (also referred to as a "licensed spectrum" and / or a "licensed band"), an unlicensed shared medium (also referred to as an "unlicensed spectrum" and / or an "unlicensed band"), or a combination thereof. For example, the licensed spectrum may include channels operating in the frequency range of approximately 400 MHz to approximately 3.8 GHz, while the unlicensed spectrum may include the 5 GHz band. The licensed spectrum may correspond to channels or frequency bands that are selected, reserved, regulated, etc. for certain types of wireless activities (e.g., wireless telecommunications network activities), while the unlicensed spectrum may correspond to one or more frequency bands that are not restricted for certain types of wireless activities. Whether a particular frequency band corresponds to a licensed medium or an unlicensed medium may depend on one or more factors, such as frequency allocations determined by a public sector organization (e.g., a government agency, a regulatory agency, etc.) or by a private sector organization involved in developing wireless communication standards and protocols.

[0050] To operate in the unlicensed spectrum, the UE 210 and the RAN node 222 may operate using standalone unlicensed operation, license-assisted access (LAA), eLAA and / or feLAA mechanisms, and / or NR unlicensed mechanisms. In these implementations, the UE 210 and the RAN node 222 may perform one or more known medium sensing operations or carrier sensing operations to determine whether one or more channels in the unlicensed spectrum are unavailable or otherwise occupied before transmitting in the unlicensed spectrum. The medium / carrier sensing operations may be performed in accordance with a listen-before-talk (LBT) protocol.

[0051] The LAA mechanism can be built on the carrier aggregation (CA) technology of the LTE-Advanced system. In CA, each aggregated carrier is called a component carrier (CC). In some cases, each CC may have a different bandwidth from other CCs. In a time division duplex (TDD) system, the number of CCs and the bandwidth of each CC may be the same for DL ​​and UL. CA also includes various serving cells to provide various CCs. The coverage of the serving cells may be different, for example, because CCs on different frequency bands will experience different path losses. The primary serving cell or PCell can provide a primary component carrier (PCC) for both UL and DL, and can handle RRC and non-access stratum (NAS) related activities. Other serving cells are called SCells, and each SCell can provide a single secondary component carrier (SCC) for both UL and DL.

[0052] The PDSCH can carry user data and higher layer signaling to UE 210. The physical downlink control channel (PDCCH) can carry information about, among other things, the transport format and resource allocation associated with the PDSCH channel. The PDCCH can also inform UE 210 about the transport format, resource allocation, and hybrid automatic repeat request (HARQ) information associated with the uplink shared channel. Typically, downlink scheduling (assigning control and shared channel resource blocks to UE 210-2 within a cell) can be performed on any of the RAN nodes 222 based on channel quality information fed back from any of the UEs 210. Downlink resource allocation information can be sent on the PDCCH for (e.g., assigned to) each of the UEs 210.

[0053] PDCCH uses control channel elements (CCE) to convey control information, where several CCEs (e.g., 6, etc.) may be composed of resource element groups (REGs), where REGs are defined as physical resource blocks (PRBs) in OFDM symbols. Before being mapped to resource elements, PDCCH complex-valued symbols may first be organized into quadruples, which may then be arranged, for example, using a sub-block interleaver for rate matching. Each PDCCH may be sent using one or more of these CCEs, where each CCE may correspond to nine sets of four physical resource elements, referred to as REGs. Four quadrature phase shift keying (QPSK) symbols may be mapped to each REG. Depending on the size of the DCI and the channel conditions, one or more CCEs may be used to send the PDCCH. Four or more different PDCCH formats may be defined in LTE with different numbers of CCEs (e.g., aggregation levels, L=1, 2, 4, 8, or 26).

[0054] Some implementations may use the concept of resource allocation for control channel information, which is an extension of the above concept. For example, some implementations may utilize an extended (E)-PDCCH that uses PDSCH resources for control information transmission. One or more ECCEs may be used to transmit EPDCCH. Similar to the above, each ECCE may correspond to a set of nine four physical resource elements, called EREGs. In some cases, an ECCE may have other numbers of EREGs.

[0055] The RAN nodes 222 may be configured to communicate with each other via an interface 223. In an implementation where the system is an LTE system, the interface 223 may be an X2 interface. In an NR system, the interface 223 may be an Xn interface. The X2 interface may be defined between two or more RAN nodes 222 (e.g., two or more eNBs / gNBs or a combination thereof) connected to an Evolved Packet Core (EPC) or CN 230, or between two eNBs connected to an EPC. In some implementations, the X2 interface may include an X2 user plane interface (X2-U) and an X2 control plane interface (X2-C). The X2-U may provide a flow control mechanism for user data packets transmitted over the X2 interface and may be used to convey information about the delivery of user data between eNBs or gNBs. For example, X2-U may provide specific sequence number information about user data transmitted from a master eNB (MeNB) to a secondary eNB (SeNB); information about successful in-sequence delivery of PDCP packet data units (PDUs) for user data from the SeNB to the UE 210; information about PDCP PDUs that were not delivered to the UE 210; information about the current minimum expected buffer size at the SeNB for sending user data to the UE; etc. X2-C may provide intra-LTE access mobility functions (e.g., including context transfer from a source eNB to a target eNB, user plane transmission control, etc.), load management functions, and inter-cell interference coordination functions.

[0056] As shown, RAN 220 can be connected (e.g., communicatively coupled) to CN 230. CN 230 can include multiple network elements 232 configured to provide various data and telecommunication services to customers / subscribers (e.g., users of UE 210) connected to CN 230 via RAN 220. In some implementations, CN 230 can include an evolved packet core (EPC), a 5G CN, and / or one or more additional or alternative types of CNs. Components of CN 230 can be implemented in one physical node or separate physical nodes, including components for reading and executing instructions from a machine-readable medium or computer-readable medium (e.g., a non-transitory machine-readable storage medium). In some implementations, network function virtualization (NFV) can be used to virtualize any or all of the above-mentioned network node roles or functions via executable instructions stored in one or more computer-readable storage media (described in further detail below). A logical instance of CN 230 can be referred to as a network slice, and a logical instance of a portion of CN 230 can be referred to as a network sub-slice. Network Function Virtualization (NFV) architecture and infrastructure can be used to virtualize one or more network functions onto physical resources including a combination of industry-standard server hardware, storage hardware, or switches (alternatively, performed by proprietary hardware). In other words, the NFV system can be used to perform virtual or reconfigurable implementations of one or more EPC components / functions.

[0057] As shown, CN 230, application server 240, and external network 250 may be connected to each other via interfaces 234, 236, and 238, which may include IP network interfaces. Application server 240 may include one or more server devices or network elements (e.g., virtual network functions (VNFs) that provide applications that use IP bearer resources via CN 230 (e.g., Universal Mobile Telecommunications System Packet Service (UMTS PS) domain, LTE PS data service, etc.). Application server 240 may also or alternatively be configured to support one or more communication services (e.g., IP voice (VoIP sessions, push-to-talk (PTT) sessions, group communication sessions, social networking services, etc.)) for UE 210 via CN 230. Similarly, external network 250 may include one or more of a variety of networks, including the Internet, thereby providing network access to various additional services, information, interconnectivity, and other network features to the mobile communication network and UE 210.

[0058] The AI-MF server 270 may include one or more servers, server devices, or network elements (e.g., VNFs) configured to transmit, receive, process, and / or store information. The AI-MF server 270 may communicate with a CN 230 interface 272, which may include an IP interface. The AI-MF server 270 may support and provide functionality regarding model validity for AI-based UE positioning as described herein. For example, the AI-MF server 270 may enable NN models for AI-based UE positioning to be configured, labeled, and / or validated for one or more conditions or scenarios, and may include model configuration functionality, model selection functionality, and / or model inference functionality. In some implementations, the AI-MF server 270 may also or alternatively perform one or more functions performed by a location management function (LMF) of the CN 230.

[0059] Figure 3 2 is a diagram of an example of a UE 210, a base station 222, a CN 230, and an AI-MF server 270 according to one or more implementations described herein. As shown, the CN 230 may include an access and mobility management function (AMF) 310, a location management function (LMF) 320, and / or one or more other types of functions or entities 330. Examples of such functions or entities may include a session management function (SMF), a unified data management (UDM) function, a gateway mobile location center (GMLC), etc. The AMF 310, LMF 320, etc. may be implemented by one or more servers in a centralized or distributed networking environment.

[0060] The AMF 310 can communicate with the base station 222 via the N2 interface and with the UE 210 via the N1 interface. The AMF 310 can manage authentication, registration, and other functions related to the UE 210 accessing the telecommunications mobile network. The AMF 310 can also handle handover, paging, and other functions related to the mobility and communication of the user equipment (UE) 210 in the telecommunications mobile network. The AMF 310 can also provide security functions for authenticating and authorizing the UE 210.

[0061] The LMF 320 may provide a positioning function to determine the geographic location of the UE 210 based on downlink (DL) position measurement radio signals and uplink (UL) position measurement radio signals. The LMF 320 may receive measurement and assistance information from the base station 222 and the UE 210 via the AMF 310 and the NL interface. The LMF 320 may use the measurement and assistance information to calculate the positioning of the UE 210. The new NR Positioning Protocol A (NRPPa) protocol may be used to carry positioning information between the base station 222 and the LMF 320 over the next generation control plane interface (NG-C). The LMF 320 may also configure the UE 210 using the LTE Positioning Protocol (LPP) via the AMF 310, and the base station 222 may configure the UE 210 using the RRC protocol over the LTE-Uu interface and / or the NR-Uu interface.

[0062] The LMF 320 and / or the AI-MF server 270 may provide positioning assistance data to the UE 210. Examples of such information may include information about the signal to be measured (e.g., expected signal timing, signal coding, signal frequency, signal Doppler, etc.), the location and identity of terrestrial transmitters (e.g., base stations 222, APs 216, etc.), and / or signal, timing, and orbit information of non-terrestrial transmitters (such as satellites and satellite systems). Doing so may improve the signal acquisition and measurement accuracy of the UE 210 and, in some cases, enable the UE 210 to better determine its current geographic location based on the position measurements. The LMF 320 may implement a protocol to transmit the AI-MF information described herein. The protocol may be part of the new NR Positioning Protocol A (NRPPa) protocol, another type of positioning protocol, or a newly developed positioning protocol.

[0063] The LMF 320 and / or the AI-MF server 270 may provide the UE 210 with information indicating the location and identity of terrestrial transmitters and / or non-terrestrial transmitters corresponding to a particular area and / or signaling information, such as transmit power, signal timing, etc. The UE 210 may obtain a measurement of the signal strength of signals received from such transceivers (e.g., a received signal strength indication (RSSI)) and / or may obtain a signal-to-noise ratio (S / N), a reference signal received power (RSRP), a reference signal received quality (RSRQ), a time of arrival (TOA), or a round-trip signal propagation time (RTT) between the UE 210 and one or more transceivers (e.g., a base station 222, an AP 216, etc.).

[0064] The UE 210 may transmit these measurements to the LMF 320 and / or the AI-MF server 270 to determine the location of the UE 210, or in some implementations, may use these measurements together with assistance data (e.g., information indicating the location and identity of terrestrial and / or non-terrestrial transmitters) received from a location server (e.g., the LMF 320 and / or the AI-MF server 270) or broadcast by the base station 222 to determine the location of the UE 210. The UE 210 may measure the reference signal time difference (RSTD) between signals such as positioning reference signals (PRS), cell-specific reference signals (CRS), or tracking reference signals (TRS) transmitted by nearby pairs of transceivers. The RSTD measurement may provide the time difference in arrival between signals (e.g., TRS, CRS, or PRS) received at the UE 210 from two different transceivers. UE 210 may return the measured RSTD to LMF 320 and / or AI-MF server 270, and LMF 320 and / or AI-MF server 270 may calculate an estimated position of UE 210 based on the measured known positions and known signal timing of the transceivers.

[0065] The LMF 320 and / or the AI-MF server 270 may support and provide functionality regarding model validity for AI-based UE positioning. In some implementations, some or all of the functionality described herein as being performed by the LMF 320 and / or the AI-MF server 270 may be performed by one or more other types of functions or entities, including the base station 222, the application server 240, and / or another function or entity of the CN 230.

[0066] Figure 4 is a diagram of an example of an AI / ML function and model 400 according to one or more implementations described herein. As shown, the example 400 may include a data collection function 410, a model training function 420, a model inference function 430, and an actor 440. In some implementations, the AI / ML function and model 400 may be implemented by one or more UEs 210, one or more base stations 222, and / or one or more elements of the CN 230, such as the LMF 320. As described herein, the AI / ML function and model 400 may be implemented to enhance throughput, robustness, accuracy, reliability, and positioning accuracy in different scenarios, such as those with severe non-line-of-sight (NLOS) conditions.

[0067] The data collection function 410 may provide input data to the model training function 420 and the model inference function 430. AI / ML algorithm-specific data preparation (e.g., data preprocessing and cleaning, formatting, and transformation) may not be performed by the data collection function 410. Examples of input data may include measurements from the UE 210 or different network entities, feedback from the actor 440, and output from the AI / ML model. The AI / ML model may include a framework capable of evaluating input data and generating features, vectors, and / or functions for output. In some implementations, the AI / ML model may include a trained neural network.

[0068] The training data may include inputs to the AI / ML model training function. The inference data may include inputs for the model inference function 430. The model training function 420 may perform AI / ML model training, validation, and testing, which may generate model performance metrics as part of the model testing process. The model training function 420 may also be responsible for data preparation (e.g., data preprocessing and cleaning, formatting, and transformation) based on the training data delivered by the data collection function 410. Model deployment / update may be used to initially deploy a trained, validated, and tested AI / ML model to the model inference function 430, or to deliver an updated model to the model inference function 430.

[0069] The model inference function 430 can provide AI / ML model inference output (e.g., prediction or decision). The model inference function 430 can provide model performance feedback to the model training function 420 when applicable. The model inference function 430 is also responsible for data preparation (e.g., data preprocessing and cleaning, formatting, and transformation) based on the inference data delivered by the data collection function 410. The inference output of the AI / ML model is generated by the model inference function 430. The details of the inference output can be use case specific. When available, model performance feedback can be used to monitor the performance of the AI / ML model. The actor function 440 can receive the output from the model inference function 430 and trigger or execute corresponding actions. The actor function 440 can trigger actions involving other entities or itself. The feedback information can be used to derive training data, inference data, or monitor the performance of the AI / ML model and its impact on the network by updating performance indicators and performance counters.

[0070] Figure 5is a diagram of an example of a neural network (NN) 500 according to one or more implementations described herein. As shown, the NN 500 may include nodes arranged in different layers, such as an input layer 510 of nodes, multiple hidden layers or intermediate layers 520 of nodes, and an output layer 530 of nodes. In some implementations, the NN 500 may be an example or part of the model training function 420, the AI / ML model, the model inference function 430, and / or the actor function 440. For example, the NN 500 may be trained on training data from the data collection function 410, deployed as an AI / ML model by the model training function 420, and used by the model inference function 430 to generate feedback for the model training function 420 and inference output for the actor function 440.

[0071] The exemplary NN 500 may include N inputs introduced into four input nodes [N, 4] of an input layer 510. This may include processing or encoding the input data into a form, shape, vector, or data structure that can be received by the NN. The four input nodes may process the inputs to generate first weights (W1) that the four input nodes provide to the five nodes [4; 5] of a first hidden layer. The five nodes of the first hidden layer may process the inputs using a first function (f1) to generate second weights (W2) that the five nodes of the first hidden layer may provide to the five nodes [5; 5] of a second hidden layer. The five nodes of the second layer may process the inputs using a second function (f2) to generate third weights (W3) that the five nodes of the second hidden layer may provide to the three nodes [5; 3] of an output layer 530. The nodes of the output layer 530 may each process the received inputs and generate an output. This may include converting or unencoding the output data from a form, shape, vector, or data structure that can be used by a subsequent algorithm, process, or procedure.

[0072] Artificial intelligence (AI) can involve the combination of computer science and datasets to solve problems. AI can encompass machine learning (ML) and deep learning (DL), which are often mentioned together. These disciplines include AI algorithms, which attempt to create expert systems that make predictions or classifications based on input data. ML, DL, and neural networks (NNs) are subfields of AI. However, NNs are actually subfields of ML, and DL is a subfield of NNs. DL differs from ML in the way each algorithm learns. Deep ML can use labeled datasets (also known as supervised learning) to inform its algorithms, but it does not necessarily require labeled datasets. DL can ingest unstructured data in its raw form (e.g., text or images) and automatically determine a set of features that distinguish different categories of data from each other. This can eliminate some of the required human intervention and enable the use of larger datasets. In a sense, DL can be considered scalable ML.

[0073] A NN or artificial NN (ANN) may comprise logically interconnected nodes arranged in a node layer. There may be an input layer, one or more hidden layers or intermediate layers, and an output layer. Each node or artificial neuron may be connected to another node or artificial neuron and has an associated weight and threshold. If the output of any individual node is above a specified threshold, the node is activated, thereby passing data to the next layer of the network. Otherwise, the node does not pass data to the next layer of the network. The "depth" in deep learning refers only to the number of layers in the NN. A NN consisting of more than three layers (which will include inputs and outputs) can be considered a deep learning algorithm or a deep NN. A neural network with only three layers is just a basic NN.

[0074] A feedforward NN or multilayer perceptron (MLP) can include an input layer, one or more hidden layers, and an output layer. Although these NNs are also called MLPs, they may include sigmoid neurons instead of perceptrons because some real-world problems can be nonlinear. Data is often fed into these models to train them, and these models can be used as the basis for computer vision, natural language processing, and other neural networks. Recurrent neural networks (RNNs) are identified by feedback loops.

[0075] Convolutional neural networks (CNNs) can be similar to feedforward NNs, but can be used for image recognition, pattern recognition, and / or computer vision. These NNs can use principles from linear algebra, particularly matrix multiplication, to identify patterns within an image. For example, linear regression analysis can be used to predict the value of one variable based on the value of another variable. This form of analysis can estimate the coefficients of a linear equation involving one or more independent variables that best predict the value of the dependent variable. Linear regression can fit a line or surface that minimizes the difference between the predicted value and the actual value. These learning algorithms can be used when making predictions about future outcomes using time series data.

[0076] Figure 6 6 is a diagram of example 600 of functions and corresponding devices and entities according to one or more specific implementations described herein. As shown, example 600 includes model configuration functionality 610, model selection functionality 620, and model inference functionality 630 (collectively referred to herein as functions 610 to 630). Example 600 also includes UE 210, base station 222, LMF 320, and AI-MF server 270.

[0077] Example 600 provides an overview of the functions described herein and the devices and device combinations that can perform each function. That is, any of the functions 610 to 630 can be performed by any combination of the devices depicted. For example, in some implementations, functions 610 to 630 can be performed by a combination of UE 210 and LMF 320. In another implementation, functions 610 to 630 can be performed by a combination of UE 210 and AI-MF server 270. In another implementation, functions 610 to 630 can be performed by base station 222 and LMF 320 and / or AI-MF server 270. In yet another implementation, functions 610 to 630 can be performed by a combination of UE 210, base station 222, LMF 320, and AI-MF server 270.

[0078] The model configuration function 610 may include the process by which one or more models are created, configured, trained, and / or applied to new scenarios or conditions. In some implementations, model configuration may include creating a NN model to be applied by the UE 210, base station 222 under certain conditions, in a certain geographic location or region, and / or the like. Model configuration may involve applying training data to the training model training function 420, updating an existing model based on model performance feedback, and / or specifying an existing model for application to a new environment, condition, or scenario. Model configuration may include associating the NN model with one or more tags, conditions, or characteristics for application of the NN model. Examples of such conditions or characteristics may include the estimated geographic location of the UE 210, cell ID, capabilities of the NN model itself, device capability information (e.g., UE capability information), assistance information, one or more reference signal configurations, date, day of the week, time of day, and / or the like. In practice, associating the NN model with one or more tags may identify the NN model as relevant to the current conditions or scenario. Some or all of the model configuration functions in the model configuration function 610 may be performed by the UE 210, base station 222, LMF 320, AI-MF server 270, and / or any combination thereof.

[0079] The model selection function 620 may include a process for selecting a model for use. Model selection may be based on one or more tags associated with the model. As described herein, a tag may include characteristics, conditions, or scenarios related to the UE 210. Examples of tags may include the capabilities of the UE 210, the location of the UE 210, the current cell of the UE 210, a measured reference signal or signal strength, and / or one or more other conditions related to the UE 210. Since the model configuration function 610 may include associating a model with one or more tags, the model selection function 620 may include a process for matching the current conditions with one or more tags of a particular model. Some or all of the model selection functions in the model selection function 620 may be performed by the UE 210, the base station 222, the LMF 320, the AI-MF server 270, and / or any combination thereof.

[0080] The model inference function 630 may include the process by which the selected NN model is used to determine the location of the UE 210. The input data for the selected NN model may include the estimated geographic location of the UE 210, the cell ID in which the UE 210 is located, one or more signal strengths measured by the UE 210, assistance information, etc. In some implementations, the location of the UE 210 may be determined solely based on the results of the NN model. In other implementations, the output of the NN model may be part of the data set used to determine the location of the UE 210. That is, the output of the NN model may be used as input to another location determination function that also uses other information as input. The execution of the model inference function 630 may result in or contribute to a more precise or accurate location of the UE 210, which would otherwise be possible due to the applied NN model. As shown, some or all of the model inference functions in the model inference function 630 may be performed by the UE 210, the BS 222, the LMF 320, the AI-MF server 270, and / or any combination thereof.

[0081] Figure 7 7 is a diagram of an example table 700 of characteristics of AI-based UE positioning according to one or more implementations described herein. As shown, table 700 includes Case 1A, Case 1B, Case 2A, Case 2B, Case 3A, and Case 3B (collectively referred to herein as "Cases 1-3"). Cases 1-3 include various example devices and entities (e.g., UE 210, base station 222, LMF 320, etc.) that can use different NN models, inputs, outputs, and other types of information to implement AI-based positioning according to the techniques described herein. In some implementations, an AI-MF server 270 can be implemented or involved instead of one or more of UE 210, base station 222, or LMF 320. Therefore, not only are Cases 1-3 non-limiting examples of one or more of the techniques described herein, but Cases 1-3 illustrate the wide range of implementations and implementation variability of the techniques described herein.

[0082] Table 700 includes columns titled Positioning Type, Assistance Type, Model, and AI / ML Type. Positioning Type may refer to a device or entity that determines the location of UE 210. As shown, Case 1A and Case 1B may be UE-based scenarios, in which UE 210 may determine the location or position of UE 210. In contrast, Cases 2A, 2B, 3A, and 3B may be LMF-based scenarios, in which LMF 320 may determine the location of UE 210.

[0083] The assistance type may refer to a device or entity that can provide information to a positioning device (e.g., UE 210 for UE-based scenarios and LMF 320 for LMF-based scenarios) to assist in determining the location of UE 210. The assistance type may not apply to Cases 1A and 1B, as UE 210 determines its location in Cases 1A and 1B. However, Cases 2A and 2B of the assistance type may include assistance information from UE 210, and Cases 3A and 3B of the assistance type may include assistance information from NG-RAN / base station 222. For example, UE 210 may provide location assistance information to assist LMF 320, and LMF 320 may use this information as input to a NN model implemented by LMF 320 or as a parameter for another type of positioning process implemented by LMF 320. A model may refer to a device or entity that implements an NN model to enable the determination or inference of UE 210's location. As shown in the figure, UE 210 may implement the NN model in Cases 1A, 1B, and 2A. LMF 320 can implement the NN model in Case 2B and Case 3B, and base station 222 can implement the NN model in Case 3 A. In some implementations, different entities can implement the NN model in different cases.

[0084] The AI / ML type may refer to whether the position of the UE 210 is determined directly by the output of the NN model (AI / ML direct), or whether the output of the NN model is used to assist in determining the position of the UE 210 (AI / ML assisted). In the AI / ML assisted scenario, the output of the NN may be used as an input or parameter to another positioning process. Examples of such processes may include positioning processes designed to determine whether the UE 210 is in line-of-sight (LOS) positioning or non-line-of-sight (NLOS) positioning. As shown, Cases 1A, 2B, and 3B may include specific implementations in which the position or location of the UE 210 is determined directly by the output of the NN model used, while the output of the NN model used in Cases 1B, 2A, and 3A may be used to assist in determining the position or location of the UE 210 (e.g., by using the output in additional algorithms, operations, processes, etc.).

[0085] In some implementations, one or more of the techniques described herein may include removing Figure 7Additional or alternative scenarios beyond those shown. For example, the base station 222 may perform one or more functions of the UE 210; the AI-MF server 270 may perform one or more functions of the LMF 320; and so on. Additionally, in some implementations (such as in scenarios 1A, 1B, and / or 2A), NN model configuration and selection may be performed by the LMF 320 and / or the AI-MF server 270, and model inference may be performed by the UE 210. In such implementations, the LMF 320 and / or the AI-MF server 270 may transmit a validity tag to the UE 210, and the UE 210 may use the validity tag to select an appropriate NN model. In other implementations of scenarios 1A, 1B, and / or 2A, the UE 210 may be pre-configured with a validity tag (e.g., with internal policies, rules, and parameters) such as region, zone, condition, time, etc., for selecting an appropriate NN model.

[0086] In some implementations, such as in Case 2B and / or Case 3B, the UE 210 and / or the base station 222 may transmit assistance information (e.g., Doppler information) to the LMF 320 and / or the AI-MF server 270 to enable verification and / or selection of an appropriate model. In other implementations of Case 2B and / or Case 3B, the LMF 320 and / or the AI-MF server 270 may be pre-configured with internal algorithms, rules, and parameters for selecting an appropriate model in a given scenario or set of conditions. In some implementations, such as in Case 3A, the LMF 320 and / or the AI-MF server 270 may transmit validity tags to the base station 222, and the base station 222 may use these tags to select an appropriate NN model. In other implementations of Case 3A, the base station 222 may be pre-configured with internal algorithms, rules, and parameters for selecting an appropriate model in a given scenario or set of conditions.

[0087] Since any combination of functions 610, 620 and / or 630 can be performed by any combination of UE 210, base station 222, LMF 320 and AI-MF server 270, one or more of the examples described herein may refer to an “entity” that performs one or more of functions 610 to 630. Figure 8 For example, it involves a model configuration and selection entity 810 and a model inference entity 820; Figure 11 Involving a model configuration entity 1110 and a model selection and inference entity 1120; and Figure 13 Involved are a model configuration entity 1310 and a model selection entity 1320. An "entity" may be interpreted as a device or combination of devices that performs the corresponding functions 610, 620 and / or 630 as described herein.

[0088] Figure 8 is a diagram of an example process 800 for obtaining an effective NN model based on model validity data according to one or more implementations described herein. The process 800 may be performed by a model configuration and selection entity 810 and a model inference entity 820. The model configuration and selection entity 810 may include a system or device that performs the model configuration function 610 and the model selection function 620, while the model inference entity 820 may include a system or device that performs the model inference function 630, as described above with reference to Figure 6 Thus, each of the model configuration and selection entity 810 and the model inference entity 820 may include one or any combination of the UE 210, the base station 222, the AI-MF server 270 and / or the LMF 320.

[0089] Additionally, the example process 800 may include Figure 8 One or more fewer, additional, or differently ordered and / or arranged operations than those shown may be included. In some implementations, some or all of the operations of example process 800 may be performed independently, sequentially, simultaneously, etc., with one or more of the other operations of example process 800. Thus, the techniques described herein are not limited to Figure 8 The number, sequence, arrangement, timing, etc. of the operations or processes depicted. Indeed, the techniques described herein may include additional and alternative versions of the example process 800 .

[0090] As shown, process 800 may include a model configuration and selection entity 810 receiving assistance information from a model inference entity 820 (at 830). For example, LMF 320 or AI-MF server 270 may receive assistance information from UE 210 and / or base station 222. As described herein, assistance information may include one or more types of information that may be relevant to determining or inferring positioning (e.g., of UE 210) using a NN model. Examples of such information may include reference signal information, reference signal measurement information, and other types of positioning information (e.g., SRS, PRS, RSSI, signal-to-noise ratio, RSRP, RSRQ, signal TOA, RTT, CRS, TRS, Doppler information, etc.). In some implementations, the assistance information may include UE capability information and / or base station capability information.

[0091] The model configuration and selection entity 810 may determine a NN model based on the assistance information and the model validity tag (at 840). For example, the LMF 320 may determine an appropriate NN model for determining the location of the UE 210 based on the assistance information and the model validity tag. For example, the model configuration and selection entity 810 may include, store, or otherwise access one or more NN models to determine the location of the UE 210.

[0092] The conditions and / or circumstances for using each NN model may be based on or defined by a model validity tag associated with the NN model. Examples of validity tags may include scenario tags, time interval tags, geographic location tags, and / or network zone tags, among others. In some implementations, the validity tag may include one or more types of auxiliary information or combinations of auxiliary information, one or more values ​​or value ranges for one or more types of auxiliary information or combinations of auxiliary information, and more. The model configuration and selection entity 810 may determine which NN model is appropriate by comparing some or all of the auxiliary information with the model validity tag of the NN model. Upon determining that the auxiliary information satisfies, maps, or matches the model validity tag of a particular NN model, the model configuration and selection entity 810 may select the NN model.

[0093] The model configuration and selection entity 810 may communicate the selected NN model to the model inference entity 820 (at 850). For example, upon identifying or selecting an appropriate NN model for a given condition or scenario, the LMF 320 may communicate the NN model to the UE 210 and / or base station 222. In some implementations, the model configuration and selection entity 810 may communicate the actual NN model that was selected. In other implementations, the model configuration and selection entity 810 may communicate an NN identifier, and the model inference entity 820 may select an NN model based on the NN identifier and may communicate appropriate configuration information for the selected model. For example, one option is that if the inference entity selects a model, the inference entity may communicate information about the selection to the configuration entity, and the configuration entity may return model-specific configuration information to the inference entity.

[0094] The model inference entity 830 may receive the NN model and may use the NN model to generate positioning inference information based on the NN model (at 860). For example, the UE 210 and / or the base station 222 may receive the NN model from the LMF 320 and may use the NN model to generate positioning inference information. As described herein, the positioning inference information may include output information of the NN model. In some implementations, the positioning inference information may include an estimated location or an actual location of the UE 210. In other implementations, the positioning inference information may be used in conjunction with one or more other types of information in an additional or subsequent position determination algorithm or process. For example, the positioning inference information may be used as part of a conventional positioning algorithm.

[0095] Figure 9is a diagram of an example process 900 for determining a NN model based on auxiliary information and a model validity label according to one or more implementations described herein. The example process 900 may be performed by the model configuration and selection entity 810. Thus, the process 1200 may be performed by any one or a combination of the UE 210, the base station 222, the AI-MF server 270, and / or the LMF 320. Additionally, the example process 900 may include the following steps: Figure 9 One or more fewer, additional, different order, and / or arrangement of operations or data sets than those shown.

[0096] As shown, the model configuration and selection entity 810 can receive auxiliary information (at 9.1). The auxiliary information can include characteristics (e.g., C_1, C_2, ..., C_N (where N is greater than or equal to 3)). The model configuration and selection entity 810 can select and / or configure a NN model with one or more validity tags (at 9.2). The NN model and / or validity tags can be determined based on the auxiliary information characteristics and one or more preconfigured network or device policies, rules and / or parameters to enable the inference entity 820 to implement the NN model. The NN model can include an NN identifier (e.g., NN_1) and one or more validity tags (e.g., VL_1.1, VL_2, ..., VL_1.N (where N is greater than or equal to 3)), which can enable the inference entity 820 to determine which NN model to apply under the conditions described by the validity tags. The model configuration and selection entity 810 can transmit the NN model to the inference entity 820 (at 9.3).

[0097] Figure 10 1 is a diagram of an example table 1000 of validity labels for NN models according to one or more implementations described herein. Example table 1000 may include a data structure used by UE 210, base station 222, AI-MF server 270, and / or LMF 320 in one or more of the processes or techniques described herein. As shown, columns of example table 1000 may include NN identifiers NN_1, NN_2, NN_3, ..., NN_Q (where Q is greater than or equal to 4). Rows of example table 1000 may include validity labels (VL) associated with NN model identifiers. For example, NN_1 may include or be associated with VL_1.1, VL_1.2, ..., VL_1.N (where Q is greater than or equal to 3).

[0098] A validity tag may include specific characteristics or conditions for using the corresponding NN model. In general, a set of validity tags may include combined characteristics or conditions (e.g., scenarios) for using the corresponding NN model. In some implementations, the NN model can be verified by satisfying one validity tag, a threshold number (e.g., two or more) of validity tags, or all validity tags for a particular NN model. The conditions for validating the NN model (e.g., how many validity tags must be satisfied as a prerequisite for selection and use) can be the same for all NN models, specific to the NN class of the NN model, or specific to the NN model (e.g., regardless of the NN class). In some implementations, the NN model can be verified and selected based on which NN model is best satisfied by the auxiliary information.

[0099] Figure 11 is a diagram of an example process 1100 for selecting a NN model based on a model validity test according to one or more implementations described herein. The process 1100 may be performed by a model configuration entity 1110 and a model selection entity 1120. Each or both of the model configuration entity 1110 and the model selection entity 1120 may include any combination of the UE 210, the base station 222, the AI-MF server 270, and / or the LMF 320. The model configuration entity 1110 may include a system or device that performs the model configuration function 610, and the model selection entity 1120 may include a system or device that performs the model selection function 620, as described above with reference to FIG. Figure 6 In some implementations, the model configuration entity 1110 or the model selection entity 1120 may also perform the operations of the model inference function 630. In some implementations, the operations of the model inference function 630 may be performed by different entities.

[0100] Additionally, the example process 1100 may include Figure 11 One or more fewer, additional, or differently ordered and / or arranged operations than those shown may be included. In some implementations, some or all of the operations of example process 1100 may be performed independently, sequentially, simultaneously, etc., with one or more of the other operations of example process 1100. Thus, the techniques described herein are not limited to Figure 11 The number, sequence, arrangement, timing, etc. of the operations or processes depicted. Indeed, the techniques described herein may include additional and alternative versions of the example process 1100 .

[0101] As shown, the model selection entity 1120 may communicate model capability data for one or more NN models to the model configuration entity 1110 (at 1130). For example, the UE 210 may communicate the model capability data to the LMF 320. The model capability data may include information describing or indicating the capabilities of the NN models that the model selection entity 1120 has selected or otherwise used to determine the location of the UE 210.

[0102] The model configuration entity 1110 may transmit the model validity data to the model selection entity 1120 (at 1140). For example, the LMF 320 may transmit the model validity data to the UE 210. As described herein, the model validity data may include information used to determine or confirm whether one or more NN models are suitable for a given scenario condition or scenario. The model validity data may include model input data, model performance requirements, model condition requirements, and / or one or more other types of information. The model validity data may also be referred to herein as model test data, validity test data, etc. The model validity data may include a model validity tag as described above with reference to the previous figures.

[0103] The model selection entity 1120 may determine model validity based on the model validity data (at 1150). For example, the UE 210 may determine model validity based on the model validity data received from the LMF 320. Model validity may mean that the NN model has been determined to be appropriate or valid for a given scenario or condition. In some implementations, the model selection entity 1120 may determine model validity by applying input data to the NN model to generate output data and confirming whether the output data is consistent with model performance requirements (e.g., within a threshold accuracy, latency, etc.) based on the received model validity data. The model input data may be derived from the model validity data received from the model configuration entity 1110 and / or based on information measured or stored by the model selection entity 1120. The model selection entity 1120 may also or alternatively determine model validity by confirming that one or more other types of conditions or requirements specified by the model validity data are met. In some implementations, the model selection entity 1120 may select a NN model for use (at 1160) if the model is determined to be valid. In other implementations, the model selection entity 1120 may mark or flag the NN model as valid for later use.

[0104] The model selection entity 1120 may provide a model validity response (1170) to the model configuration entity 1110. For example, the UE 210 may generate a model validity response and communicate it to the LMF 320. The model validity response may include an indication of a result of determining or testing the NN model based on the model validity data, which may include an indication of whether the tested NN model is valid or invalid. In some implementations, such as when the model selection entity 1120 tests multiple NN models, the model validity response information may include a performance metric or validity metric for each NN model, and the model configuration entity 1110 may select a NN model based on the metric, and indicate and configure the selected NN model for use by the model selection entity 1120 (at 1180).

[0105] Figure 12 is a diagram of an example of a process 1200 for model selection according to one or more implementations described herein. The example process 1200 may be performed by the model selection entity 1110 (not shown). Thus, the process 1200 may be performed by any one or a combination of the UE 210, the base station 222, the AI-MF server 270, and / or the LMF 320. Additionally, the example process 1200 may include Figure 12 One or more fewer, additional, different order, and / or arrangement of operations or data sets than those shown.

[0106] The model selection entity 1110 may receive validity data and test one or more NN models based on the validity data (at 12.1). As shown, the model selection entity 1110 may receive validity data, which is represented as VD_2.1, VD_2.2, VD_3.1, VD_1.S, VD_3.2, and VD_3.3. The validity data may include information for determining whether one or more NN models are suitable for use. The validity data may include NN model inputs to be used for testing, acceptable NN model outputs, and / or other conditions or scenarios for using the NN model. In some embodiments, the model selection entity 1110 may test the validity of the NN model based solely on the validity data received from the model configuration entity 1110 (e.g., without using validity tags).

[0107] In other specific implementations, since the validity tag can be used to indicate when the NN model can be used, the NN model can be valid when the conditions indicated by the validity tag of the NN model are met. Example 12 includes NN models represented as NN_1, NN_2, NN_3, ..., NN-Q (where Q is greater than or equal to 1). Each NN model includes a validity tag. For example, NN_1 includes VL_1.1, VL_1.2, and VL_1.3. The model selection entity 1110 can compare the validity data and / or the results of applying the validity test data to the validity tags of one or more NN models to determine which NN model is valid to use.

[0108] As shown, when the test data is applied to some of the validity labels of some of the NN models, the model selection entity 1110 may select NN_3 (at 12.2) because NN_3 is the only NN model with the conditions fully satisfied by the test data. In some implementations, when all the validity data correspond to the validity labels of the NN models, the NN model may be verified and selected. After selecting NN_3, the model selection entity 1110 may transmit a notification or indication of the selection to the model configuration entity 1110 (at 12.3).

[0109] Figure 13 1 is a diagram of an example process 1300 for selecting a NN model based on a model class according to one or more implementations described herein. The process 1300 may be performed by a model configuration entity 1310 and a model selection entity 1320. Each or both of the model configuration entity 1310 and the model selection entity 1320 may include any combination of the UE 210, the base station 222, the AI-MF server 270, and / or the LMF 320. The model configuration entity 1310 may include a system or device that performs the model configuration function 610, and the model selection entity 1320 may include a system or device that performs the model selection function 620, as described above with reference to FIG. Figure 6 In some implementations, the model configuration entity 1310 or the model selection entity 1320 may also perform the operations of the model inference function 630. In some implementations, the operations of the model inference function 630 may be performed by different entities.

[0110] Additionally, the example process 1300 may include Figure 13 One or more fewer, additional, or differently ordered and / or arranged operations than those shown. In some implementations, some or all of the operations of example process 1300 may be performed independently, sequentially, simultaneously, etc., with one or more of the other operations of example process 1300. Thus, the techniques described herein are not limited to Figure 13The number, sequence, arrangement, timing, etc. of the operations or processes depicted. Indeed, the techniques described herein may include additional and alternative versions of the example process 1300 .

[0111] As shown, the model configuration entity 1310 may provide model validity data for a plurality of NN model classes to the model selection entity 1320 (at 1330). As described herein, the NN model class may include a type or category of NN models. For example, the LMF 320 may provide an indication to the UE 210 of an NN model class that the LMF 320 may use to monitor the location of the UE 210 (at 1330). The LMF 320 may also transmit validity data (e.g., validity test data) for each model of the NN class to the UE 210. The NN models of the NN model class may be associated with each other in one or more ways, including an NN model class identifier, NN model functionality, NN model tags, NN model requirements, and / or a type of validity data set.

[0112] The model selection entity 1320 may receive the validity data and select a NN model for each NN model class (at 1340). For example, the UE 210 may receive the validity data, determine the number of validity data sets contained therein, determine the NN model class associated with each validity data set, and map or determine one or more NN models for each validity data set. The NN model classes and the identified NN models may each correspond to a different type or scenario of AI-based positioning. In some implementations, the model selection entity 1320 may confirm or verify the validity of each NN model before selecting it. As described above, doing so may include applying an appropriate validity data set to an NN model of an appropriate type or class. Additionally or alternatively, the model selection entity 1320 may select a NN model based on a one-to-many mapping or a many-to-one mapping between model validity data sets and NN model classes.

[0113] The model selection entity 1320 may communicate the selected model to the model configuration entity 1310 (at 1350). By doing so, the model configuration entity 1310 (e.g., LMF 320) may know the NN model that models the selection entity 1320 (e.g., UE 210). Thus, the model configuration entity 1310 may periodically cause the model selection entity 1310 or a model inference entity (not shown) to switch between NN models. In some implementations, the model selection entity 1320 may indicate the selected NN model by providing a mapping of a validity dataset or NN model class to the selected NN model. In such implementations, the mapping may include a one-to-many mapping or a many-to-one mapping between model validity datasets and NN model classes. The mapping may also include indications of which NN models are to be ranked or prioritized over other NN models, which NN models are to be used under certain circumstances or conditions, and the like. Thus, the model configuration entity 1310 may periodically (eg, as time, conditions, or scenarios change) cause the model selection entity 1320 to switch from one of the selected NN models to another NN model (at 1360).

[0114] Figure 14 is a diagram of an example data structure 1400 of NN model classes and validity labels according to one or more specific implementations described herein. As shown, the example data structure 1400 may include data structure 1410, data structure 1420, and data structure 1430. Data structure 1410 may include NN models arranged by NN model class. For example, the NN model class corresponding to the NN model class identifier NN_CLASS_C1 may include NN models corresponding to NN model identifiers C1_NN_1, C1_NN_2, ..., C1_NN_N (where N is greater than or equal to 2). Other NN model classes and corresponding NN models are similarly represented.

[0115] Data structure 1420 may include a validity tag for each NN model of the NN model class NN_CLASS_C1. For example, the NN model of the NN model identifier C1_NN_1 may be associated with validity tags VL_1.1, VL_1.2, ..., VL_1.R (where R is greater than or equal to 3). The NN model identifiers C1_NN_2 and C1_NN_N are also represented by corresponding validity tags. Similarly, data structure 1430 may include a validity tag for each NN model of the NN model class NN_CLASS_C2. For example, the NN model of the NN model identifier C2_NN_1 may be associated with validity tags VL_1.1, VL_1.2, ..., VL_1.U (where U is greater than or equal to 3). The NN models of identifiers C2_NN_2 and C2_NN_N are similarly represented by validity tags. Thus, the techniques described herein may include a model configuration entity 1310 , a model selection entity 1320 , and / or a model inference entity 1320 that organize and maintain NN model classes, NN models, and validity labels for NN models.

[0116] Figure 15 1 is a diagram of an example process 1500 for selecting an NN model based on validity data received for an NN model class according to one or more implementations described herein. As shown, the example process 1500 may include validity data 1510 arranged according to NN model classes represented by NN model identifiers NN_CLASS_C1, NN_CLASS_C2, etc. In addition, each NN model class in the validity data 1510 may be associated with one or more types of validity data (VD) represented by VD_1.1, VD_1.2, VD_1.R, etc.

[0117] The model selection entity 1320 (not shown) can receive the validity data 1510 and select one or more NN models based on the validity data 1510. For example, the model selection entity 1320 can compare the NN model class of the validity data 1510 with NN models of different NN model classes 1520 and NN model classes 1530 or map the NN model class of the validity data 1510 to NN models of different NN model classes 1520 and NN model classes 1530 (at 15.1). More specifically, the model selection entity 1320 can compare the type of validity data (VD) of a given NN model class with the validity labels of the NN models of each class or map the type of validity data (VD) of a given NN model class to the validity labels of the NN models of each class. As shown, the model selection entity 1320 can determine that the validity data received for NN_CLASS_C1 corresponds to the validity label of NN_CLASS_C1 (at 15.2), and therefore can select the NN model of C1_NN_1 for AI-based positioning purposes (at 15.3). In contrast, the model selection entity 1320 may determine that the validity data received for NN_CLASS_C2 does not correspond to the validity tag of NN_CLASS_C2 (at 15.4), and therefore the NN model of C1_NN_2 may not be selected for AI-based positioning purposes.

[0118] Figure 16 1 is a diagram of an example process 1600 for selecting a plurality of NN models based on validity data received for an NN model class according to one or more implementations described herein. As shown, the example process 1600 may include validity data 1610 arranged according to NN model classes represented by NN model identifiers NN_CLASS_C1, NN_CLASS_C2, etc. In addition, each NN model class in the validity data 1610 may be associated with one or more types of validity data (VD) represented by VD_1.1, VD_1.2, VD_1.R, etc.

[0119] The model selection entity 1320 (not shown) may receive the validity data 1610 and select one or more NN models based on the validity data 1610. For example, the model selection entity 1320 may compare the NN model class of the validity data 1610 with NN models of a different NN model class 1620 or map the NN model class of the validity data 1610 to NN models of a different NN model class 1620 (at 16.1). More specifically, the model selection entity 1320 may compare the type of validity data (VD) of a given NN model class with the validity labels of the NN models of each class or map the type of validity data (VD) of a given NN model class to the validity labels of the NN models of each class. As shown, the model selection entity 1320 may determine that the validity data received for NN_CLASS_C1 corresponds to the validity label of NN_CLASS_C1 (at 16.2), and may therefore select the NN model of C1_NN_1 for AI-based positioning purposes (at 16.3).

[0120] The model selection entity 1320 may compare the NN model class of the validity data 1610 with the NN models of the other NN model classes 1630 or map the NN model class of the validity data 1610 to the NN models of the other NN model classes 1630 (at 16.4). As shown, the model selection entity 1320 may determine that the validity data received for NN_CLASS_C2 corresponds to the validity label of NN_CLASS_C2 (at 16.5), and therefore may select the NN model of C1_NN_2 for AI-based positioning purposes (at 16.6). Thus, validity data of multiple NN model classes may be mapped to multiple NN validity models and selected for use.

[0121] Figure 17 1 is a diagram of an example process 1700 for model validation according to one or more implementations described herein. As shown, process 1700 is described below as being performed by UE 210 and LMF 320. However, one or more other devices may perform some or all of process 1700. For example, in some implementations, process 1700 may include base station 222 instead of UE 210, and LMF 320 may additionally or alternatively be replaced by AI-MF server 270. Additionally, example process 1700 may include Figure 17 In some implementations, some or all of the operations of example process 1700 may be performed independently, sequentially, concurrently, etc., with one or more of the other operations of example process 1700.

[0122] As shown, process 1700 may include transmitting model capability data (block 1710). For example, UE 210 or base station 222 may transmit model capability data to LMF 320 and / or AI-MF server 270. The model capability data may include information describing the ability of an inference entity (e.g., UE 210 or base station 222) to use one or more types of NN models to determine the location of UE 210. The model capability data may include the number of NN models and information about the characteristics of each NN model, such as target accuracy, latency, complexity, size, etc. The target accuracy may refer to how accurate the NN model is in determining the location of UE 210 and may be expressed as a distribution of errors or a mean-variance between the inferred location of UE 210 and the actual location.

[0123] The latency may refer to a model inference latency, which may be the amount of time typically involved in using an NN model to determine the location of the UE 210. The model capability data may also include a signaling type for each NN model (such as a reference signal type or configuration) and an indication of the training of each NN model (e.g., whether each NN model is already trained, trained offline, trained online, etc.). The model capability data may also include a model type for each NN model. The model type may indicate one or more characteristics, such as the input data that can be provided to the NN model, the resource intensity of the NN model, and the type and amount of output data generated by the NN model.

[0124] Process 1700 may include transmitting model condition and requirement data (block 1720). For example, UE 210 may transmit the model condition and requirement data to LMF 320. The model and requirement data may include conditions that UE 210 may support when operating the NN models. Examples of model condition data may include auxiliary signaling received by UE 210, reference signaling configurations supported by UE 210, and the type of training required for each NN model (e.g., whether the model is trained online or offline). Additional examples of model and requirement data may include information such as the type, quantity, quality, and frequency of NN model input data, one or more validity labels associated with each NN model, and an indication (e.g., by LMF 320) of whether each model is supervised, partially supervised, or unsupervised.

[0125] Process 1700 may include determining model configuration and validity data (block 1730). For example, the LMF 320 may determine the model configuration and validity data of the NN model indicated by the UE 210. The LMF 320 may do so by matching the model capability data and the model condition and requirement data with a NN model data set stored locally or otherwise available to the LMF 320. In some implementations, the LMF 320 may have a data repository that includes information about NN models configured to assist in determining the location of the UE 210. Each NN model may be associated with configuration information indicating various characteristics about the NN model. Examples of such information may include a model identifier (ID), an index of the capabilities of each NN model, and reference signal configurations supported by each NN model. Additional examples of such information may include whether the NN model is trained online or offline, a schedule for updating the NN model with additional training data, and assistance data that each NN model may use.

[0126] Process 1700 may include transmitting model configuration and validation data (block 1740). For example, LMF 320 may transmit the model configuration and validation data to UE 210. The model configuration and validation data may indicate one or more NN models to be tested by UE 210, configuration parameters for configuring each NN model, and validation data for testing each NN model. The NN models, configuration parameters, and validation data may be selected to enable UE 210 to run the NN models under test conditions to confirm or verify that the NN models operate as expected. Thus, the NN models, configuration parameters, and validation data may correspond to different conditions or scenarios in which the location of UE 210 may be determined (or facilitated) by the appropriate NN model.

[0127] Process 1700 may include testing NN models based on configuration and test data (at 1750). For example, UE 210 may test the NN models indicated by LMF 320 based on the configuration and test data provided by LMF 320. By evaluating whether the NN models operate as expected and produce acceptable output, UE 210 may determine whether each NN model is valid (i.e., is appropriate for the set of conditions specifying the NN model) and select a valid NN model. In some implementations, UE 210 may alternatively provide feedback information to LMF 320. Process 1700 may include transmitting the feedback information to LMF 320 (block 1760). For example, UE 210 may report the results of the validity test to LMF 320. In response, LMF 320 may determine which models are valid and select a valid NN model. Process 1700 may continue by LMF 320 providing an indication of the valid NN models to UE 210 (block 1770), which may cause UE 210 to select the indicated NN model for use.

[0128] Figure 18 is a diagram of an example process 1800 for model validation according to one or more implementations described herein. As shown, process 1800 is described below as being performed by UE 210 and LMF 320. However, one or more other devices may perform some or all of process 1800. For example, in some implementations, process 1800 may include base station 222 instead of UE 210, and additionally or alternatively, LMF 320 may be replaced by AI-MF server 270. Additionally, example process 1800 may include Figure 18 In some implementations, some or all of the operations of example process 1800 may be performed independently, sequentially, concurrently, etc., with one or more of the other operations of example process 1800.

[0129] As shown, process 1800 may include UE 210 providing assistance information and model capability information to LMF 320 (at 1810). Assistance information and model capability data are described above. In some implementations, the assistance information may be current assistance information or assistance information indicating a condition under which UE 210 will use a NN model to determine the location of UE 210. Process 1800 may also include LMF 320 determining a NN model configuration and a model tag based on the assistance information and the model capability information (at 1820). For example, LMF 320 may determine a condition of UE 210 based on the assistance information and select an appropriate NN model based on the current condition and the model capability data. LMF 320 may also determine an appropriate validity tag for UE 210 to use the NN model based on the assistance information, the model capability data, and / or the selected NN model. Process 1800 may include LMF 320 communicating the model configuration and validity tag to UE 210 at 1830, and UE 210 may respond to LMF 320 with a confirmation message at 1840. Although not shown, UE 210 may monitor conditions for using the NN model and use the NN model when the monitored conditions are met.

[0130] Figure 19 1900 is a diagram illustrating an example of components of a device according to one or more implementations described herein. In some implementations, device 1900 may include at least application circuitry 1902, baseband circuitry 1904, RF circuitry 1906, front-end module (FEM) circuitry 1908, one or more antennas 1910, and power management circuitry (PMC) 1912, coupled together as shown. The components of the illustrated device 1900 may be included in a UE or a RAN node. In some implementations, device 1900 may include fewer components (e.g., a RAN node may not utilize application circuitry 1902 but instead include a processor / controller to process IP data received from a CN or Evolved Packet Core (EPC)). In some implementations, device 1900 may include additional components, such as memory / storage, a display, a camera, sensors (including one or more temperature sensors, such as a single temperature sensor or multiple temperature sensors at different locations within device 1900), or input / output (I / O) interfaces. In other implementations, the components described below may be included in more than one device (e.g., the circuitry may be separately included in more than one device for a Cloud-RAN (C-RAN) implementation).

[0131] The application circuitry 1902 may include one or more application processors. For example, the application circuitry 1902 may include circuitry such as, but not limited to, one or more single-core processors or multi-core processors. The processors may include any combination of general-purpose processors and specialized processors (e.g., graphics processors, application processors, etc.). The processors may be coupled to or include a memory / storage device and may be configured to execute instructions stored in the memory / storage device to enable various applications or operating systems to run on the device 1900. In some implementations, the processors of the application circuitry 1902 may process IP data packets received from the EPC.

[0132] The baseband circuitry 1904 may include circuitry such as, but not limited to, one or more single-core processors or multi-core processors. The baseband circuitry 1904 may include one or more baseband processors or control logic components to process baseband signals received from the receive signal path of the RF circuitry 1906 and generate baseband signals for the transmit signal path of the RF circuitry 1906. The baseband circuitry 1904 may interact with the application circuitry 1902 to generate and process baseband signals and control the operation of the RF circuitry 1906. For example, in some implementations, the baseband circuitry 1904 may include a 3G baseband processor 1904A, a 4G baseband processor 1904B, a 5G baseband processor 1904C, or other baseband processors 1904D of other current, developing, or future generations (e.g., 5G, 6G, etc.).

[0133] Baseband circuitry 1904 (e.g., one or more of baseband processors 1904A through 1904D) may handle various radio control functions that enable communication with one or more radio networks via RF circuitry 1906. In other implementations, some or all of the functions of baseband processors 1904A-D may be included in modules stored in memory 1904G and executed via central processing unit (CPU) 1904E. Radio control functions may include, but are not limited to, signal modulation / demodulation, encoding / decoding, radio frequency shifting, and the like. In some implementations, the modulation / demodulation circuitry of baseband circuitry 1904 may include fast Fourier transform (FFT), precoding, or constellation mapping / demapping functions. In some implementations, the encoding / decoding circuitry of baseband circuitry 1904 may include convolution, tail-biting, turbo, Viterbi, or low-density parity check (LDPC) encoder / decoder functions. The implementation of the modulation / demodulation and encoder / decoder functions is not limited to these examples and, in other aspects, may include other suitable functions.

[0134] In some specific implementations, the memory 1904G may receive and / or store information and instructions for determining the location of the UE 210 using the NN model. The output of the NN may be the location of the UE 210, or may be used to determine the location of the UE in a subsequent process. The information and instructions may enable the NN model to be configured, marked, and verified for use in one or more scenarios. One or more of the UE 210, the base station 222, the AI-MF server 270, and the LMF 320 may be involved, and the processes described herein may include model configuration functionality, model selection functionality, and model inference functionality.

[0135] In some implementations, the baseband circuitry 1904 may include one or more audio digital signal processors (DSPs) 1904F. The audio DSPs 1904F may include components for compression / decompression and echo cancellation, and in other implementations may include other suitable processing elements. In some implementations, the components of the baseband circuitry may be appropriately combined in a single chip, a single chipset, or provided on the same circuit board. In some implementations, some or all of the components of the baseband circuitry 1904 and the application circuitry 1902 may be implemented together, such as on a system on a chip (SOC).

[0136] In some implementations, the baseband circuitry 1904 can provide communications compatible with one or more radio technologies. For example, in some implementations, the baseband circuitry 1904 can support communications with NG-RAN, Evolved Universal Terrestrial Radio Access Network (EUTRAN), or other wireless metropolitan area network (WMAN), wireless local area network (WLAN), wireless personal area network (WPAN), etc. Implementations in which the baseband circuitry 1904 is configured to support radio communications of more than one wireless protocol can be referred to as multi-mode baseband circuitry.

[0137] RF circuitry 1906 can communicate with a wireless network using modulated electromagnetic radiation through a non-solid medium. In various implementations, RF circuitry 1906 can include switches, filters, amplifiers, and the like to facilitate communication with the wireless network. RF circuitry 1906 can include a receive signal path that can include circuitry for down-converting RF signals received from FEM circuitry 1908 and providing a baseband signal to baseband circuitry 1904. RF circuitry 1906 can also include a transmit signal path that can include circuitry for up-converting baseband signals provided by baseband circuitry 1904 and providing an RF output signal to FEM circuitry 1908 for transmission.

[0138] In some implementations, the receive signal path of RF circuitry 1906 may include mixer circuitry 1906A, amplifier circuitry 1906B, and filter circuitry 1906C. In some implementations, the transmit signal path of RF circuitry 1906 may include filter circuitry 1906C and mixer circuitry 1906A. RF circuitry 1906 may also include synthesizer circuitry 1906D for synthesizing frequencies for use by mixer circuitry 1906A in the receive and transmit signal paths. In some implementations, mixer circuitry 1906A in the receive signal path may be configured to downconvert the RF signal received from FEM circuitry 1908 based on the synthesized frequency provided by synthesizer circuitry 1906D. Amplifier circuitry 1906B may be configured to amplify the downconverted signal, and filter circuitry 1906C may be a low-pass filter (LPF) or a band-pass filter (BPF) configured to remove unwanted signals from the downconverted signal to generate an output baseband signal. The output baseband signal can be provided to baseband circuitry 1904 for further processing. In some implementations, the output baseband signal can be a zero-frequency baseband signal, but this is not required. In some implementations, the mixer circuitry 1906A of the receive signal path can include a passive mixer, but the scope of the implementation is not limited in this respect.

[0139] In some implementations, the mixer circuit 1906A of the transmit signal path can be configured to upconvert an input baseband signal based on a synthesized frequency provided by the synthesizer circuit 1906D to generate an RF output signal for the FEM circuit 1908. The baseband signal can be provided by the baseband circuit 1904 and can be filtered by the filter circuit 1906C.

[0140] In some implementations, the mixer circuit 1906A of the receive signal path and the mixer circuit 1906A of the transmit signal path can include two or more mixers and can be arranged for quadrature down-conversion and up-conversion, respectively. In some implementations, the mixer circuit 1906A of the receive signal path and the mixer circuit 1906A of the transmit signal path can include two or more mixers and can be arranged for image rejection (e.g., Hartley image rejection). In some implementations, the mixer circuit 1906A of the receive signal path and the mixer circuit 1406A can be arranged for direct down-conversion and direct up-conversion, respectively. In some implementations, the mixer circuit 1906A of the receive signal path and the mixer circuit 1906A of the transmit signal path can be configured for superheterodyne operation.

[0141] In some implementations, the output baseband signal and the input baseband signal can be analog baseband signals, although the scope of the implementation is not limited in this respect. In some alternative implementations, the output baseband signal and the input baseband signal can be digital baseband signals. In these alternative implementations, RF circuitry 1906 can include analog-to-digital converter (ADC) circuitry and digital-to-analog converter (DAC) circuitry, and baseband circuitry 1904 can include a digital baseband interface to communicate with RF circuitry 1906.

[0142] In some dual-mode implementations, separate radio IC circuitry may be provided to process signals for each spectrum, although the scope of the implementations is not limited in this respect.

[0143] In some implementations, synthesizer circuit 1906D can be a fractional-N synthesizer or a fractional N / N+1 synthesizer, but the scope of the implementation is not limited in this respect, as other types of frequency synthesizers may also be suitable. For example, synthesizer circuit 1906D can be a delta-sigma synthesizer, a frequency multiplier, or a synthesizer including a phase-locked loop with a frequency divider.

[0144] The synthesizer circuit 1906D can be configured to synthesize an output frequency based on the frequency input and the divider control input for use by the mixer circuit 1906A of the RF circuit 1906. In some implementations, the synthesizer circuit 1906D can be a fractional-N / N+1 synthesizer.

[0145] In some implementations, the frequency input can be provided by a voltage-controlled oscillator (VCO), although this is not required. The divider control input can be provided by the baseband circuit 1904 or the application circuit 1902 depending on the desired output frequency. In some implementations, the divider control input (e.g., N) can be determined from a lookup table based on the channel indicated by the application circuit 1902.

[0146] The synthesizer circuit 1906D of the RF circuit 1906 may include a frequency divider, a delay-locked loop (DLL), a multiplexer, and a phase accumulator. In some implementations, the frequency divider may be a dual-modulus divider (DMD), and the phase accumulator may be a digital phase accumulator (DPA). In some implementations, the DMD may be configured to divide the input signal by N or N+1 (e.g., based on a carry-out) to provide a fractional division ratio. In some example implementations, the DLL may include a cascaded, tunable delay element, a phase detector, a charge pump, and a set of D-type flip-flops. In these implementations, the delay elements may be configured to divide the VCO cycle into Nd equal phase groups, where Nd is the number of delay elements in the delay line. In this way, the DLL provides negative feedback to help ensure that the total delay through the delay line is one VCO cycle.

[0147] In some implementations, the synthesizer circuit 1906D can be configured to generate a carrier frequency as the output frequency, while in other implementations, the output frequency can be a multiple of the carrier frequency (e.g., twice the carrier frequency, four times the carrier frequency) and can be used with a quadrature generator and divider circuit to generate multiple signals at the carrier frequency with multiple different phases relative to each other. In some implementations, the output frequency can be the LO frequency (fLO). In some implementations, the RF circuit 1906 can include an IQ / polarity converter.

[0148] The FEM circuitry 1908 may include a receive signal path that may include circuitry configured to operate on RF signals received from one or more antennas 1910, amplify the receive signal, and provide an amplified version of the receive signal to the RF circuitry 1906 for further processing. The FEM circuitry 1908 may also include a transmit signal path that may include circuitry configured to amplify transmit signals provided by the RF circuitry 1906 for transmission via one or more of the one or more antennas 1910. In various implementations, amplification by either the transmit signal path or the receive signal path may be performed only in the RF circuitry 1906, only in the FEM circuitry 1908, or in both the RF circuitry 1906 and the FEM circuitry 1908.

[0149] In some implementations, the FEM circuitry 1908 may include a TX / RX switch to switch between transmit and receive modes of operation. The FEM circuitry may include a receive signal path and a transmit signal path. The receive signal path of the FEM circuitry may include an LNA to amplify a received RF signal and provide the amplified received RF signal as an output (e.g., to the RF circuitry 1906). The transmit signal path of the FEM circuitry 1908 may include a power amplifier (PA) to amplify an input RF signal (e.g., provided by the RF circuitry 1906) and one or more filters to generate an RF signal for subsequent transmission (e.g., via one or more of the one or more antennas 1910).

[0150] In some implementations, the PMC 1912 can manage the power provided to the baseband circuitry 1904. Specifically, the PMC 1912 can control power source selection, voltage scaling, battery charging, or DC-DC conversion. When the device 1900 is capable of being powered by a battery, such as when the device is included in a UE, the PMC 1912 is typically included. The PMC 1912 can improve power conversion efficiency while providing a desired implementation size and heat dissipation characteristics.

[0151] Although Figure 19 The PMC 1912 is shown coupled only to the baseband circuit 1904. However, in other implementations, the PMC 1912 may additionally or alternatively be coupled to other components (such as, but not limited to, the application circuit 1902, the RF circuit 1906, or the FEM 1908) and perform similar power management operations.

[0152] In some implementations, the PMC 1912 can control or otherwise be part of various power saving mechanisms of the device 1900. For example, if the device 1900 is in the RRC_Connected state, in which the device remains connected to the RAN node because it expects to receive traffic soon, then after a period of inactivity, the device can enter a state known as discontinuous reception mode (DRX). During this state, the device 1900 can be powered off for short intervals to save power.

[0153] If there is no data traffic activity for an extended period of time, the device 1900 may transition to the RRC_Idle state, in which the device is disconnected from the network and does not perform operations such as channel quality feedback, handover, etc. The device 1900 enters a very low power state and performs paging, in which the device periodically wakes up again to listen to the network and then powers down again. The device 1900 may not receive data in this state; to receive data, the device may transition back to the RRC_Connected state.

[0154] An additional power saving mode can prevent a device from using the network for a period exceeding the paging interval (ranging from a few seconds to several hours). During this period, the device cannot connect to the network and can be completely powered down. Any data transmitted during this period will incur significant latency, assuming that latency is acceptable.

[0155] The processor of application circuitry 1902 and the processor of baseband circuitry 1904 can be used to execute elements of one or more instances of a protocol stack. For example, the processor of baseband circuitry 1904 can be used alone or in combination to perform layer 3, layer 2, or layer 1 functions, while the processor of baseband circuitry 1904 can utilize data received from these layers (e.g., packet data) and further perform layer 4 functions (e.g., transmit communication protocol (TCP) and user datagram protocol (UDP) layers). As mentioned herein, layer 3 may include an RRC layer, which will be described in further detail below. As mentioned herein, layer 2 may include a medium access control (MAC) layer, a radio link control (RLC) layer, and a packet data convergence protocol (PDCP) layer, which will be described in further detail below. As mentioned herein, layer 1 may include a physical (PHY) layer of a UE / RAN node, which will be described in further detail below.

[0156] Figure 20 is a block diagram illustrating components that can read instructions from a machine-readable medium or computer-readable medium (e.g., a non-transitory machine-readable storage medium) and perform any one or more of the methods discussed herein, according to some example implementations. Specifically, Figure 20 A schematic diagram of hardware resources 2000 is shown, including one or more processors (or processor cores) 2010, one or more memory / storage devices 2020, and one or more communication resources 2030, each of which may be communicatively coupled via a bus 2040. For implementations in which node virtualization (e.g., NFV) is utilized, a hypervisor 2002 may be executed to provide an execution environment for one or more network slices / subslices to utilize the hardware resources 2000.

[0157] Processor 2010 (e.g., a central processing unit (CPU), a reduced instruction set computing (RISC) processor, a complex instruction set computing (CISC) processor, a graphics processing unit (GPU), a digital signal processor (DSP) (such as a baseband processor), an application-specific integrated circuit (ASIC), a radio frequency integrated circuit (RFIC), another processor, or any suitable combination thereof) may include, for example, processor 2012 and processor 2014.

[0158] The memory / storage device 2020 may include main memory, disk storage, or any suitable combination thereof. The memory / storage device 2020 may include, but is not limited to, any type of volatile or non-volatile memory, such as dynamic random access memory (DRAM), static random access memory (SRAM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), flash memory, solid-state storage, etc.

[0159] In some implementations, the memory / storage device 2020 receives and / or stores information and instructions 2055 for using a NN model to determine the location of the UE 210. The output of the NN can be the location of the UE 210, or can be used in a subsequent process to determine the location of the UE. The information and instructions 2055 can enable the NN model to be configured, labeled, and validated for use in one or more scenarios. One or more of the UE 210, the base station 222, the AI-MF server 270, and the LMF 320 can be involved, and the processes described herein can include model configuration functionality, model selection functionality, and model inference functionality.

[0160] The communication resources 2030 may include an interconnect component or a network interface component or other suitable device to communicate with one or more peripheral devices 2004 or one or more databases 2006 via the network 2008. For example, the communication resources 2030 may include a wired communication component (e.g., for coupling via a universal serial bus (USB)), a cellular communication component, an NFC component, Components (e.g. Low power consumption), components and other communication components.

[0161] The instructions 2050 may include software, programs, applications, applet programs, apps, or other executable code for causing at least one of the processors 2010 to perform any one or more of the methodologies discussed herein. The instructions 2050 may reside entirely or partially within at least one of the processors 2010 (e.g., within a cache memory of a processor), the memory / storage device 2020, or any suitable combination thereof. In addition, any portion of the instructions 2050 may be transferred to the hardware resource 2000 from any combination of the peripheral device 2004 or the database 2006. Thus, the memory of the processor 2010, the memory / storage device 2020, the peripheral device 2004, and the database 2006 are examples of computer-readable and machine-readable media.

[0162] The embodiments herein may include subject matter such as a method, components for performing the actions or blocks of the method, and at least one machine-readable medium comprising executable instructions that, when executed by a machine (e.g., a processor with memory, an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), etc.), cause the machine to perform the actions of a method or apparatus or system for concurrent communication using multiple communication technologies according to the described implementations and embodiments.

[0163] In embodiment 1, which may also include one or more of the embodiments described herein, a device includes: a memory; and a processor, wherein the processor is configured to cause the device, when executing instructions stored in the memory, to: obtain a neural network (NN) model having one or more validity tags corresponding to conditions for determining a location of a user equipment (UE) using the NN model; obtain auxiliary information corresponding to the conditions of the UE; select the NN model by matching the auxiliary information with the validity tags; and determine the location of the UE using the NN model.

[0164] In embodiment 2, which may also include one or more of the embodiments described herein, the device is the UE. In embodiment 3, which may also include one or more of the embodiments described herein, the device is a base station. In embodiment 4, which may also include one or more of the embodiments described herein, the device receives the NN model with the one or more validity labels from a server device.

[0165] In embodiment 5, which may also include one or more of the embodiments described herein, the apparatus obtains the NN model having the one or more validity labels by creating the NN model having the one or more validity labels. In embodiment 6, which may also include one or more of the embodiments described herein, the apparatus obtains the NN model having the one or more validity labels during an NN model building process.

[0166] In embodiment 7, which may also include one or more embodiments described herein, a device includes: a memory; and a processor, wherein the processor is configured to, when executing instructions stored in the memory, cause the server device to: receive auxiliary information; determine a neural network (NN) model having one or more validity tags based on the auxiliary information, wherein the one or more validity tags correspond to conditions for determining a location of a user equipment (UE) using the NN model; and communicate the NN model to a device to implement positioning of the UE or determine the location of the UE using the NN model.

[0167] In embodiment 8, which may also include one or more of the embodiments described herein, the server device includes a location management function of a core network. In embodiment 9, which may also include one or more of the embodiments described herein, the server device includes an artificial intelligence (AI)-management function (MF) (AI-MF) server. In embodiment 10, which may also include one or more of the embodiments described herein, the device is the UE, and the assistance information is received from the UE. In embodiment 11, which may also include one or more of the embodiments described herein, the device is a base station, and the assistance information is received from the base station.

[0168] In embodiment 12, which may also include one or more embodiments of the embodiments described herein, one or more devices may include: a memory; and a processor, wherein the processor is configured to, when executing instructions stored in the memory, cause the one or more devices to: obtain validity test data based on model capability data; and determine the validity of a neural network (NN) model based on the validity test data, wherein: the NN model is configured to enable determination of a location of a user equipment (UE), and when the NN model is valid, use the NN to determine the location of the UE.

[0169] In embodiment 13, which may also include one or more of the embodiments described herein, the one or more devices include at least one of the following: a UE; a base station; a location management function (LMF); or an artificial intelligence (AI)-management function (MF) (AI-MF) server. In embodiment 14, which may also include one or more of the embodiments described herein, testing the validity of the NN model includes determining whether the NN model performs appropriately for the current conditions of the UE. In embodiment 15, which may also include one or more of the embodiments described herein, the model capability data is received from the UE or the base station; and the NN model is used by the UE or the base station.

[0170] In embodiment 16, which may also include one or more embodiments described herein, the model capability data is received from the UE or base station; and the NN model is used by one of the UE, the base station, a location management function (LMF), or an artificial intelligence (AI)-management function (MF) (AI-MF) server.

[0171] In embodiment 17, which may also include one or more of the embodiments described herein, when the NN model is invalid, a message indicating that the NN model is invalid is generated. In embodiment 18, which may also include one or more of the embodiments described herein, the model capability data corresponds to a plurality of NN models, the validity test data corresponds to a plurality of NN models, and the validity of the plurality of NN models is determined. In embodiment 19, which may also include one or more of the embodiments described herein, the plurality of NN models correspond to different NN model classes.

[0172] In embodiment 20, which may also include one or more of the embodiments described herein, a plurality of NN models may correspond to one or more NN model classes. In embodiment 21, which may also include one or more of the embodiments described herein, when more than one NN model is valid for a validity test data set, a selection criterion may be applied to select a best NN model among the more than one NN models. In embodiment 22, which may also include one or more of the embodiments described herein, a device comprises: a memory; and a processor configured to, when executing instructions stored in the memory, cause the device to: communicate mobility capability data to a location management function (LMF); communicate neural network (NN) condition and requirement data to the LMF; receive model configuration and validity data from the LMF; and determine whether one or more NN models are valid based on the model configuration and validity data.

[0173] In embodiment 23, which may also include one or more of the embodiments described herein, the apparatus comprises a user equipment (UE). In embodiment 24, which may also include one or more of the embodiments described herein, the apparatus comprises a base station. In embodiment 25, which may also include one or more of the embodiments described herein, the capability data comprises quality of accuracy, latency, reference signal configuration, and model type corresponding to the NN model. In embodiment 26, which may also include one or more of the embodiments described herein, the model configuration and validity data comprises a model identifier (ID), reference signal configuration, feedback configuration, and assistance data.

[0174] In embodiment 27, which may also include one or more of the embodiments described herein, the device is further configured to provide feedback data regarding the effectiveness of the one or more NN models to the LMF. In embodiment 28, which may also include one or more of the embodiments described herein, the device is further configured to receive a model indication of whether to use the one or more NN models in response to transmitting the feedback data.

[0175] In embodiment 29, which may also include one or more of the embodiments described herein, the server device may include: a memory; and a processor configured to, when executing instructions stored in the memory, cause the device to: receive model capability data and auxiliary data from a device; determine a neural network (NN) model configuration and validity label based on the model capability data and the auxiliary data; communicate the model configuration and validity data to the device; and receive validity confirmation for the model configuration and validity data from the device.

[0176] In embodiment 30, which may also include one or more of the embodiments described herein, the server device comprises a location management function of a core network. In embodiment 31, which may also include one or more of the embodiments described herein, the server device comprises an artificial intelligence (AI)-management function (MF) (AI-MF) server. In embodiment 32, which may also include one or more of the embodiments described herein, the model capability data comprises reference signal configurations supported by the device.

[0177] In embodiment 33, which may also include one or more embodiments described herein, a method performed by a device, the method comprising: obtaining a neural network (NN) model having one or more validity tags corresponding to conditions for determining a location of a user equipment (UE) using the NN model; obtaining auxiliary information corresponding to the conditions of the UE; selecting the NN model by matching the auxiliary information with the validity tags; and determining the location of the UE using the NN model.

[0178] In embodiment 34, which may also include one or more of the embodiments described herein, the device receives the NN model having the one or more validity tags from a server device. In embodiment 35, which may also include one or more of the embodiments described herein, the device obtains the NN model having the one or more validity tags by creating the NN model having the one or more validity tags. In embodiment 36, which may also include one or more of the embodiments described herein, the device obtains the NN model having the one or more validity tags during an NN model building process.

[0179] In embodiment 37, which may also include one or more embodiments described herein, a method performed by a server device, the method comprising: receiving auxiliary information; determining a neural network (NN) model having one or more validity tags based on the auxiliary information, the one or more validity tags corresponding to conditions for using the NN model to determine a location of a user equipment (UE); and communicating the NN model to a device to implement positioning of the UE or using the NN model to determine the location of the UE.

[0180] In embodiment 38, which may also include one or more of the embodiments described herein, the server device comprises a location management function of a core network. In embodiment 39, which may also include one or more of the embodiments described herein, the server device comprises an artificial intelligence (AI)-management function (MF) (AI-MF) server. In embodiment 40, which may also include one or more of the embodiments described herein, the device is the UE, and the assistance information is received from the UE.

[0181] In embodiment 41, which may also include one or more of the embodiments described herein, a computer-readable medium includes instructions that, when executed by one or more processors, cause the one or more processors to: obtain a neural network (NN) model having one or more validity tags corresponding to conditions for determining a location of a user equipment (UE) using the NN model; obtain auxiliary information corresponding to the conditions of the UE; select the NN model by matching the auxiliary information with the validity tags; and determine the location of the UE using the NN model.

[0182] In embodiment 42, which may also include one or more of the embodiments described herein, a computer-readable medium includes instructions that, when executed by one or more processors, cause the one or more processors to: receive auxiliary information; determine a neural network (NN) model having one or more validity tags based on the auxiliary information, the one or more validity tags corresponding to conditions for determining a location of a user equipment (UE) using the NN model; and communicate the NN model to a device to implement positioning of the UE or determine the location of the UE using the NN model.

[0183] The above description of illustrative examples, implementations, aspects, etc. of the subject matter of the present disclosure, including what is described in the Abstract, is not intended to be exhaustive or to limit the disclosed aspects to the precise forms disclosed. Although specific examples, implementations, aspects, etc. are described herein for illustrative purposes, various modifications are contemplated within the scope of such examples, implementations, aspects, etc., as those skilled in the relevant art will recognize.

[0184] In this regard, although the subject matter of the present disclosure has been described in conjunction with various examples, implementations, aspects, etc. and corresponding figures, it should be understood that other similar aspects may be used, or modifications and additions may be made to the disclosed subject matter, where applicable, to perform the same, similar, alternative, or alternative functions of the disclosed subject matter without departing from the disclosed subject matter. Accordingly, the disclosed subject matter should not be limited to any single example, implementation, or aspect described herein, but should be construed in accordance with the breadth and scope of the claims appended hereto.

[0185] In particular, with respect to the various functions performed by the components or structures (assemblies, devices, circuits, systems, etc.) described above, unless otherwise indicated, the terms used to describe such components (including references to "parts") are intended to correspond to any component or structure that performs the specified function of the described component (e.g., functionally equivalent), even if not structurally equivalent to the disclosed structures that perform the functions in the exemplary implementations illustrated herein. In addition, although particular features have been disclosed with respect to only one of several implementations, for any given application, such features may be combined with one or more other features of other implementations, as may be desirable and advantageous.

[0186] As used herein, the term "or" is intended to mean an inclusive "or" rather than an exclusive "or". That is, unless otherwise specified or clear from the context, "X employs A or B" is intended to mean any of the natural inclusive permutations. That is, if X employs A; X employs B; or X employs both A and B, then "X employs A or B" is satisfied in any of the foregoing cases. In addition, the articles "a" and "an" used in this application and the appended claims should generally be construed to mean "one or more" unless otherwise specified or clear from the context to be directed to the singular form. Moreover, to the extent that the terms "including," "comprising," "having," "having," "with," or variations thereof are used in the detailed description or claims, such terms are intended to be inclusive in a manner similar to the term "comprising." Additionally, where one or more numbered items are discussed (e.g., "a first X," "a second X," etc.), generally, the one or more numbered items may be different or they may be the same, but in some cases, the context may indicate that they are different or that they are the same.

[0187] It is understood that the use of personally identifiable information should be subject to privacy policies and practices that are generally recognized to meet or exceed industry or government requirements for maintaining user privacy. In particular, personally identifiable information data should be managed and handled to minimize the risk of inadvertent or unauthorized access or use, and the nature of authorized use should be clearly indicated to users.

Claims

1. A device, comprising: Memory; and a processor configured to, when executing instructions stored in the memory, cause the device to: obtaining a neural network (NN) model having one or more validity tags corresponding to conditions for determining a location of a user equipment (UE) using the NN model; Obtaining auxiliary information corresponding to the condition of the UE; selecting the NN model by matching the auxiliary information with the validity label; as well as The location of the UE is determined using the NN model. The device according to claim 1 , wherein the device is the UE. The device of claim 1 , wherein the device is a base station.

4. The apparatus of claim 1, wherein the apparatus receives the NN model with the one or more validity labels from a server device. 5 . The apparatus according to claim 1 , wherein the apparatus obtains the NN model having the one or more validity labels by creating the NN model having the one or more validity labels.

6. The apparatus of claim 1, wherein the apparatus obtains the NN model with the one or more validity labels during a NN model building process.

7. A server device, comprising: Memory; and a processor configured to, when executing instructions stored in the memory, cause the server device to: receiving auxiliary information; determining a neural network (NN) model having one or more validity tags based on the assistance information, the one or more validity tags corresponding to conditions for determining a location of a user equipment (UE) using the NN model; as well as The NN model is communicated to a device to implement positioning of the UE or the NN model is used to determine the position of the UE.

8. The server device according to claim 7, wherein the server device comprises a location management function of a core network.

9. The server device of claim 7, wherein the server device comprises an artificial intelligence (AI)-management function (MF) (AI-MF) server. 10 . The server device according to claim 7 , wherein the device is the UE, and the assistance information is received from the UE.

11. The server device of claim 7, wherein the device is a base station, and the assistance information is received from the base station.

12. One or more devices, comprising: Memory; and a processor configured, when executing instructions stored in the memory, to cause the one or more devices to: Obtain validity test data based on model capability data; as well as determining the validity of a neural network (NN) model based on the validity test data, in: The NN model is configured to enable determination of a location of a user equipment (UE), and When the NN model is valid, the location of the UE is determined using the NN.

13. The one or more devices of claim 12, wherein the one or more devices include at least one of the following: the UE; base stations; Location Management Function (LMF); or Artificial Intelligence (AI)-Management Function (MF) (AI-MF) server.

14. The one or more devices of claim 12, wherein testing the validity of the NN model comprises determining whether the NN model performs appropriately for current conditions of the UE.

15. One or more devices according to claim 12, wherein: The model capability data is received from the UE or base station; and The NN model is used by one of the UE, the base station, a location management function (LMF), or an artificial intelligence (AI)-management function (MF) (AI-MF) server.

16. The one or more devices of claim 12, wherein when the NN model is invalid, a message is generated indicating that the NN model is invalid.

17. One or more devices according to claim 12, wherein the model capability data corresponds to a plurality of NN models, the validity test data corresponds to a plurality of NN models, and the validity of the plurality of NN models is determined.

18. The one or more devices of claim 17, wherein the plurality of NN models correspond to different NN model classes.

19. The one or more devices of claim 17, wherein the plurality of NN models may correspond to one or more NN model classes.

20. The one or more devices of claim 17, wherein when more than one NN model is valid for the validity test data set, a selection criterion can be applied to select the best NN model among the more than one NN models.

Images