WO2024141153A1 - User equipment positioning - Google Patents

Abstract

A method (1000) for estimating the location of a first UE. The method includes obtaining a trained model configured to map input data to a location, wherein the input data comprises a set of reference signal, RS, measurements. The method also includes obtaining a first set of RS measurements associated with the first UE. The method also includes inputting the first set of reference signal measurements into the model. The method also includes, after inputting the first set of reference signal measurements into the model, obtaining from the model location information indicating an estimated location of the first UE. Obtaining the first set of RS measurements associated with the first UE comprises: using a first beam within a first GoB associated with a first antenna array, to receive an RS transmission from the first UE; measuring the RS transmission as received using the first beam to produce a first RS measurement associated with the first UE; using a second beam within the GoB to receive the RS transmission from the first UE; and measuring the RS transmission as received using the second beam to produce a second RS measurement associated with the first UE.

Description

TITLE

USER EQUIPMENT POSITIONING

TECHNICAL FIELD

[001] Disclosed are embodiments related to user equipment (UE) positioning.

BACKGROUND

[002] In many use cases, a UE’s position is determined with the aid of a navigation satellite system (NSS), which provides autonomous geo-spatial positioning with either global or regional coverage. A Global NSS (GNSS) generally refers to NSS that achieves global coverage. As noted in section 8.1 of the 3rd Generation Partnership Project (3GPP) Technical Specification (TS) 38.305 V17.2.0 (“TS 38.305”), the following are some of the NSSs currently supported by 3 GPP: Global Positioning System (GPS), Galileo, Satellite Based Augmentation Systems (SB AS), Quasi-Zenith Satellite System (QZSS), BeiDou Navigation Satellite System (BDS). Each NSS can be used individually or in combination with others, including regional navigation systems and augmentation systems. When used in combination, the effective number of navigation satellite signals would be increased.

[003] 3 GPP Radio Network

[004] High-band deployment by a network is referred to by 3GPP as deployment on frequencies higher than 6 GHz. To cope with the coverage challenge at these high frequencies more antenna elements are needed. In 3 GPP New Radio (NR) the notion of massive antenna arrays has been introduced to achieve both increased coverage and increased level of throughput. These antenna arrays are sometimes referred to as Advanced Antenna Systems (AAS). In 3GPP the AAS is referred to as a Transmission/Reception Point (TRP) and is simply a collection of antenna elements, like a panel of elements. To reduce the cost of an AAS, analog beamforming may be used. This means transmissions involving the base station can only happen on one such analog beam at a time. Also, at the UE side, analog beamforming is expected for high-band deployment; which means the UE can only receive a transmission from one beam at a time because its spatial reception filter applies to all resource elements of an Orthogonal Frequency Division Multiplexing (OFDM) symbol (per polarization). Typically for high-band deployment, bands are TDD.

[005] Analog beamforming is an example of time-domain beamforming, meaning one beamform applies to all frequency resources being part of a transmission from the base station. Hybrid beamforming based on different sub-arrays of antenna elements connected to separate RF chains is another version of time-domain beamforming. Compared to strict digital beamforming, hybrid beamforming can be seen as the digital domain operating an array of subarrays of antennas. The subarrays of antennas are subject to analog beamforming and act as physical antennas except that the beamforms of the subarrays can each be pointing in different directions given appropriate analog weights for a specific point in time. Here we refer to these sub-arrays as analog antenna subarrays.

[006] Analog beamforming is one example of time-domain beamforming, meaning one beamform applies to all frequency resources being part of a transmission from the base station. Hybrid beamforming based on different sub-arrays of antenna elements connected to separate RF chains is another version of time-domain beamforming. Compared to strict digital beamforming, hybrid beamforming can be seen as the digital domain operating an array of subarrays of antennas as shown in FIG. 1 A and FIG. IB. The subarrays of antennas are subject to analog beamforming and act as physical antennas except that the beamforms of the subarrays can each be pointing into different directions given appropriate analog weights for a specific point in time. Here we refer to these sub-arrays as analog antenna subarrays. An example sub-array is labeled as 101 in FIG. 1A.

[007] Theoretically, the analog beamforms of the analog antenna subarrays can point into different directions; however, a typical deployment would have these beamforms targeting similar directions, as shown in FIG. IB, to increase the coverage of a certain area.

[008] For at least one of the analog directions of FIG. IB, there is an option of digital beamforming. The case of no analog beamforming can be seen as there is simply only one analog beam available of FIG. IB. The digital beamforming may introduce a grid of beams (GoB), both in horizontal and vertical dimension. Seen from above this would look like the beams shown in FIG. 2A, showing a GoB covering the horizontal dimension (in the example shown the GoB consists of 23 beams, but this is just an example as the GoB may have M number of beams where M is greater than or less than 23). A similar grid exists also for the vertical dimension as shown in FIG. 2B. The resolution of the grid (the number of beams) depends on number of analog antenna sub arrays.

[009] FIG. 2B is a front view of an antenna panel and of the GoB. In this case, digital beamforming introduces a grid both for the azimuth and the elevation. For illustration, it is indicated how the beams in previous figure are associated to the upper elevation. There is one GoB like this per analog elevation shown in FIG. 1.

[0010] Compared to a full digital solution hybrid beamforming reduces the need to transfer data between the frontend and baseband. Another strategy to minimize data transfer between baseband and frontend is to limit number of layers allowed at a specific time occasion. A third strategy is to only receive data from all the analog antenna subarrays on a fraction of the full bandwidth. Whereas this disregards from frequency-related information it still allows the base station to spatially resolve the received signal from the grid of all the analog antenna subarrays elements (on a reduced bandwidth) using digital reception assuming a certain analog beamform.

[0011] As the number of analog antenna subarrays grows this allows for more beams in the grid (and each beam is narrower) which in turns allows for increasingly more accurate information on the UE location. This is an alternative way of tracking UE location compared to using GPS coordinates which sometimes is not accurate.

[0012] In 3GPP terminology, sounding is referred to as a Sounding Reference Signal (SRS), see 3GPP Technical Specification (TS) 38.211 V17.3.0 (“TS 38.211”). The SRS transmitted by a UE on the uplink is a specifically defined reference signal received by the base station, repeatedly using each analog beamform (of the analog antenna subarrays as shown in FIG. 1) applicable for the intended coverage area. There could be a number of these analog beamforms; therefore, the UE needs to transmit SRS on several symbols such that the base station can receive on some symbols using a certain beamform until SRS has been received on all of the analog beamforms considered relevant for the UE. The base station allocates SRS resources individually for each UE. The UEs are provided with SRS resources orthogonal to each other, such there is no risk of interference. SUMMARY

[0013] Certain challenges presently exist. For instance, some systems rely on a UEs NSS (e.g. GPS) capability for determining the position of the UE, but if the UE’s NSS module is not activated, this would mean that the UE’s position would not be available. On the other hand, if the NSS module is always activated, then this would be a drain on the UE’s battery. Additionally, the NSS may not work well if the UE is in a location where it cannot receive the satellite signals.

[0014] Accordingly, in one aspect there is provided a method for estimating the location of a UE. The method includes obtaining a trained model configured to map input data to a location, wherein the input data comprises a set of reference signal, RS, measurements. The method also includes obtaining a first set of RS measurements associated with the first UE. The method also includes inputting the first set of reference signal measurements into the model. The method also includes, after inputting the first set of reference signal measurements into the model, obtaining from model location information indicating an estimated location of the first UE. Obtaining the first set of RS measurements associated with the first UE comprises: using a first beam within a first GoB associated with a first antenna array, to receive an RS transmission from the first UE; measuring the RS transmission as received using the first beam to produce a first RS measurement associated with the first UE; using a second beam within the GoB to receive the RS transmission from the first UE; and measuring the RS transmission as received using the second beam to produce a second RS measurement associated with the first UE.

[0015] In another aspect there is provided a computer program comprising instructions which when executed by processing circuitry of a network node causes the network node to perform any of the methods disclosed herein. In one embodiment, there is provided a carrier containing the computer program wherein the carrier is one of an electronic signal, an optical signal, a radio signal, and a computer readable storage medium. In another aspect there is provided a network node that is configured to perform the methods disclosed herein. The network node may include memory and processing circuitry coupled to the memory.

[0016] An advantage of the embodiments disclosed herein is that they enable robust positioning (for instance when satellite signal is blocked). That is, it is possible to determine a UE’s location without having to rely on the UE receiving the GPS signals; this way the UE consumes less power.

BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings, which are incorporated herein and form part of the specification, illustrate various embodiments.

[0018] FIG. 1 A illustrates a TRP.

[0019] FIG. IB illustrates beams produced by a TRP.

[0020] FIG. 2 A illustrates a GoB.

[0021] FIG. 2B is a front view of an antenna panel.

[0022] FIG. 3 illustrates an example of supervised learning.

[0023] FIG. 4 illustrates steps of a machine learning process.

[0024] FIG. 5. illustrates an artificial neural network.

[0025] FIG. 6 is a message flow diagram according to an embodiment.

[0026] FIG. 7 is a message flow diagram according to an embodiment.

[0027] FIG. 8 illustrates a machine learning process for a multi-TRP use case.

[0028] FIG. 9 illustrates a node of an artificial neural network.

[0029] FIG. 10 is a flowchart illustrating a process according to an embodiment.

[0030] FIG. 11 is a block diagram of a network node according to an embodiment.

[0031] FIG. 12 is a block diagram of a network node according to another embodiment.

DETAILED DESCRIPTION

[0032] There are many ways to creating models for use with Machine Learning (ML) algorithms, with three main categories: 1) supervised learning, in which algorithms are trained using labeled datasets; 2) unsupervised learning, using a clustering or grouping technique without labeled datasets; and 3) reinforcement learning, using a target reward strategy that allows guidance toward an optimal set of actions. A labeled dataset is a dataset comprising many training examples wherein each training example consists of input data paired with output data (a.k.a., a label).

[0033] In this disclosure, supervised learning is used to predict a UE’s position (i.e., the output of the ML model is a predicted location for the UE). As used herein, a UE is any device capable of wireless communication with a base station. In FIG. 3 illustrates an example of supervised learning. Supervised learning has three steps: 1) a training step, 2) a validation step, and 3) a prediction step. The collected historical data is split into two parts: 1) a labeled dataset used for training the ML model (a.k.a., training dataset or training examples) and 2) a test dataset used to verify the ML model.

[0034] First, the training examples are used to produce a trained model. Next, the test dataset is used by the trained model for validation, which is a check of whether the trained model has reached the learning target (e.g., prediction precision). Finally, new input data is used by the trained model to produce a prediction.

[0035] In one embodiment, each training example in the training dataset includes input data that comprises SRS measurements of GoBs. The input data may also comprise other measurements (e.g., downlink reference signal (RS) measurements). Each training example also includes output data that comprises UE location information (e.g., GPS coordinates). That is, the position information is used as labels of the input data (e.g., SRS measurements).

[0036] Many such training examples from many UEs are collected and fed into the machine learning algorithm for training. GoB SRS measurements (typically on narrow beams such as the ones shown in FIG. 2A) can be replaced or complemented by Reference Signal Received Power (RSRP) measurements on System Synchonization Blocks (SSBs) which are typically transmitted on wider beams and/or on channel state information reference signals (CSI- RS). After training and validation, the trained machine learning model is used for prediction. This means that GoB SRS measurements, RSRP measurements or other input data of new UEs will be provided to the trained model, and the output from the model is the predicted positions of these new UEs. These procedural steps are shown in FIG. 4.

[0037] There are many supervised machine learning algorithms. Artificial neural network is one of the most widely used algorithms. In FIG. 5 an example is given how neural networks are used for UE position prediction. [0038] An artificial neural network is an interconnected group of nodes. Each circular node in FIG. 5 represents an artificial neuron and each line between two circular nodes represents a connection from the output of one artificial neuron to the input of another. A typical neural network has three layers: input layer, hidden layer and output layer.

[0039] In FIG. 5, the input data of the input layer is the SRS measurements, where N is the maximum number of beams in the GoB. The index “i” means the ith training example. The index “n” refers to a measurement of SRS on the nth beam (within that training example). The hidden layer connects input layer and output layer, and it contains the main learning model information. The hidden layer could have one layer or multiple layers of neuron nodes. The output of the output layer is location information (e.g., coordinates). During training, the information of UE coordinates is provided in UE reports; after training, the output will be the predicted UE coordinates.

[0040] To connect a UE’s location (e.g., coordinates as acquired from a positioning system such as GPS) to RS measurements (e.g., SRS measurements), communication between the UE and the gNB is needed. To this end, support from a Location Management Function (LMF) can be useful. An LMF is responsible for, among other things, positioning of UEs (see, e.g., 3GPP TS 38.305 section 5.4.4). A schematic signaling procedure allowing ML training and positioning is shown in FIG. 6.

[0041] In FIG. 6, an LMF 606 requests a UE 602 to report its coordinates by sending to the UE a “UE position measurement request” message. The UE reports its coordinates in a “UE position measurement report.” In 3GPP, the information sent from the UE to the LMF is shown in table 8.1.2.2-1 of TS 38.305. Information of interest in this table could be the position represented by Latitude/Longitude/ Altitude. The message names “UE position measurement request” and “UE position measurement report” in FIG. 6 are generic names for messages defined in 3 GPP.

[0042] As illustrated in FIG. 6, in one embodiments, the LMF transmits to the network node station 604, which is serving the UE, a UE position measurement report message containing UE position like Latitude/Longitude/ Altitude (as previously sent from the UE to LMF) to connect the SRS measurements to the coordinates in the network node. Network node may be, but is not limited to, an access point (AP) (e.g., radio access point), a base station (BS) (e.g., radio base station, Node B, evolved Node B (eNB) and NR NodeB (gNB)), 0-RAN nodes or components of an 0-RAN node (e.g., 0-RU, 0-DU, O-CU), a core network node, etc. For shorthand, the network node 604 will be referred to herein as a “gNB”, but is not limited to gNBs.

[0043] This message from the LMF to the gNB may trigger the gNB to trigger the UE to perform SRS transmissions (if no measurement of SRS transmissions recent enough is available). After the gNB has measured the SRS transmissions, the measurements together with coordinates are used for training the ML model.

[0044] The message “UE position measurement report” from LMF to gNB could be a message part of a proprietary interface. Another option would be to have the coordinates sent from the UE straight to gNB (which would be a new addition to Radio Resource Control (RRC) protocol).

[0045] After the ML model is trained, the LMF may send to the gNB a UE position request message identifying a particular UE, a particular group of UEs (e.g., UEs of a particular type), or any UE. For simplicity, assume the UE position request identified UE 602. After receiving the UE position request, the gNB measures SRS transmission performed by UE 602 and inputs those measurements into the trained ML model, and the ML model, based on that input, would produce UE location information indicating a predicted location of UE 602. The gNB may then provide this predicted UE location information to the LMF. In FIG. 6 this is represented by a message “Predicted UE position report.” The intention is that transferring coordinates creates less load on the interface between gNB and LMF compared to large amount of SRS measurements. Moreover, this means LMF will be provided UE coordinates also during phases when UEs are out of NSS coverage. FIG. 7 illustrates an embodiment in which a network function (which in this example is the LMF, but it could be another network function) performs the ML model training. Accordingly, as shown in FIG. 7, the LMF obtains the necessary reference signal measurements (e.g., the SRS measurements).

[0046] In the embodiments described above, only a single TRP was used (all sounding measurements are from one TRP). One TRP may allow a unique connection between coordinates and sounding measurements since not only the direction of the GoB is detected by the gNB but also the amount of energy received on the beam. Accordingly, gNB would register less energy for sounding measurements from UEs far away (compared to UEs close to gNB). More than one TRP would make the connection between coordinates and sounding measurements even more reliable; however, more data would be processed by the ML algorithms.

[0047] In FIG. 8 illustrates the ML methodology for the multi-TRP case. Note that the base stations could be the same or at different locations. If some of the base stations are the same it means these sounding measurements are received by one gNB with many TRPs; if a base station is different than the others it means it is a neighboring base station provided SRS measurements by its own TRP.

[0048] With respect to introducing more TRPs, there are two main options considering how to configure the sounding resources. One option is to separate sounding resource per TRP such that one sounding resource is allocated to one TRP (and the associated SRS transmission is received by this TRP) and another sounding resource is allocated to another TRP (and the associated SRS transmission is received by other TRP). The sounding resources in this case are on different frequency and/or time resources. This consumes more air interface resources; also, the more distant TRP may not receive its sounding. Another option is that TRPs all receive the same sounding resource. This consumes less air interface resources.

[0049] It is expected that any of the NSS positioning methods could support the LMF (with the help of the UE) to get the coordinates (to be transferred to gNB during training as described above).

[0050] In the following details are provided on the training and validation steps.

[0051] Training:

[0052] There are a number of standard procedures in the training phase. One variant is described here, based on FIG. 5. Each node in the hidden layer in FIG. 5 takes several inputs, multiplied with weights as shown in FIG. 9. Next, the inputs multiplied with respective weight is accumulated. The accumulated sum is then multiplied by an activation function with the intention to only consider the output from this hidden-layer node if it exceeds a certain value (the case when the activation function is a threshold function). This activation function can be smooth such that it allows smaller output to have some impact and larger output from the node not as much impact. In the end the output is compared with the coordinates reported from the positioning system.

[0053] Effectively an iterative procedure is applied. First, a training example (e.g., SRS measurements paired with coordinates) is processed according to the estimated weights Wi (weights from all nodes in FIG. 5) from the previous training example. If this was the first training example (in other words the first SRS measurement entering the training phase) the weights Wi would be chosen randomly (since there is no previous training example). One would then calculate a first error, epsilon (a) (since the coordinates are available). Then a second set of weights Wi' would be selected (somewhat) different from the first set. A second error, epsilon prime (s’), is calculated based on Wi' to estimate the gradient as seen in the equation below. The gradient is then multiplied by a weight step A (in the opposite direction of the gradient to increase likelihood of convergence towards the error minimum); the negative weight step multiplied by the gradient is then added to the old previous Wi weight to obtain a new weight Wi':

[0054] The procedure is then repeated with the new weight Wi' considered the previous weight Wi until close enough to the minimum. The procedure above is repeated for all training examples (SRS measurements together with coordinates) entering the training phase. In the end weights of the nodes in FIG. 5 are expected to be reasonably well estimated, given large amounts of groups of data.

[0055] Validation:

[0056] After training has finished another set of training examples is provided to the nodes of FIG. 5. The input data of the training example (e.g., SRS measurements) together with weights is used to calculate estimates of the coordinates. If estimates consistently come out close enough to real coordinates this phase is done (i.e., the model does not need additional training).

[0057] FIG. 10 is a flow chart illustrating a process 1000, according to an embodiment, for estimating the location of a UE (e.g., UE 602). Process 1000 may begin in step sl002.

[0058] Step si 002 comprises obtaining a trained model (e.g., trained artificial neural network) configured to map input data to a location (e.g., 2D or 3D coordinates, address, room id), wherein the input data comprises a set of reference signal (RS) (e.g., SRS, PRS, SSB, CSI- RS, uplink (UL) DMRS, etc.) measurements.

[0059] Step si 004 comprises obtaining a first set of RS measurements associated with the first UE. This step of obtaining the first set of RS measurements associated with the first UE comprises: using a first beam within a first GoB associated with a first antenna array, to receive an RS transmission from the first UE; measuring the RS transmission as received using the first beam to produce a first RS measurement associated with the first UE; using a second beam within the GoB to receive the RS transmission from the first UE; and measuring the RS transmission as received using the second beam to produce a second RS measurement associated with the first UE.

[0060] Step si 006 comprises inputting the first set of reference signal measurements into the model. Step sl008 comprises, after inputting the first set of reference signal measurements into the model, obtaining from model location information indicating an estimated location of the first UE.

[0061] In some embodiments, the RS transmission is a sounding reference signal, SRS, transmission.

[0062] In some embodiments, obtaining a trained model comprises producing the trained model using a training process that comprises: obtaining a set of training examples; and training the model using the set of training examples. A training example within the set of training examples comprises: i) output data comprising location information indicating the location of a second UE at a point in time; and ii) input data paired with the output data, wherein the input data comprises: a first measurement of an RS transmitted by the second UE and received using the first beam within the first GoB, wherein the RS transmitted by the second UE was transmitted within X amount of time from said point in time, where X is greater than or equal to 0, and a second measurement of the RS transmitted by the second UE and received using the second beam within the first GoB.

[0063] In some embodiments, obtaining the training example comprises receiving the output data from the LMF and generating or retrieving the input data. In some embodiments, obtaining the training example further comprises, in response to receiving the output data from the location management function, determining whether the input data exists; if the input data exists, then retrieving the input data; and if the input data does not exist, then generating the input data.

[0064] In some embodiments, obtaining the training example comprises generating the input data. In some embodiments, generating the input data comprises: using the first beam within the first GoB to receive the RS transmission from the second UE; measuring the RS transmission from the second UE as received using the first beam to produce the first RS measurement associated with the second UE; using the second beam within the GoB to receive the RS transmission from the second UE; and measuring the RS transmission from the second UE as received using the second beam to produce the second RS measurement associated with the second UE. In some embodiments, generating the input data further comprises triggering the second UE to perform the RS transmission.

[0065] In some embodiments, the method is performed by a base station. In some embodiments, the set of training examples further comprises a second training example comprising second output data and second input data, and obtaining the second training example comprises receiving the second output data from the location management function; and receiving the second input data from a second base station.

[0066] In some embodiments, the method is performed by the LMF (or other network function).

[0067] FIG. 11 is a block diagram of network node 604, according to some embodiments for performing methods disclosed herein. As shown in FIG. 11, network node 604 may comprise: processing circuitry (PC) 1102, which may include one or more processors (P) 1155 (e.g., a general purpose microprocessor and/or one or more other processors, such as an application specific integrated circuit (ASIC), field-programmable gate arrays (FPGAs), and the like), which processors may be co-located in a single housing or in a single data center or may be geographically distributed (i.e., the network node may be a distributed computing apparatus); a network interface 1168 comprising a transmitter (Tx) 1165 and a receiver (Rx) 1167 for enabling network node 604 to transmit data to and receive data from other nodes connected to a network 110 (e.g., an Internet Protocol (IP) network) to which network interface 1168 is connected; communication circuitry 1148 (e.g., radio transceiver circuitry comprising an Rx 1147 and a Tx 1145) coupled to an antenna system 1149 for wireless communication with UEs or other nodes; and a storage unit (a.k.a., “data storage system”) 1108, which may include one or more nonvolatile storage devices and/or one or more volatile storage devices. In embodiments where PC 1102 includes a programmable processor, a computer readable storage medium (CRSM) 1142 may be provided. CRSM 1142 may store a computer program (CP) 1143 comprising computer readable instructions (CRI) 1144. CRSM 1142 may be a non-transitory computer readable medium, such as, magnetic media (e.g., a hard disk), optical media, memory devices (e.g., random access memory, flash memory), and the like. In some embodiments, the CRI 1144 of computer program 1143 is configured such that when executed by PC 1102, the CRI causes network node 604 to perform steps described herein (e.g., steps described herein with reference to one or more flow charts). In other embodiments, network node 604 may be configured to perform steps described herein without the need for code. That is, for example, PC 1102 may consist merely of one or more ASICs. Hence, the features of the embodiments described herein may be implemented in hardware and/or software.

[0068] FIG. 12 is a block diagram of network node 1200, according to some embodiments, for implementing any network function, such as LMF 606. As shown in FIG. 12, network node 1200 may comprise: processing circuitry (PC) 1202, which may include one or more processors (P) 1255 (e.g., one or more general purpose microprocessors and/or one or more other processors, such as an application specific integrated circuit (ASIC), field-programmable gate arrays (FPGAs), and the like), which processors may be co-located in a single housing or in a single data center or may be geographically distributed (i.e., network node 1200 may be a distributed computing apparatus); at least one network interface 1248 (e.g., a physical interface or air interface) comprising a transmitter (Tx) 1245 and a receiver (Rx) 1247 for enabling network node 1200 to transmit data to and receive data from other nodes connected to a network 110 (e.g., an Internet Protocol (IP) network) to which network interface 1248 is connected (physically or wirelessly) (e.g., network interface 1248 may be coupled to an antenna arrangement comprising one or more antennas for enabling network node 1200 to wirelessly transmit/receive data); and a storage unit (a.k.a., “data storage system”) 1208, which may include one or more non-volatile storage devices and/or one or more volatile storage devices. In embodiments where PC 1202 includes a programmable processor, a computer readable storage medium (CRSM) 1242 may be provided. CRSM 1242 may store a computer program (CP) 1243 comprising computer readable instructions (CRI) 1244. CRSM 1242 may be a non-transitory computer readable medium, such as, magnetic media (e.g., a hard disk), optical media, memory devices (e.g., random access memory, flash memory), and the like. In some embodiments, the CRI 1244 of computer program 1243 is configured such that when executed by PC 1202, the CRI causes network node 1200 to perform steps described herein (e.g., steps described herein with reference to the flow charts). In other embodiments, network node 1200 may be configured to perform steps described herein without the need for code. That is, for example, PC 1202 may consist merely of one or more ASICs. Hence, the features of the embodiments described herein may be implemented in hardware and/or software.

[0069] While various embodiments are described herein, it should be understood that they have been presented by way of example only, and not limitation. Thus, the breadth and scope of this disclosure should not be limited by any of the above-described exemplary embodiments. Moreover, any combination of the above-described elements in all possible variations thereof is encompassed by the disclosure unless otherwise indicated herein or otherwise clearly contradicted by context.

[0070] As used herein transmitting a message “to” or “toward” an intended recipient encompasses transmitting the message directly to the intended recipient or transmitting the message indirectly to the intended recipient (i.e., one or more other nodes are used to relay the message from the source node to the intended recipient). Likewise, as used herein receiving a message “from” a sender encompasses receiving the message directly from the sender or indirectly from the sender (i.e., one or more nodes are used to relay the message from the sender to the receiving node). Further, as used herein “a” means “at least one” or “one or more.”

[0071] Additionally, while the processes described above and illustrated in the drawings are shown as a sequence of steps, this was done solely for the sake of illustration. Accordingly, it is contemplated that some steps may be added, some steps may be omitted, the order of the steps may be re-arranged, and some steps may be performed in parallel.

Claims

1. A method (1000) for estimating the location of a first user equipment, UE, (602) the method comprising: obtaining (si 002) a trained model configured to map input data to a location, wherein the input data comprises a set of reference signal, RS, measurements; obtaining (si 004) a first set of RS measurements associated with the first UE; inputting (si 006) the first set of reference signal measurements into the model; and after inputting the first set of reference signal measurements into the model, obtaining (si 008) from model location information indicating an estimated location of the first UE, wherein obtaining the first set of RS measurements associated with the first UE comprises: using a first beam within a first group of beams, GoB, associated with a first antenna array, to receive an RS transmission from the first UE; measuring the RS transmission as received using the first beam to produce a first RS measurement associated with the first UE; using a second beam within the GoB to receive the RS transmission from the first UE; and measuring the RS transmission as received using the second beam to produce a second RS measurement associated with the first UE.

2. The method of claim 1, wherein the RS transmission is a sounding reference signal, SRS, transmission.

3. The method of claim 1 or 2, wherein obtaining a trained model comprises producing the trained model using a training process that comprises: obtaining a set of training examples; and training the model using the set of training examples, wherein a training example within the set of training examples comprises: i) output data comprising location information indicating the location of a second UE at a point in time; and ii) input data paired with the output data, wherein the input data comprises: a first measurement of an RS transmitted by the second UE and received using the first beam within the first GoB, wherein the RS transmitted by the second UE was transmitted within X amount of time from said point in time, where X is greater than or equal to 0, and a second measurement of the RS transmitted by the second UE and received using the second beam within the first GoB.

4. The method of claim 3, wherein obtaining the training example comprises: receiving the output data from a location management function, LMF; and generating or retrieving the input data.

5. The method of claim 4 wherein obtaining the training example further comprises: in response to receiving the output data from the location management function, determining whether the input data exists; if the input data exists, then retrieving the input data; and if the input data does not exist, then generating the input data.

6. The method of claim 4 or 5, wherein obtaining the training example comprises generating the input data; and generating the input data comprises: using the first beam within the first GoB to receive the RS transmission from the second UE; measuring the RS transmission from the second UE as received using the first beam to produce the first RS measurement associated with the second UE; using the second beam within the GoB to receive the RS transmission from the second UE; and measuring the RS transmission from the second UE as received using the second beam to produce the second RS measurement associated with the second UE.

7. The method of claim 6, wherein generating the input data further comprises triggering the second UE to perform the RS transmission.

8. The method of any one of claims 1-7, wherein the method is performed by a first base station (604).

9. The method of claim 8, wherein the set of training examples further comprises a second training example comprising second output data and second input data, and obtaining the second training example comprises: receiving the second output data from the location management function; and receiving the second input data from a second base station.

10. The method of any one of claims 1-7, wherein the method is performed by a network function (606).

11. A network node (602, 1200) for estimating the location of a first user equipment, UE, (602) the network node being configured to perform a method comprising: obtaining (si 002) a trained model configured to map input data to a location, wherein the input data comprises a set of reference signal, RS, measurements; obtaining (si 004) a first set of RS measurements associated with the first UE; inputting (si 006) the first set of reference signal measurements into the model; and after inputting the first set of reference signal measurements into the model, obtaining (si 008) from model location information indicating an estimated location of the first UE, wherein obtaining the first set of RS measurements associated with the first UE comprises: using a first beam within a first group of beams, GoB, associated with a first antenna array, to receive an RS transmission from the first UE; measuring the RS transmission as received using the first beam to produce a first RS measurement associated with the first UE; using a second beam within the GoB to receive the RS transmission from the first UE; and measuring the RS transmission as received using the second beam to produce a second RS measurement associated with the first UE.

12. The network node of claim 11, wherein the RS transmission is a sounding reference signal, SRS, transmission.

13. The network node of claim 11 or 12, wherein obtaining a trained model comprises producing the trained model using a training process that comprises: obtaining a set of training examples; and training the model using the set of training examples, wherein a training example within the set of training examples comprises: i) output data comprising location information indicating the location of a second UE at a point in time; and ii) input data paired with the output data, wherein the input data comprises: a first measurement of an RS transmitted by the second UE and received using the first beam within the first GoB, wherein the RS transmitted by the second UE was transmitted within X amount of time from said point in time, where X is greater than or equal to 0, and a second measurement of the RS transmitted by the second UE and received using the second beam within the first GoB.

14. The network node of claim 13, wherein obtaining the training example comprises: receiving the output data from a location management function, LMF; and generating or retrieving the input data.

15. The network node of claim 14 wherein obtaining the training example further comprises: in response to receiving the output data from the location management function, determining whether the input data exists; if the input data exists, then retrieving the input data; and if the input data does not exist, then generating the input data.

16. The network node of claim 14 or 15, wherein obtaining the training example comprises generating the input data; and generating the input data comprises: using the first beam within the first GoB to receive the RS transmission from the second UE; measuring the RS transmission from the second UE as received using the first beam to produce the first RS measurement associated with the second UE; using the second beam within the GoB to receive the RS transmission from the second UE; and measuring the RS transmission from the second UE as received using the second beam to produce the second RS measurement associated with the second UE.

17. The network node of claim 16, wherein generating the input data further comprises triggering the second UE to perform the RS transmission.

18. The network node of any one of claims 11-17, wherein the network node is a first base station (604).

19. The network node of claim 18, wherein the set of training examples further comprises a second training example comprising second output data and second input data, and obtaining the second training example comprises: receiving the second output data from the location management function; and receiving the second input data from a second base station.

20. The network node of any one of claims 11-17, wherein the network node is a network function (606).

21. A computer program (1143, 1243) comprising instructions (1144, 1244) which when executed by processing circuitry (1102, 1202) of a network node (604, 1200) causes the network node to perform the method of any one of claims 1-10.

22. A carrier containing the computer program of claim 21, wherein the carrier is one of an electronic signal, an optical signal, a radio signal, and a computer readable storage medium (1142, 1242).