Cellular positioning with local sensors using neural networks

Abstract

A DNN or other neural network is used to perform RAT-assisted positioning of the UE.SOLUTION: The TXDNN (120) in the BS (108) generates reference signals (138) and performs wireless transmission of the reference signals to the UE (110). The RXDNN (134) at the UE receives the reference signals and local UE sensor data (140) as inputs and generates UE measurements and sensor reports (144) from the inputs. The TXDNN (148) at the UE receives the report (144) as input and generates therefrom an RF signal (154) representing the UE measurements and sensor report for transmission to the BS. The RXDNN (128) at the BS receives the reports (144) from the RF signals (154) as input and generates a location estimate (130) of the UE from this input.SELECTED DRAWING: Figure 1

Description

[Background technology]

[0001] background Accurate and robust positioning of cellular network devices, such as user equipment (UE), often contributes significantly to the effective and efficient operation of cellular networks. High-precision (i.e., centimeter-level or less) UE positioning is particularly interesting for a variety of applications, such as augmented / virtual reality, sensor-based, and industrial applications. One technology that provides high-precision UE positioning is the Global Navigation Satellite System (GNSS). However, GNSS typically suffers from interference, multipath loss, and poor signal-to-noise ratio (SNR) in urban and indoor environments. Therefore, cellular networks often supplement or even replace GNSS and similar technologies with one or more other UE positioning technologies, such as radio access technology (RAT)-assisted UE positioning. For example, current and emerging cellular networks implement signaling or reference signals that network components can use to perform UE positioning. Upon receiving the reference signal, the UE (or base station (BS)) performs various measurements on the reference signal. The UE (or BS) transmits reference signal measurements to one or more other network components, such as the BS (or location server), which use the measurements to calculate an estimate of the UE's position. Summary of the Invention [Means for solving the problem]

[0002] Overview of the embodiment According to some embodiments, a computer-implemented method on a first device includes receiving reference signal information as input to a transmit neural network of the first device, the transmit neural network generating a first output representing a reference signal based on the reference signal information, controlling a radio frequency (RF) antenna interface of the first device to transmit a first RF signal representing the first output for reception by a second device, receiving, at a receive neural network of the first device, input representing one or more RF signals associated with the second device in response to transmitting the first RF signal, and the receive neural network generating a second output representing a position estimate of the second device based on the input to the receive neural network.

[0003] In various embodiments, the method may further include one or more of the following aspects: receiving an input representing one or more RF signals associated with the second device includes receiving, from the second device, a second RF signal representing a signal measurement associated with the first RF signal; the second RF signal received from the second device further represents local sensor data generated at the second device; the position estimate indicates a position of the second device and an orientation of the second device; the first output further represents a downlink position reference signal including symbols dedicated to user equipment positioning; generating a first output includes generating the first output at the transmit neural network based on a first neural network architecture configuration in the transmit neural network; the method further includes selecting a first neural network architecture configuration from a plurality of neural network architecture configurations based on one or more capabilities of at least one of the first device or the second device; selecting the first neural network architecture configuration includes receiving, from the second device, information representing one or more capabilities of the second device and selecting the first neural network architecture configuration using the information. Generating the second output is based on a second neural network architecture configuration in the receiving neural network. generating a second output at the receiving neural network based on the received neural network architecture configuration. The method also includes selecting a second neural network architecture configuration from a plurality of neural network architecture configurations based on one or more capabilities of at least one of the first device or the second device. Selecting the second neural network architecture configuration includes receiving information from the second device representing one or more capabilities of the second device and selecting the second neural network architecture configuration using the information. The method further includes receiving a command from a management infrastructure component to implement at least one of the first neural network architecture configuration in the transmitting neural network or the second neural network architecture configuration in the receiving neural network. The method also includes selecting at least one of a third neural network architecture configuration in the transmitting neural network or a fourth neural network architecture configuration in the receiving neural network in response to changes in the one or more capabilities of at least one of the first device or the second device. At least one of the transmitting neural network and the receiving neural network includes a deep neural network (DNN). The method further includes engaging in joint training of a transmit neural network and a receive neural network of the first device with a transmit neural network and a receive neural network of the second device. The method also includes communicating with a third device implementing the transmit neural network and configuring the transmit neural network of the third device to generate an output representative of the reference signal for reception by the second device.The method further includes generating, at a receiver neural network of the first device, a third output representing a position estimate of the fourth device based on the one or more RF signals received from the fourth device; determining, at the receiver neural network of the first device, that the second output and the third output indicate that the second device and the fourth device occupy the same space; and refining one or more parameters of the receiver neural network of the first device in response to the second output and the third output indicating that the second device and the fourth device occupy the same space.

[0004] According to some embodiments, a computer-implemented method on a first device includes receiving a first RF signal representing a reference signal from a second device at a radio frequency (RF) antenna interface of the first device; providing a representation of the first RF signal as a first input to a receiving neural network of the first device; and generating, by the receiving neural network, a first output representing a measurement report at the first device based on the first input to the receiving neural network.

[0005] In various embodiments, the method may further include one or more of the following aspects. That is, the method may include receiving, at a transmit neural network of the first device, a first output from the receive neural network as an input; the transmit neural network generating a second output representing a measurement report; and controlling an RF antenna interface of the first device to transmit a second RF signal representing the second output for reception by the second device. Generating the first output includes performing one or more reference signal measurements on the first input representing a reference signal, and the measurement report includes at least one of the one or more reference signal measurements. The method may further include providing, as a second input to the receive neural network of the first device, a representation of sensor data generated by one or more sensors of the first device. The measurement report includes one or more reference signal measurements fused with the sensor data. Generating the first output includes generating the first output in the receive neural network based on a first neural network architecture configuration in the receive neural network. The method may include transmitting a second RF signal representing the second output to the first device or the second device. The method further includes selecting a first neural network architectural configuration from the plurality of neural network architectural configurations based on one or more capabilities of at least one of the first device or the second device. Selecting the first neural network architectural configuration includes generating information representing that one or more capabilities of the first device have been changed and providing the information as an input to the receiving neural network. Generating a second output includes generating a second output at the transmitting neural network based on the second neural network architectural configuration at the transmitting neural network. The method further includes selecting a second neural network architectural configuration from the plurality of neural network architectural configurations based on one or more capabilities of at least one of the first device or the second device. Selecting the second neural network architectural configuration includes generating information representing that one or more capabilities of the first device have been changed and providing the representing information as an input to the transmitting neural network. The method also includes receiving a command from a network infrastructure component to implement at least one of the first neural network architectural configuration at the receiving neural network or the second neural network architectural configuration at the transmitting neural network. The method further includes, in response to the change in one or more capabilities of the first device, transmitting a message to the network infrastructure component indicating the change in the one or more capabilities, and, in response to transmitting the message, receiving from the network infrastructure component a second neural network architecture configuration of at least one of the receiving neural network or the transmitting neural network, wherein at least one of the receiving neural network and the transmitting neural network includes a deep neural network (DNN).The method also includes engaging in joint training of a receiving neural network and a transmitting neural network of the first device and a receiving neural network and a transmitting neural network of the second device.

[0006] According to some embodiments, a computer-implemented method includes receiving capability information from at least one of a first device or a second device, and selecting a pair of neural network architecture configurations from a set of candidate neural network architecture configurations based on the capability information, the pair of neural network architecture configurations being jointly trained to perform a cellular device positioning estimation process between the first device and the second device, the method further including sending first instructions of the pair of first neural network architecture configurations to the first device for implementation in one or more of a transmitting neural network and a receiving neural network of the first device, and sending second instructions of the pair of second neural network architecture configurations to the second device for implementation in one or more of a receiving neural network and a transmitting neural network of the second device.

[0007] In various embodiments, the method may further include one or more of the following aspects: The at least one capability includes at least one of an antenna array capability, a processing capability, a power supply capability, a temperature-related capability, or a sensor capability; The transmit neural network and the receive neural network of the first device and the transmit neural network and the receive neural network of the second device each include a deep neural network (DNN).

[0008] In some embodiments, a device includes a radio frequency (RF) antenna interface, at least one processor coupled to the RF antenna interface, and a memory storing executable instructions, the executable instructions operating the at least one processor. and configured to perform any of the methods described hereinabove.

[0009] The present disclosure may be better understood, and its numerous features and advantages made apparent to those skilled in the art by referencing the accompanying drawings, in which: The use of the same reference numbers in different drawings indicates similar or identical items. [Brief explanation of the drawings]

[0010] [Figure 1] FIG. 1 illustrates an example wireless system that uses a UE positioning neural network architecture to calculate position estimates for one or more UEs, in accordance with some embodiments. [Figure 2] FIG. 2 illustrates an example hardware configuration of a UE in the wireless system of FIG. 1 according to some embodiments. [Figure 3] FIG. 2 is a diagram illustrating an example of a hardware configuration of a BS in the wireless system of FIG. 1 according to some embodiments. [Figure 4] FIG. 2 illustrates an example hardware configuration of management infrastructure components of the wireless system of FIG. 1 according to some embodiments. [Figure 5] FIG. 1 illustrates a machine learning (ML) module employing a neural network for use in a UE positioning neural network architecture according to some embodiments. [Figure 6] FIG. 1 illustrates a pair of jointly trained neural networks for processing and transmitting reference signals between one or more BSs and a UE in accordance with some embodiments. [Figure 7] FIG. 1 illustrates a pair of jointly trained neural networks for processing and transmitting UE measurements and sensor reports, including reference signal measurements fused with local UE sensor data, between a UE and a BS in accordance with some embodiments. [Figure 8]FIG. 1 is a flow diagram illustrating an example method for joint training of a set of neural networks to facilitate UE positioning in a wireless system according to some embodiments. [Figure 9] FIG. 1 is a flow diagram illustrating an example method for calculating a UE position estimate using a set of selected jointly trained neural networks according to some embodiments. [Figure 10] 10 is a ladder signaling diagram illustrating an example operation of the method of FIG. 9 according to some embodiments. [Figure 11] FIG. 10 is a flow diagram illustrating another exemplary method for calculating a UE position estimate using a set of selected jointly trained neural networks according to some embodiments. [Figure 12] 12 is a ladder signaling diagram illustrating an example operation of the method of FIG. 11 according to some embodiments. DETAILED DESCRIPTION OF THE INVENTION

[0011] Detailed Description RAT-assisted UE positioning in conventional wireless communication systems typically relies on a series of processing steps / blocks, such as reference signal transmission, reference signal measurement, reference signal measurement reporting, and UE position estimation. The design, testing, and implementation of these processing steps are relatively independent of each other. This custom and independent design approach to each process step typically results in excessive complexity, resource consumption, and overhead. Furthermore, conventional RAT-assisted UE positioning techniques are generally based on reference signal measurements calculated by the UE, or in some cases, by a BS or other infrastructure network component. However, the UE often requires a number of RAT-assisted UE positioning techniques, including a global positioning satellite (GPS) / global navigation satellite system (GNSS) chipset, a camera, object detection sensors, an accelerometer, and inertial measurement devices. The RAT often includes various local sensors such as an IMU (Instrumental Measurement Unit), altimeter, temperature sensor, barometer, etc. Information or data from these UE sensors can improve the accuracy of RAT-assisted UE positioning techniques.

[0012] Thus, rather than taking a handcrafted approach for each process step, exemplary systems and techniques are described below that utilize an end-to-end neural network configuration for RAT-assisted UE positioning, providing improved efficiency and accuracy over traditional RAT-assisted UE positioning techniques, as well as rapid development and deployment. Traditional processing stages in RAT-assisted UE positioning are replaced or supplemented by jointly trained neural networks that operate to fuse sensor data from the UE's available sensors with UE reference signal measurements (or signals) to generate more accurate and meaningful UE position estimates. For example, by fusing and processing UE-provided reference signal measurements (or signals) with UE local sensor information, a BS (or other network component, such as a location server) can generate a UE position estimate that incorporates the UE's local context, can indicate the UE's orientation, and / or can include secondary information such as motion (e.g., rotation, heading, velocity, etc.). Thus, the jointly trained neural network architecture includes a set of neural networks, each trained to provide more accurate and efficient UE positioning than conventional RAT-assisted UE positioning phase sequences, without having to be effectively specially designed or tested with respect to the RAT-assisted UE positioning phase sequence. In at least some embodiments, the jointly trained neural network architecture implements one or more processes of the RAT-assisted UE positioning technique, such as a reference signal transmission process, a reference signal measurement process, a local UE sensor information collection and fusion process, a reference signal measurement and sensor reporting process, and a UE location estimation process.

[0013] In at least some embodiments, the wireless system may employ joint training of multiple candidate neural network architecture configurations for use between the BS and the UE based on any of a variety of parameters, such as BS operating characteristics (e.g., frequency, bandwidth, etc.), UE reported reference signal received power (RSRP), Doppler estimates, deployment information, computational resources, sensor resources, power resources, antenna resources, and other capabilities. Thus, the particular neural network configuration used at each of the BS and UE may be selected based on a correlation between the particular configurations of these devices and the parameters used to train the corresponding neural network architecture configuration.

[0014] 1 illustrates a wireless communication system 100 employing neural network-facilitated UE positioning according to some embodiments. As illustrated, the wireless communication system 100 is a cellular network including a core network 102 coupled to one or more wide area networks (WANs) 104 or other packet data networks (PDNs), such as the Internet. The wireless communication system 100 further includes one or more base stations 108 (designated as BSs 108-1 and 108-2), each of which supports wireless communication with one or more user equipment (UEs) 110 (designated as UEs 110-1 and 110-2) via one or more wireless communication links 112 (designated as communication links 112-1 and 112-2), which may be unidirectional or bidirectional. In at least some embodiments, each BS 108 is configured to communicate with the user equipment (UEs) 110 via the wireless communication links 112 by radio frequency (RF) signaling using one or more applicable RATs specified by one or more communication protocols or standards. Thus, each BS 108 acts as a radio interface between the UE 110 and the various networks and services provided by the core network 102 and other networks, such as packet-switched (PS) data services, circuit-switched (CS) services, etc. The communication of data or signaling from the BS 108 to the UE 110 is referred to as the "downlink" or "DL," while the communication of data or signaling from the UE 110 to the BS 108 is referred to as the "uplink" or "UL." In at least some embodiments, the BS 108 also includes an inter-base station interface 114, such as an Xn and / or X2 interface, configured to exchange user plane and control plane data between other BSs 108.

[0015] Each BS 108 may use any of a variety of RATs or combinations of RATs, such as operating as a NodeB (or base station (BTS)) in the Universal Mobile Telecommunications System (UMTS) RAT (also known as “3G”), operating as an evolved NodeB (eNodeB) in the Third Generation Partnership Project (3GPP®) Long Term Evolution (LTE) RAT, operating as a 5G NodeB (“gNB”) in the 3GPP Fifth Generation (5G) New Radio (NR) RAT, etc. The UEs 110, in turn, may implement any of a variety of electronic devices operable to communicate with the BSs 108 via the appropriate RAT, including, for example, a mobile phone, a cellular-enabled tablet or laptop computer, a desktop computer, a cellular-enabled video game system, a server, a cellular-enabled appliance, a cellular-enabled automotive communication system, a cellular-enabled smart watch or other wearable device, etc.

[0016] In at least some embodiments, the UE 110 uses one or more positioning technologies, such as GNSS, to obtain high-precision positioning information associated with the UE 110. However, GNSS typically suffers from interference, multipath, and low signal-to-noise ratios in urban and indoor environments. The wireless communication system 100, in at least some embodiments, can supplement or even replace GNSS and similar technologies with one or more other UE positioning technologies, such as RAT-assisted UE positioning, to overcome the challenges associated with GNSS. RAT-assisted UE positioning is based at least in part on signaling or reference signals generated and transmitted, for example, by the BS 108 or the UE 110. Examples of these reference signals include a positioning reference signal (PRS), a channel state information reference signal (CSI-RS), a synchronization / physical broadcast channel block (SS / PBCH), and a sounding reference signal (SRS). RAT-assisted UE positioning, in at least some embodiments, typically involves the BS 108 transmitting a reference signal to the UE 110 (or vice versa). The UE 110 (or BS 108) then performs various measurements on the reference signal. Examples of reference signal measurements include signal strength, reference signal time difference measurement (RSTD) in observed time difference of arrival (OTDoA), uplink time difference of arrival (UTDoA), timing advance (TDAV), angle of arrival (AoA), angle of departure (AoD), round trip time (RTT), etc. The UE 110 (or BS 108) transmits the reference signal measurements to one or more other network components, such as the BS 108 (or a location server), which uses the measurements to calculate an estimate of the UE's location.

[0017] As previously mentioned, RAT-assisted UE positioning in conventional wireless communication systems typically relies on a series of processing stages / blocks that result in excessive complexity, resource consumption, and overhead. Furthermore, conventional RAT-assisted UE positioning techniques generally do not consider UE sensor data when computing a UE position estimate. Accordingly, in at least one embodiment, the BS 108 and the UE 110 implement transmitter (TX) and receiver (RX) processing paths that incorporate one or more neural networks (NNs) that are trained or otherwise configured to facilitate RAT-assisted UE positioning. The NN, in at least one configuration, fuses sensor data from available sensors of the UE 110 with UE reference signal measurements (or signals) to provide more accurate UE positioning than conventional RAT-assisted UE positioning mechanisms. BS 108 generates meaningful UE position estimates. For example, as shown with respect to a RAT-assisted UE positioning path 116 (or, for brevity, "UE positioning path 116") established between one or more BSs 108 and the UE 110, the BS 108 employs a BS position reference TX DNN 120 (denoted as TX DNNs 120-1 and 120-2) or a TX processing path 118 (denoted as processing paths 118-1 and 118-2) with other neural networks. The BS position reference TX DNN 120 has an input configured to receive reference signal information 122 (denoted as information 122-1 and 122-2) for generating a reference signal 138, such as a position reference signal (PRS). The BS position reference TX DNN 120 also includes an output coupled to an RF front end 124 (denoted as RF front ends 124-1 and 124-2) of the BS 108. The BS 108 uses a BS position RX It further uses an RX processing path 126 having a DNN 128 or other neural network. The BS location RX DNN 128 has an input coupled to the RF front end 124 and an output configured to generate a UE location estimate 130.

[0018] The UE 110 uses an RX processing path 132 having a UE position reference RX DNN 134 or other neural network. The UE position reference RX DNN 134 has an input coupled to an RF front-end 136. The input of the UE position reference RX DNN 134 is configured to receive, for example, at least one reference signal 138 (denoted as reference signals 138-1 and 138-2) or other reference signals from one or more BSs 108, local sensor data 140, etc. The UE position reference RX DNN 134 also has an output configured to generate UE measurements and sensor reports 144 based on the input to the UE position reference RX DNN 134. The UE 110 further uses a TX processing path 146 having a UE position feedback TX DNN 148 or other neural network. The UE position feedback TX DNN 148 has an input coupled to the output of the UE position reference RX DNN 134 and further has an output coupled to the RF front-end 136. In at least some embodiments, the serving BS 108-1 (or another cellular network component) configures the UE position reference RX DNN 134 and the UE position feedback TX DNN 148 for the UE 110 based on the serving cell's operating characteristics, the UE-reported RSRP, Doppler estimates, deployment information, etc. The UE 110, in at least some embodiments, receives the particular neural network architecture from the serving BS 108-1 (or other network component) via one or more control messages, such as RRC messages.

[0019] In operation, the BS position reference TX DNN 120, the BS position RX DNN 128, the UE position reference RX DNN 134, the UE position feedback TX DNN 148, or a combination thereof, are trained or otherwise configured together to perform one or more RAT-assisted UE positioning operations. In at least some embodiments, the BS position reference TX DNN 120 receives reference signal information 122 as an input. The BS position reference TX DNN 120 generates a reference signal 138 output from the reference signal information 122 (and any other inputs) suitable for RF transmission to the UE 110 and processing by the UE position reference RX DNN 134 of the UE 110. The reference signal 138 output represents a positioning reference signal, which in at least one embodiment is a downlink position reference signal including symbols dedicated to UE positioning. However, the reference signal 138 output may represent other types of reference signals, such as CSI-RS or SS / PBCH. As part of this joint training or other configuration, the BS position reference TX DNN 120 is, in at least one embodiment, trained or otherwise configured to effectively generate and configure reference signal 138 or other reference signals for transmission by the BS 108 to the UE 110. Thus, the BS position reference TX DNN 120 provides the reference signal 138 as an output to the RF front end 124 of the BS 108. The RF front end 124 processes the output, converts the processed output to an analog signal, and RF transmits 152 ( The analog signal is modulated onto a carrier frequency suitable for RF transmission (shown as 152-1 and 152-2).

[0020] In at least some embodiments, multiple BSs 108 are configured with corresponding BS location reference TX DNNs 120 to generate and transmit reference signals 138 to the UE 110. In such embodiments, a first BS acts as a serving / reference BS 108-1, and the remaining BSs are neighboring BSs 108-2. The serving BS 108-1 can communicate with each neighboring BS 108-2 via the inter-base station interface 114 to configure the BS location reference TX DNN 120-2 for the neighboring BS 108-2. For example, the serving BS 108-1 can configure the BS location reference TX DNN 120-2 for the neighboring BS 108-2 based on the serving cell's operating characteristics (e.g., frequency, bandwidth, etc.), UE-reported RSRP, Doppler estimates, deployment information (e.g., urban / rural deployment, or whether angle estimation should be performed by the BS 108), etc. In another embodiment, if the BS 108 includes multiple antenna arrays, each antenna array may be associated with a BS position reference TX DNN 120. The serving BS 108-1, in at least some embodiments, implements a BS position RX DNN 128 in addition to the BS position reference TX DNN 120.

[0021] At the UE 110, one or more components perform reference signal measurements 142, such as RSRP, RSTD, OTDoA, UTDoA, TDAV, AoA, AoD, or RTT, on a received reference signal 138. The UE 110 provides the reference signal measurements 142 as inputs to the UE position reference RX DNN 134. Alternatively, the RF front end 136 may provide the reference signal 138 (or a representation thereof) as input to the UE position reference RX DNN 134. The UE position reference RX DNN 134 may then calculate one or more reference signal measurements 142 on the reference signal 138. In at least some embodiments, other inputs, such as sensor data 140 from sensors at the UE 110, are simultaneously provided as inputs to the UE position reference RX DNN 134. Examples of sensor data 140 inputs include GPS data, camera data, accelerometer data, IMU data, altimeter data, temperature data, barometer data, object detection sensors (e.g., radar sensors, lidar sensors, imaging sensors, or structured light-based depth sensors), etc. From these inputs, based on joint training or other configuration, the UE location reference RX DNN 134 operates to output UE measurements and sensor reports 144 associated with the UE 110. For example, the UE location reference RX DNN 134 processes the reference signal measurements 142 or the reference signal 138 itself to generate outputs representing the UE measurements and sensor reports 144. In other embodiments, the UE location reference RX DNN 134 also processes the sensor data 140 inputs and fuses the sensor data 140 inputs with the reference signal measurements to generate outputs representing the UE measurements and sensor reports 144.

[0022] The UE position reference RX DNN 134 provides an output representing the UE measurements and sensor reports 144 as an input to the UE position feedback TX DNN 148. From this input, the UE position feedback TX DNN 148 generates an output representing the UE measurements and sensor reports 144 and provides the output to the RF front end 136 of the UE 110. The RF front end 136 transceiver processes the output to generate and transmit (over the air) an RF signal 154 including the UE measurements and sensor reports 144 to the serving BS 108-1. The UE 110 can use various messaging mechanisms, such as the Radio Resource Control (RRC) protocol, the Long Term Evolution (LTE) Positioning Protocol (LPP), etc., to configure and transmit the RF signal 154 to the serving BS 108. Accordingly, the UE position feedback TX DNN 148 of the UE 110 provides the generated output to the RF front end 136, where it is processed, converted to an analog signal, and then transmitted over a carrier frequency suitable for RF transmission to the serving BS 108-1. is modulated by

[0023] At the serving BS 108-1, the RF front end 124 receives the RF signal 154 from the UE 110 and converts the RF signal 154 into a digital signal representing the UE measurements and sensor reports 144. The RF front end 124 then provides the digital signal as an input to the BS location RX DNN 128 of the serving BS 108-1. From this input, and based on joint training or other configuration, the BS location RX DNN 128 operates to output a UE location estimate 130 associated with the UE 110. For example, the BS location RX DNN 128 processes the reference signal measurements and UE sensor data 140 from the received UE measurements and sensor reports 144 as inputs. From these inputs, the BS location RX DNN 128 generates an output representing the location estimate 130 at the UE 110. UE position estimate 130, in at least some embodiments, not only incorporates reference signal measurements 142 provided by UE 110, but also incorporates UE sensor data 140, resulting in a UE position estimate that includes, for example, secondary information such as the UE's local conditions, an indication of the UE's orientation, movement (e.g., rotation, orientation, etc.), and / or the like. Thus, by considering UE sensor data 140, serving BS 108-1 can generate a more accurate and meaningful UE position estimate than conventional RAT-assisted positioning techniques. In at least some embodiments, serving BS 108-1 processes UE position estimate 130 or transmits UE position estimate 130 to one or more other components of wireless communication system 100 for further processing.

[0024] In at least some embodiments, the serving BS 108-1 can receive signals from multiple UEs 110, each containing UE measurement and sensor reports 144 or UE measurement reports (without UE sensor data) associated with a different UE 110. In these embodiments, the BS position RX DNN 128 of the serving BS 108-1 compares UE position estimates 130 calculated for two or more distinct UEs 110 to determine whether the position estimates 130 indicate that the distinct UEs 110 occupy the same space. If the position estimates 130 indicate that two or more distinct UEs 110 occupy the same space, the serving BS 108-1 (or another cellular network component) determines that the BS position RX DNN 128 is experiencing a positioning error and must be refined because multiple distinct objects cannot occupy the same physical space. The serving BS 108-1 (or another cellular network component) proceeds to adjust one or more parameters of the BS position RX DNN 128, such as weights, to correct the identified positioning error.

[0025] Although the described techniques include the BS 108 transmitting a reference signal to the UE 110, the UE 110 can similarly transmit a reference signal to the BS 108. In this configuration, the UE location feedback TX DNN 148 or other TX DNN of the UE 110 is configured similarly to the BS location reference TX DNN 120 of the BS 108 for transmitting a reference signal, such as an SRS. The UE location feedback TX DNN 148 of the UE 110 can also augment the reference signal with sensor data 140 from one or more sensors available to the UE 110 and generate an output representing the augmented reference signal. The UE 110 then transmits the augmented reference signal to the serving BS 108-1. The BS location RX DNN 128-1 of the serving BS 108-1, in at least one configuration, performs one or more measurements on the reference signal received from the UE 110 and calculates a UE position estimate 130 based on the reference signal measurements and the UE sensor data 140 received as part of the augmented reference signal transmitted by the UE. In an alternative embodiment, rather than processing the augmented reference signal or measurements and sensor reports at the BS 108, the TX neural network at the BS 108 may, in at least one configuration, process locally generated augmented reference signal measurements. (or UE-provided measurements and sensor reports) to management component 150. In this embodiment, management component 150 implements an RX neural network configured to process reference signal measurements and UE sensor data received from BS 108 to compute a UE position estimate.

[0026] As previously mentioned and described in more detail herein, both the BS 108 and the UE 110 each employ one or more DNNs or other neural networks that are jointly trained and selected based on context-specific parameters to facilitate the overall RAT-assisted UE positioning process. To manage the joint training, selection, and maintenance of these neural networks, the system 100, in at least one embodiment, further includes a management infrastructure component 150 (or, for brevity, “management component 150”). This management component 150 may include a server or other component within the network infrastructure 106 of the wireless communication system 100, such as, for example, within the core network 102 or within the WAN 104. Furthermore, although shown as separate components in the illustrated example, in at least some embodiments, the BS 108 implements the management component 150. The monitoring functions provided by management component 150 may include, for example, some or all of the following: monitoring the joint training of neural networks; managing the selection of a particular neural network architecture configuration at BS 108 or UE 110 based on their particular capabilities or other component-specific parameters; receiving and processing capability updates for neural network configuration selection; receiving and processing feedback for neural network training or selection;

[0027] As described in more detail below with respect to FIG. 4, management component 150, in some embodiments, maintains a set 412 (FIG. 4) of candidate neural network architecture configurations 414 (FIG. 4). Management component 150 (or other network component) may select a candidate neural network architecture configuration 414 to be used for a particular component in a corresponding RAT-assisted UE positioning path based at least in part on the current capabilities of the component implementing the corresponding neural network, the current capabilities of other components in the transmit chain, the current capabilities of other components in the receive chain, or a combination thereof. These capabilities may include, for example, sensor capabilities, processing resource capabilities, battery / power capabilities, RF antenna capabilities, capabilities of one or more accessories of the component, etc. Information representing these capabilities at BS 108 and UE 110 is obtained by management component 150 and stored in management component 150 as BS capability information 420 (FIG. 4) and UE capability information 422 (FIG. 4), respectively. The management component 150 may further consider parameters of the corresponding channel or propagation channel or other aspects of the environment, such as the carrier frequency of the channel, the known presence of objects or other interference, etc.

[0028] To support this approach, in some embodiments, management component 150 can manage the joint training of different combinations of candidate neural network architecture configurations 414 in different capability / context combinations. Management component 150 can then obtain capability information 420 from BS 108, capability information 422 from UE 110, or both, from which management component 150 selects, for each component, a neural network architecture configuration from set 412 of candidate neural network architecture configurations 414 based at least in part on the corresponding indicated capabilities, RF signaling environment, etc. In at least some embodiments, management component 150 (or other network component) manages the joint training of different combinations of candidate neural network architecture configurations 414 in different capability / context combinations. The configurations are jointly trained as paired subsets, such that each candidate neural network architecture configuration for a particular capability set at BS 108 is jointly trained with a single corresponding candidate neural network architecture configuration for a particular capability set at UE 110. In other embodiments, management component 150 (or other network component) determines the candidate neural network architecture configurations such that each candidate configuration at BS 108 has a one-to-many correspondence with multiple candidate configurations at UE 110, and vice versa.

[0029] Thus, system 100 utilizes a RAT-assisted UE positioning approach that relies on a set of neural networks that are managed, jointly trained, and selectively used between one or more BSs 108 and one or more UEs 110 for UE positioning, rather than independently designed process blocks that may not be specifically designed with compatibility in mind. This not only provides greater flexibility but also, in some circumstances, allows for faster processing at each device, more accurate UE position estimates, and more efficient transmission and processing of reference signals and UE measurements and sensor reports.

[0030] 2 illustrates an example hardware configuration for a UE 110 according to some embodiments. Note that the depicted hardware configuration represents the processing and communication components most directly related to the neural network-based processes of one or more embodiments, and omits certain components that are well understood to be frequently implemented in such electronic devices, such as displays, non-sensor peripherals, external power sources, etc.

[0031] In the illustrated configuration, the UE 110 includes an RF front end 136 having one or more antennas 202 and an RF antenna interface 204 having one or more modems supporting one or more RATs. The RF front end 136 effectively acts as a physical (PHY) transceiver interface, performing and processing signaling between the one or more processors 206 of the UE 110 and the antennas 202 to facilitate various types of wireless communications. The antennas 202 may be arranged in one or more arrays of similarly or differently configured antennas and may be tuned to one or more frequency bands associated with corresponding RATs. The one or more processors 206 may include, for example, one or more central processing units (CPUs), graphics processing units (GPUs), tensor processing units (TPUs), or other application-specific integrated circuits (ASICs). Illustratively, the processors 206 may include an application processor (AP) utilized by the UE 110 to run an operating system and various user-level software applications, as well as one or more processors utilized by a modem or baseband processor of the RF front end 136. The UE 110 further includes one or more computer-readable media 208, which may include any of a variety of media used by electronic devices to store data and / or executable instructions, such as random access memory (RAM), read-only memory (ROM), cache, flash memory, solid-state drive (SSD), or other mass storage device. For ease of explanation and brevity, the computer-readable media 208 will be referred to herein as “memory 208” in light of the frequent use of system memory or other memory to store data and instructions for execution by the processor 206, although it will be understood that references to “memory 208” apply to other types of storage media as well, unless otherwise noted.

[0032] In at least one embodiment, the UE 110 further includes a plurality of sensors, referred to herein as a sensor set 210, at least some of which may be implemented in one or more The sensors in sensor set 210 are utilized in the neural network-based scheme of the embodiment. Generally, sensors in sensor set 210 include sensors that sense some aspect of the UE 110's environment or the user's use of the UE 110, potentially sensing parameters that have at least some influence on or reflect the UE 110's position, orientation, movement, or a combination thereof, for example. The sensors in sensor set 210 may include one or more sensors for object detection, such as a radar sensor, a lidar sensor, an imaging sensor, a structured light-based depth sensor, etc. Sensor set 210 may also include one or more sensors for determining the position or attitude / orientation of UE 110, such as a satellite positioning sensor such as a GPS sensor, a global navigation satellite system (GNSS) sensor, an internal measurement unit (IMU) sensor, a visual odometry sensor, a gyroscope, a tilt sensor or other inclinometer, an ultra-wideband (UWB)-based sensor, etc. Other examples of types of sensors in sensor set 210 may include environmental sensors such as a temperature sensor, a barometer, an altimeter, or an imaging sensor such as a camera for image capture by a user, a camera for face detection, a camera for stereoscopic vision or visual odometry, a light sensor for detecting objects in proximity to features of the device, an object detection sensor (e.g., a radar sensor, a lidar sensor, an imaging sensor, or a structured light-based depth sensor). UE 110 may further include one or more batteries 212 or other portable power source, as well as one or more user interface (UI) components 214, such as a touchscreen, a user-operable input / output device (e.g., a “button” or keyboard) or other touch / contact sensor, a microphone or other audio sensor for capturing audio content, an image sensor for capturing video content, a thermal sensor (e.g., for detecting proximity to a user), etc.

[0033] The one or more memories 208 of the UE 110 store one or more sets of executable software instructions and associated data that operate the one or more processors 206 and other components of the UE 110 to perform various functions ascribed to the UE 110. The set of executable software instructions may include, for example, an operating system (OS) and various drivers (not shown), as well as various software applications. The set of executable software instructions further includes one or more of a neural network management module 216, a capabilities management module 218, or a reference signal measurement module 220. The neural network management module 216 implements one or more neural networks in the UE 110, which are described in detail below. The capabilities management module 218 determines various capabilities of the UE 110 that may relate to neural network configuration or selection and reports such capabilities to the management component 150, as well as monitors the UE 110 for changes in such capabilities, including changes in RF and processing capabilities, changes in accessory availability or capabilities, changes in sensor availability, etc., and manages reporting of such capabilities and capability changes to the management component 150. Similar to that described above, the reference signal measurement module 220 operates to generate signal measurements such as RSRP, RSTD, OTDoA, UTDoA, TDAV, AoA, AoD, RTT, etc., for reference signals received from one or more BSs 108.

[0034] To facilitate operation of the UE 110, the one or more memories 208 of the UE 110 may further store data associated with these operations. This data may include, for example, device data 222 and one or more neural network architecture configurations 224. The device data 222 may represent, for example, user data, multimedia data, beamforming codebooks, software application configuration information, etc. The device data 222 may further include, with respect to the UE 110, capability information such as sensor capability information for one or more sensors of the sensor set 210, including the presence or absence of particular sensors or sensor types, and, for those sensors that are present, one or more indications of their corresponding capabilities, e.g., labeling. The capability information may further include information regarding, for example, the capability or status of the battery 212, the capability or status of the UI 214 (e.g., the screen resolution, color gamut, or frame rate of the display), etc.

[0035] The one or more neural network architecture configurations 224 represent UE implementations selected from a set 412 of candidate neural network architecture configurations 414 maintained by the management component 150. Each neural network architecture configuration 224 includes one or more data structures containing data and other information representing a corresponding architecture and / or parameter configuration used by the neural network management module 216 to form a corresponding neural network for the UE 110. Information included in the neural network architecture configurations 224 may include, for example, parameters specifying a fully connected layer neural network architecture, a convolutional layer neural network architecture, recurrent neural network layers, the number of connected hidden neural network layers, the input layer architecture, the output layer architecture, the number of nodes used in the neural network, the coefficients (e.g., weights and biases) used in the neural network, kernel parameters, the number of filters used in the neural network, the stride / pooling configuration used in the neural network, the activation function for each neural network layer, the interconnections between neural network layers, which neural network layers to skip, etc. Thus, the neural network architecture configuration 224 includes any combination of NN formation components (e.g., architecture and / or parameter configurations) to create a NN formation configuration (e.g., a combination of one or more NN formation components) that defines and / or forms a DNN.

[0036] 3 illustrates an example hardware configuration for a BS108 according to some embodiments. Note that the depicted hardware configuration represents the processing and communication components most directly related to the neural network-based processes of one or more embodiments, and omits certain components that are well understood to be frequently implemented in such electronic devices, such as displays, non-sensor peripherals, external power sources, etc. Furthermore, while the illustrated diagram represents an implementation of the BS108 as a single network node (e.g., a 5G NR NodeB, or "gNB"), it should be noted that the functionality, and therefore the hardware components, of the BS108 may instead be distributed across multiple network nodes or devices, and may be distributed in a manner that performs the functionality of one or more embodiments.

[0037] In the illustrated configuration, the BS 108 includes an RF front end 124 having one or more antennas 302 and an RF antenna interface (or front end) 304 having one or more modems supporting one or more RATs. The RF antenna interface operates as a PHY transceiver interface to perform and process signaling between one or more processors 306 of the BS 108 and the antennas 302 to facilitate various types of wireless communications. The antennas 302 may be arranged in one or more arrays of similarly or differently configured antennas and may be tuned to one or more frequency bands associated with corresponding RATs. The one or more processors 306 may include, for example, one or more CPUs, GPUs, TPUs, or other ASICs. The BS 108 further includes one or more computer-readable media 308, including any of a variety of media used by electronic devices to store data and / or executable instructions, such as RAM, ROM, cache, flash memory, SSD, or other mass storage device. Similar to the memory 208 of the UE 110, for ease of explanation and brevity, the computer-readable media 308 may be referred to as a memory. The computer-readable medium 308 is referred to herein as "memory 308" in light of the frequent use of system memory or other memory to store data and instructions for execution by the processor 306, although it can be understood that references to "memory 308" apply to other types of storage media as well unless otherwise specified.

[0038] In at least one embodiment, the BS 108 further includes a plurality of sensors, referred to herein as sensor set 310, at least some of which are utilized in the neural network-based scheme of one or more embodiments. Generally, the sensors in sensor set 310 include sensors that sense some aspect of the BS 108's environment and have the potential to sense parameters that have at least some impact on or reflect the BS 108's RF propagation path or RF transmission / reception performance by the BS 108 to the corresponding UE 110. The sensors in sensor set 310 may include one or more sensors for object detection, such as a radar sensor, a lidar sensor, an imaging sensor, a structured light-based depth sensor, etc. If the BS 108 is a mobile BS, the sensor set 310 may also include one or more sensors for determining the BS 108's position or attitude / orientation. Other examples of types of sensors in sensor set 310 may include imaging sensors, optical sensors for detecting objects near features of the BS 108, etc.

[0039] The one or more memories 308 of the BS 108 store one or more sets of executable software instructions and associated data that operate the one or more processors 306 and other components of the BS 108 to perform various functions ascribed to the BS 108 of one or more embodiments. The set of executable software instructions includes, for example, an OS and various drivers (not shown), as well as various software applications. The set of executable software instructions further includes one or more of a neural network management module 314, a reference signal management module 316, a UE positioning management module 318, or a capability management module 320.

[0040] The neural network management module 314 implements one or more neural networks for the BS 108, as described in more detail below. The reference signal management module 316 manages the generation and transmission of one or more reference signals, which in some embodiments are based on one or more neural networks implemented by the neural network management module 314. The UE positioning management module 318 manages the generation of UE position estimates, which in some embodiments are based on one or more neural networks implemented by the neural network management module 314. The capabilities management module 320 determines various capabilities of the BS 108 that may be related to neural network configuration or selection and reports such capabilities to the management component 150, as well as monitors the BS 108 for changes in such capabilities, including changes in RF and processing capabilities, and manages the reporting of such capabilities and changes in capabilities to the management component 150.

[0041] To facilitate the operation of the BS 108, the one or more memories 308 of the BS 108 may further store data associated with these operations. This data may include, for example, BS data 322 and one or more neural network architecture configurations 324. The BS data 322 may represent, for example, a beamforming codebook, software application configuration information, etc. The BS data 322 may further include capability information, such as sensor capability information for one or more sensors of the sensor set 310, including the presence or absence of a particular sensor or sensor type, and, for those sensors that are present, one or more indications of their corresponding capabilities, such as the range and resolution of a lidar or radar sensor, the image resolution and color depth of an imaging camera, etc. One or more neural network architecture configurations 324 may further include, for example, BS data 322 and one or more neural network architecture configurations 324. The BS data 322 may represent, for example, a beamforming codebook, software application configuration information, etc. The BS data 322 may further include, for the BS 108, capability information, such as sensor capability information for one or more sensors of the sensor set 310, including the presence or absence of a particular sensor or sensor type, and, for those sensors that are present, one or more indications of their corresponding capabilities, such as the range and resolution of a lidar or radar sensor, the image resolution and color depth of an imaging camera, etc. The neural network architecture configurations 324 represent example BS implementations selected from the set 412 of candidate neural network architecture configurations 414 maintained by the management component 150. Thus, similar to the neural network architecture configurations 224 of FIG. 2, each neural network architecture configuration 324 includes one or more data structures containing data and other information representing the corresponding architecture and / or parameter configuration used by the neural network management module 314 to form the corresponding neural network of the BS 108.

[0042] 4 illustrates an example hardware configuration for management component 150 according to some embodiments. Note that the depicted hardware configuration represents the processing and communication components most directly related to the neural network-based processes of one or more embodiments, and omits certain components that are well understood to be frequently implemented in such electronic devices. Furthermore, while the hardware configuration is shown as being located in a single component, the functionality of management component 150, and therefore the hardware components, may instead be distributed across multiple infrastructure components or nodes, and may be distributed in a manner that performs the functionality of one or more embodiments.

[0043] As previously mentioned, any of a variety of components or combinations of components within network infrastructure 106 may implement management component 150. For ease of explanation, management component 150 is described with reference to an example implementation as a server or other component in one of core networks 102, although in other embodiments management component 150 may be implemented as part of BS 108, for example.

[0044] As shown, management component 150 includes one or more network interfaces 402 (e.g., Ethernet interfaces) for coupling to one or more networks of system 100, one or more processors 404 coupled to the one or more network interfaces 402, and one or more non-transitory computer-readable storage media 406 (referred to herein as “memory 406” for brevity) coupled to the one or more processors 404. The one or more memories 406 store one or more sets of executable software instructions and associated data that operate the one or more processors 404 and other components of management component 150 to perform various functions of one or more embodiments attributed to management component 150. The sets of executable software instructions include, for example, an OS and various drivers (not shown). The software stored in the one or more memories 406 may further include one or more of a training module 408 or a neural network selection module 410. The training module 408 operates to manage the joint training of candidate neural network architecture configurations 414 in a set 412 of candidate neural networks available for use at transmitting and receiving devices in the UE positioning path using one or more training datasets 416. The training may include training the neural networks while offline (i.e., not actively involved in processing communications) and / or online (i.e., while actively involved in processing communications). Furthermore, the training may be individual or separate, whereby each neural network is trained individually based on its own training dataset without the results being communicated to or otherwise affecting the DNN training at the opposite end of the transmission path, or the training may be joint, whereby neural networks in the data stream transmission path are jointly trained based on the same or complementary datasets. .

[0045] The neural network selection module 410 operates to obtain, filter, and otherwise process selection-related information 418 from one or both of the BS 108 and the UE 110 in the RAT-assisted UE positioning path and use this selection-related information 418 to select a pair of jointly trained neural network architecture configurations 414 from the candidate set 412 for implementation at the transmitting and receiving devices in the RAT-assisted UE positioning path. As previously mentioned, this selection-related information 418 may include, for example, one or more of BS capability information 420 or UE capability information 422, current propagation path information, channel-specific parameters, etc. After the neural network selection module 410 makes a selection, the neural network selection module 410 then begins transmitting an indication of the selected neural network architecture configuration 414 for each network component, such as via transmitting an index number associated with the selected configuration, transmitting one or more data structures representing the neural network architecture configuration itself, or a combination thereof.

[0046] 5 illustrates an example machine learning (ML) module 500 for implementing a neural network according to some embodiments. At least one of the BS 108 and the UE 110 in the UE positioning path 116 implements one or more DNNs or other neural networks for one or more of transmitting reference signals, performing measurements on the reference signals, fusing the reference signal measurements with UE sensor data, generating UE measurements and sensor reports, and generating UE positioning estimates. Thus, the ML module 500 illustrates an example module for implementing one or more of these neural networks.

[0047] In the illustrated example, the ML module 500 implements at least one deep neural network (DNN) 502 with a group of connected nodes (e.g., neurons and / or perceptrons) organized into three or more layers. The nodes between layers can be configured in various ways, such as a partially connected configuration in which a first subset of nodes in a first layer is connected to a second subset of nodes in a second layer, or a fully connected configuration in which each node in the first layer is connected to each node in the second layer. Neurons process input data and generate continuous output values, such as any real number between 0 and 1. In some cases, the output value indicates how close the input data is to a desired category. Perceptrons perform linear classification, such as binary classification, on the input data. Nodes, whether neurons or perceptrons, can generate output information based on adaptive learning using various algorithms. Using the DNN 502, the ML module 500 performs a variety of different types of analysis, including single linear regression, multiple linear regression, logistic regression, stepwise regression, binary classification, multi-class classification, multivariate adaptive regression splines, locally estimated scatterplot smoothing, and the like.

[0048] In some embodiments, the ML module 500 adaptively learns based on supervised learning. In supervised learning, the ML module 500 receives various types of input data as training data. The ML module 500 processes the training data to learn how to map the inputs to desired outputs. As an example, when implemented in a BS position reference signal TX mode, the ML module 500 receives as input one or more of reference signals, such as a PRS, capability information of the BS 108, capability information of the UE 110, operating environment characteristics of the BS 108, or operating environment characteristics of the UE 110, and learns how to map this input training data to one or more output reference signals configured to be transmitted to the UE 110. As another example, when implemented in a UE position reference signal RX mode, the ML module 500 receives as input one or more of the following reference signals: an indication of the received reference signal, a UE reference signal measurement, a UE sensor signal, or a UE position reference signal. The ML module 500 receives as input one or more of the following as inputs: UE sensor data, UE reference signal measurements, BS location information, UE reference signal measurements, and the like; and learns how to map this input training data to an output that represents the UE measurements and sensor reports and fuses the UE sensor data and the UE reference signal measurements. In another example, when implemented in a UE location feedback TX mode, the ML module 500 receives as input outgoing UE measurements and sensor reports and learns how to generate an output that is, e.g., at least channel coded, suitable for wireless transmission over an RF antenna interface. As yet another example, when implemented in a BS location RX mode, the ML module 500 receives as input one or more of the following: UE sensor data, UE reference signal measurements, BS location information, UE location information, and the like; and learns how to generate an output that represents at least one UE location estimate. In at least some embodiments, the training process trains the ML module 500 to minimize the mean squared error (MSE) between the estimated UE location and the UE's actual location. Additionally, training in either or both of the TX mode or the RX mode may further include training using sensor data as input, capability information as input, RF antenna configuration as input, other operational parameter information, and the like.

[0049] During the training procedure, the ML module 500 uses labeled or known data as input to the DNN 502. The DNN 502 analyzes the input using nodes and generates corresponding outputs. The ML module 500 compares the corresponding outputs with the truth data and adapts the algorithms implemented by the nodes to improve the accuracy of the output data. The DNN 502 then applies the adapted algorithm to the unlabeled input data to generate corresponding output data. The ML module 500 uses one or both of statistical analysis and adaptive learning to map inputs to outputs. For example, the ML module 500 uses characteristics learned from the training data to correlate unknown inputs with outputs that are statistically likely to be within a threshold range or value. This allows the ML module 500 to receive complex inputs and identify corresponding outputs. In some embodiments, the training process trains the ML module 500 on characteristics of communications transmitted over a wireless communication system (e.g., time / frequency interleaving, time / frequency deinterleaving, convolutional coding, convolutional decoding, power levels, channel equalization, inter-symbol interference, quadrature amplitude modulation / demodulation, frequency division multiplexing / demultiplexing, transmission channel characteristics) that match the characteristics of the data encoding / decoding schemes used in such systems, thereby enabling the trained ML module 500 to receive samples of a signal as input and recover information from the signal, such as binary data embedded in the signal.

[0050] In the illustrated example, the DNN 502 includes an input layer 504, an output layer 506, and one or more hidden layers 508 positioned between the input layer 504 and the output layer 506. Each layer has any number of nodes, and the number of nodes between layers may be the same or different. That is, the input layer 504, for example, may have the same and / or a different number of nodes as the output layer 506, the output layer 506 may have the same and / or a different number of nodes as the one or more hidden layers 508, etc.

[0051] Node 510 corresponds to one of several nodes included in input layer 504, which perform separate and independent computations. As will be further described, the node receives input data, processes the input data using one or more algorithms, and generates output data. Typically, the algorithms include weights and / or coefficients that change based on adaptive learning. Thus, the weights and / or coefficients reflect information learned by the neural network. Each node can optionally decide whether to pass the processed input data to one or more subsequent nodes. For illustrative purposes, after processing the input data, node 510 passes the processed input data to nodes 512 and 513 in hidden layer 508. Alternatively or additionally, node 510 may pass the processed input data to a node based on the layer connection architecture. This process may be repeated across multiple layers until DNN 502 generates an output using a node in the output layer 506 (e.g., node 516).

[0052] Neural networks can also employ various architectures that determine which nodes in the neural network are connected, how data is advanced and / or retained within the neural network, the weights and coefficients the neural network uses to process input data, how the data is processed, and the like. These various elements collectively describe a neural network architectural configuration, such as the neural network architectural configuration briefly described above. By way of example, a recurrent neural network, such as a long short-term memory (LSTM) neural network, forms cycles between node connections to retain information from previous portions of an input data sequence. The recurrent neural network then uses the retained information for subsequent portions of the input data sequence. As another example, a feedforward neural network passes information forward to a forward connection without forming a cycle that retains information. While described in the context of node connections, it should be understood that a neural network architectural configuration can include various parameter configurations that affect how the DNN 502 or other neural network processes input data.

[0053] The neural network architecture configuration of a neural network can be characterized by various architecture and / or parameter configurations. For illustrative purposes, consider an example in which the DNN 502 implements a convolutional neural network (CNN). Generally, a convolutional neural network corresponds to a type of DNN in which layers filter input data by processing the data using convolution operations. Thus, the CNN architecture configuration can be characterized by, for example, pooling parameters, kernel parameters, weights, and / or layer parameters.

[0054] Pooling parameters correspond to parameters that specify a pooling layer within a convolutional neural network that reduces the dimensionality of input data. For example, a pooling layer can combine the outputs of nodes in a first layer with the inputs of nodes in a second layer. Alternatively or additionally, pooling parameters specify where and how the neural network pools data within the data processing layer. For example, a pooling parameter indicating "max pooling" configures the neural network to select and pool the maximum values ​​from a group of data generated by nodes in the first layer and use the maximum values ​​as input to a single node in the second layer. A pooling parameter indicating "average pooling" configures the neural network to generate an average value from a group of data generated by nodes in the first layer and use the average value as input to a single node in the second layer.

[0055] The kernel parameters indicate the filter size (e.g., width and height) used to process the input data. Alternatively or additionally, the kernel parameters specify the type of kernel method used to filter and process the input data. For example, a support vector machine corresponds to a kernel method that uses regression analysis to identify and / or classify data. Other types of kernel methods include Gaussian processes, canonical correlation analysis, spectral clustering techniques, etc. Thus, the kernel parameters can indicate the filter size and / or type of kernel method to apply in the neural network. The weight parameters specify the weights and biases used by the algorithm in the node to classify the input data. In some implementations, the weights and biases are learned parameter configurations, such as parameter configurations generated from training data. The layer parameters are used to determine the weights and biases of all the first layer (e.g., output layer 506). The layer parameters specify layer connections and / or layer types, such as a fully connected layer type indicating that a node is connected to all nodes in a second layer (e.g., hidden layer 508), a partially connected layer type indicating which nodes in the first layer are disconnected from the second layer, an activation layer type indicating which filters and / or layers are active within the neural network, etc. Alternatively or additionally, the layer parameters specify the type of node layer, such as a normalization layer type, a convolutional layer type, a pooling layer type, etc.

[0056] Although described with reference to pooling parameters, kernel parameters, weight parameters, and layer parameters, it can be understood that other parameter configurations can be used to form DNNs consistent with the guidelines provided herein. Thus, a neural network architecture configuration can include any suitable type of configuration parameter that can be applied by a DNN that affects how the DNN processes input data and generates output data.

[0057] The architectural configuration of the ML module 500 may be based on the capabilities of the node implementing the ML module 500, the capabilities (including sensors) of one or more nodes upstream or downstream of the node implementing the ML module 500, or a combination thereof. For example, the UE 110 may have one or more sensors enabled or disabled or may have limited battery power; thus, the ML modules 500 in both the UE 110 and the BS 108 may be trained based on the different sensor configuration of the UE 110 or battery power as input, e.g., to facilitate the ML modules 500 at both ends to take advantage of RAT-assisted UE positioning techniques that are more suitable for the different sensor configuration of the UE 110 or lower power consumption.

[0058] Thus, in some embodiments, a device implementing ML module 500 may be configured to implement different neural network architecture configurations for different combinations of capability parameters, sensor parameters, RF environment parameters, operational parameters, etc. For example, the device may have access to one or more neural network architecture configurations for use when an imaging camera is available to UE 110 and may have access to a different set of one or more neural network architecture configurations for use when an imaging camera is not available to UE 110.

[0059] In at least some embodiments, a device implementing the ML module 500 locally stores some or all of a set of candidate neural network architecture configurations that the ML module 500 can use. For example, a component may receive as input one or more parameters, such as one or more BS capability parameters, one or more UE capability parameters, one or more BS operational parameters, one or more UE operational parameters, or one or more channel parameters, and index the candidate neural network architecture configurations by a look-up table (LUT) or other data structure that outputs an identifier associated with a corresponding locally stored candidate neural network architecture configuration that is suitable for operation given the input parameters. However, in some embodiments, the neural network used at the BS 108 and the neural network used at the UE 110 are jointly trained, and therefore, a mechanism may need to be employed between the BS 108 and the UE 110 to help each device select a neural network architecture configuration for its ML module 500 that is jointly trained, or at least operationally compatible, with the neural network architecture configuration selected by the other device for its complementary ML module 500. This mechanism may include, for example, coordination signaling sent between the BS 108 and the UE 110 directly or via the management component 150, or the management component 150 may be configured by each device. The proposed subset can then act as a referee to select a pair of compatible co-trained architecture configurations from the proposed subset.

[0060] However, in other embodiments, it may be more efficient or otherwise advantageous to have management component 150 operate to select an appropriate pair of co-trained neural network architecture configurations to be used in the corresponding ML modules 500 of the sending and receiving devices. In this approach, management component 150 obtains information from the sending and receiving devices representing some or all of the parameters that may be used in the selection process, and from this information, selects a pair of co-trained neural network architecture configurations 414 from a set 412 of such configurations maintained by management component 150. Management component 150 (or other network component) may implement this selection process using, for example, one or more algorithms, LUTs, etc. Management component 150 may then send to each device an identifier or other indication of the neural network architecture configuration selected for that device's ML module 500 (if each device has a locally stored copy), or management component 150 may send one or more data structures representing the neural network architecture configuration selected for that device.

[0061] To facilitate the process of selecting an appropriate pair of neural network architecture configurations for the transmitting and receiving devices, in at least one embodiment, management component 150 trains ML module 500 in the UE positioning path using an appropriate combination of neural network management and training modules. Training can occur offline when no active communication exchange is taking place, or online during an active communication exchange. For example, management component 150 can mathematically generate training data, access files that store training data, acquire real-world communication data, etc. Management component 150 then extracts and stores the various learned neural network architecture configurations for subsequent use. Some implementations store input characteristics with each neural network architecture configuration, whereby the input characteristics describe various characteristics of one or both of the BS 108 or UE 110 operating characteristics and capability configurations corresponding to the respective neural network architecture configuration. In an embodiment, the neural network manager selects a neural network architecture configuration by adapting the current operating environment of one or more of the BS108 or UE110 to the input characteristics, where the current operating environment includes an indication of the capabilities of one or more nodes along the training UE positioning path, such as sensor capabilities, RF capabilities, processing capabilities, etc.

[0062] As previously mentioned, network devices in wireless communication, such as the BS 108 and the UE 110, may be configured to process wireless communication exchanges using one or more DNNs at each network device, with each DNN replacing and / or adding new functionality to one or more functions traditionally implemented by one or more hard-coded or fixed design blocks to facilitate the RAT-assisted UE positioning process. Additionally, each DNN may further incorporate current sensor data from one or more sensors of the networked device's sensor set and / or capability data from some or all nodes in the UE positioning path 116, effectively modifying or otherwise adapting its operation to account for the current operating environment.

[0063] To this end, Figures 6 and 7 together illustrate an exemplary operating environment 600 for a DNN implementation in the exemplary UE positioning path 116 of Figure 1. In the illustrated example, operating environment 600 employs a neural network-based approach to facilitate RAT-assisted UE positioning. In at least one embodiment, the neural network management module 314 of one or more BSs 108 implements a BS position reference signal TX processing module 602 (denoted as TX processing modules 602-1 and 602-2), while the neural network management module 216 of the UE 110 implements a UE position reference signal receiver (RX) processing module 604. The neural network management module 216 of the UE 110 further implements a UE position feedback TX processing module 702, while the neural network management module 314 of the serving BS 108-1 further implements a BS position RX processing module 704.

[0064] In at least one embodiment, each of these processing modules implements one or more DNNs via a corresponding ML module implementation, as described above in connection with one or more DNNs 502 of ML module 500 of FIG. 5. Thus, the BS position reference signal TX processing module 602 of one or more BSs 108 and the UE position reference signal RX processing module 604 of UE 110 interoperate to support a downlink neural network-based wireless communication path between BS 108 and UE 110 and generate and communicate data to facilitate RAT-assisted UE positioning. Similarly, the UE position feedback TX processing module 702 of UE 110 and the BS position RX processing module 704 of serving BS 108-1 interoperate to support an uplink neural network-based wireless communication path between UE 110 and serving BS 108-1 and generate and communicate data to facilitate RAT-assisted UE positioning.

[0065] One or more DNNs of the BS position reference signal TX processing module 602 of at least one BS 108 are trained to receive as input reference signal information 122 (denoted as information 122-1 and 122-2) from the BS reference signal management module 316 (denoted as modules 316-1 and 316-2). In one example, the BS position reference signal TX processing module 602 receives the reference signal information 122 as input in response to a component, such as a UE 110, a location management server (not shown), or a remote application, requesting UE location information. The reference signal information 122, in at least some embodiments, includes one or more different types of information that the DNN of the BS position reference signal TX processing module 602 utilizes as input to generate and configure one or more reference signals. Examples of reference signal information 122 include reference signal-related parameters or attributes such as transmit power, antenna mapping, number of physical downlink control channel (PDCCH) symbols, number of transmissions in consecutive downlink subframes, PRS bandwidth, PRS transmission time offset, PRS configuration index, PRS periodicity, PRS subframe offset, PRS muting sequence, PRS muting sequence length, time domain behavior, time / frequency resource element density, quasi-co-location (QCL) information, RX panel information of the UE 110, etc. Other examples of reference signal information 122 include serving cell operating characteristics (e.g., frequency, bandwidth, etc.), UE reported reference signal received power (RSRP), Doppler estimate, deployment information (e.g., urban / rural deployment or whether angle estimation should be performed by the BS 108), UE capability information, etc.

[0066] From the reference signal information 122 input, one or more DNNs in the BS position reference signal TX processing module 602 are trained to generate and configure one or more corresponding reference signal 138 outputs (denoted as outputs 138-1 and 138-2), such as PRS outputs. For example, the BS position reference signal TX processing module 602 generates and configures the reference signal 138 to include certain parameters or characteristics (e.g., bandwidth, resource or resource set, repetition, periodicity, interference suppression, etc.) based on the reference signal information 122 input. The BS 108's RF antenna interface 304 (denoted as interfaces 304-1 and 304-2) and one or more antennas 302 (denoted as antennas 302-1 and 302-2) are connected to the RF antenna interface 304 of the BS 108. 2-1 and 302-2) convert the reference signal 138 output into a corresponding RF signal 606 (denoted as RF signals 606-1 and 606-2) that is transmitted over the air for reception by the UE 110. In particular, in some embodiments, one or more DNNs of the BS position reference signal TX processing module 602 are trained to provide processing that effectively results in a structured and modulated reference signal for transmission by the BS 108 to the UE 110, with such processing being trained to the one or more DNNs via joint training rather than requiring laborious and inefficient hard-coding of algorithms or separate discrete processing blocks to perform the reference signal generation, structure, and modulation.

[0067] An RF signal 606 is received and processed at the UE 110 via one or more antennas 202 and RF antenna interface 204, and the resulting acquired signal 608 is analyzed by a reference signal measurement module 220 to generate one or more reference signal measurements 142, such as RSRP, RSTD, OTDoA, UTDoA, TDAV, AoA, AoD, or RTT. One or more DNNs in a UE position reference signal RX processing module 604 of the UE 110 are trained to receive the reference signal measurements 142 and other inputs as inputs and generate corresponding UE measurement and sensor report 144 outputs from these inputs. In at least some embodiments, the UE position reference signal RX processing module 604 does not receive other inputs, and as a result, the UE position reference signal RX processing module 604 generates UE measurement reports (rather than UE measurements and sensor reports). The UE position reference signal RX processing module 604 may also receive the acquisition signal 608 signal as an input and calculate the reference signal measurements 142 thereon, as compared to receiving the reference signal measurements 142 from the reference signal measurement module 220.

[0068] Other inputs provided to the UE position reference signal RX processing module 604 may include, for example, sensor data 140 from the sensor set 210. Examples of sensor data 140 inputs include GPS data, camera data, accelerometer data, IMU data, altimeter data, temperature data, barometer data, object detection data (e.g., radar data, lidar data, imaging sensor data, structured light-based depth sensor data, etc.). Additionally, it may be understood that the capabilities of the UE 110, including available sensors, may change over time. For example, the UE 110 may disable one or more sensors based on the UE 110's current battery level, thermal state, or other condition. To compensate for the changing sensor capabilities, one or more DNNs of the RX processing module 604 may be trained with different sensor data 140 inputs to provide UE measurement and sensor report 144 outputs that account for the UE 110's different sensor capabilities. Thus, in some embodiments, one or more DNNs of the UE position reference signal RX processing module 604 are trained to provide processing that effectively results in UE measurements and sensor reports 144 that fuse sensor data 140 from available sensors of the UE 110 with UE reference signal measurements 142, and such processing is trained to one or more DNNs via joint training rather than requiring laborious and inefficient hard-coding of algorithms or separate discrete processing blocks to implement the same process.

[0069] As described in the example shown in FIG. 7 , the UE position reference signal RX processing module 604 provides the UE measurement and sensor report 144 output as input to the UE position feedback TX processing module 702 of the UE 110, which generates from this input a corresponding output signal 706 representative of the UE measurement and sensor report 144. The RF antenna interface 204 and one or more antennas 202 convert the output signal 706 into a corresponding RF signal 708 representative of a wireless communication to be transmitted over the air for reception by the serving BS 108-1. The UE 110 uses a radio resource control (RRC) protocol to configure and transmit the wireless communication. Various messaging mechanisms may be used, such as RFC 2224, Long Term Evolution (LTE) Positioning Protocol (LPP), etc. In particular, in some embodiments, one or more DNNs of the UE position feedback TX processing module 702 are trained to provide processing that effectively results in at least a channel-coded (including modulated) representation of the input UE measurements and sensor reports 144 suitable for wireless transmission over, for example, the RF antenna interface 204, with such processing being trained to the one or more DNNs via joint training rather than requiring laborious and inefficient hard-coding of algorithms or separate discrete processing blocks to implement the same process.

[0070] The RF signals 708 propagated from the UE 110 are received and initially processed by the serving BS 108-1's antenna 302-1 and RF antenna interface 304-1, for example, to convert the RF signals 708 into digital signals representing the UE measurements and sensor reports 144. One or more DNNs in the BS position RX processing module 704 receive as input the resulting outputs 710 of the RF antenna interface 304 representing the UE measurements and sensor reports 144 and are trained to generate from this input a corresponding UE position estimate 130. For example, the one or more DNNs in the BS position RX processing module 704 receive as input the UE position estimate 130, which includes the UE reference signal measurements 142 and, in some embodiments, the UE sensor data 140. From these inputs, the one or more DNNs in the BS position RX processing module 704 generate an output 712 representing the position estimate 130 at the UE 110. In at least some embodiments, the UE position estimate 130 not only incorporates reference signal measurements 142 provided by the UE 110, but also incorporates UE sensor data 140, resulting in a UE position estimate 130 that includes, for example, the geographic location of the UE 110, the local conditions of the UE 110, secondary information such as an indication of the UE's orientation, movement (e.g., rotation, orientation, etc.), etc. In particular, in some embodiments, one or more DNNs of the BS position RX processing module 704 are trained to provide processing that effectively results in an output representing the UE position estimate 130 based on, for example, the UE reference signal measurements 142 fused with the UE sensor data 140, with such processing being trained to the one or more DNNs via joint training rather than requiring laborious and inefficient hard-coding of algorithms or separate discrete processing blocks to implement the same process. Thus, by considering the UE sensor data 140 in addition to the UE reference signal measurements 142, the BS position RX processing module 704 can generate a more accurate and meaningful UE position estimate than conventional RAT-aided positioning techniques.

[0071] In at least some embodiments, the serving BS 108-1 processes the UE position estimate 130 or transmits the UE position estimate 130 to the UE 110 or one or more other components of the wireless communication system 100, such as a Location Management Function (LMF) server (not shown), for further processing. The UE position estimate 130 may be transmitted in a variety of standard or non-standard formats and may include additional information, such as an estimation error (uncertainty), the method used to obtain the UE position estimate, etc. When the serving BS 108-1 transmits the UE position estimate 130 to one or more other components, the RF antenna interface 304 and one or more antennas 302-1 of the serving BS 108-1 convert an output 712 representing the position estimate 130 into a corresponding RF signal 714 that is transmitted over the air for reception by the UE 110 (or other network component). In at least some embodiments, the serving BS 108-1 implements a UE position estimation TX processing module (not shown) having one or more DNNs trained to provide processing that effectively results in coded (e.g., compressed) data and / or a channel coded representation of the UE position estimate 130 suitable for wireless transmission over the RF antenna interface 304, and such processing is performed jointly via joint training. Or trained on multiple DNNs.

[0072] A DNN or other neural network for implementing a RAT-assisted UE positioning path between the BS 108 and the UE 110 provides design flexibility and facilitates efficient updates compared to traditional block-by-block design and testing approaches, while allowing devices in the UE positioning path to quickly adapt the generation, transmission, and processing of reference signals, UE measurements and sensor reports, and UE position estimates based on current operating parameters and capabilities. However, before a DNN can be deployed and begin operation, it is typically trained or otherwise configured to provide appropriate outputs for a predetermined set of one or more inputs. To this end, FIG. 8 illustrates an exemplary method 800 for developing one or more jointly trained DNN architecture configurations as options for devices in a RAT-assisted UE positioning path for different operating environments or capabilities, according to some embodiments. Note that the order of operations described in connection with FIG. 8 is for illustrative purposes only; a different order of operations may be performed, and further, one or more operations may be omitted or one or more additional operations may be included in the illustrated method. Furthermore, while FIG. 8 illustrates an offline training approach using one or more test nodes, it should be noted that a similar approach can be implemented for online training using one or more nodes in active operation.

[0073] As described above, the operation of a DNN used in one or both devices in a DNN chain forming a corresponding RAT-assisted UE positioning path may be based on the specific capabilities and current operating parameters of the RAT-assisted UE positioning path, such as the operating parameters and / or capabilities of the device using the corresponding DNN, one or more upstream or downstream devices, or a combination thereof. These capabilities and operating parameters may include, for example, the type of sensors used to sense the current condition of the device, the capabilities of such sensors, the power capacity of one or more devices, the processing capacity of one or more devices, the RF antenna interface configuration of one or more devices (e.g., number of beams, antenna ports, supported frequencies), etc. Because the described DNNs utilize such information to determine their operation, it can be understood that in many cases the specific DNN configuration implemented in one of the nodes is based on the specific capabilities and operating parameters currently used in that device or devices on the other side of the RAT-assisted UE positioning path; i.e., the specific DNN configuration implemented reflects the capability information and operating parameters currently indicated by the RAT-assisted UE positioning path implemented by the BS 108 and the UE 110.

[0074] Thus, method 800 begins at block 802 with identifying the expected capabilities (including expected operating parameters or parameter ranges) of one or more test nodes of a test RAT-assisted UE positioning path, including one or more test BSs and one or more test UEs (also referred to as "test devices" for brevity). In the following, it is assumed that training module 408 of management component 150 is managing the joint training, and thus capability information at the test devices is known to training module 408 (e.g., via a database or other locally stored data structure that stores this information). However, since management component 150 likely does not have a priori knowledge of the capabilities of any given UE, the test UE provides management component 150 with an indication of its capabilities, such as an indication of the types of sensors available at the test UE, an indication of various parameters of these sensors (e.g., imaging resolution and image data format of an imaging camera, type of satellite positioning and format of satellite-based location sensor, etc.), accessories available at the device, and applicable parameters (e.g., number of audio channels). For example, the test UE may have at least 4 UECapabilityEnquiry sent by the BS according to the 5G LTE and 5G NR specifications An indication of this capability may be provided as part of a UECapabilityInformation RRC message, which is typically provided by the UE in response to a Radio Resource Control (RRC) message. Alternatively, the test UE may provide the indication of the sensor capability as a separate side-channel or control channel communication. Furthermore, in some embodiments, the capabilities of the test device may be stored in a local or remote database available to management component 150, which may then query this database based on some form of identifier for the test device, such as an International Mobile Subscriber Identity (IMSI) value associated with the test device.

[0075] In at least some embodiments, the training module 408 may attempt to train permutations of every RAT-assisted UE positioning configuration (or, for brevity, “UE positioning configuration”). However, in implementations in which the BS 108 and the UE 110 are likely to have a relatively large number and variety of capabilities and other operating parameters, this approach may be impractical. Thus, at block 804, the training module 408 may select a particular UE positioning configuration from a designated set of candidate RAT-assisted UE positioning configurations for jointly training the DNN of the test device. Thus, each candidate UE positioning configuration may represent a particular combination of UE positioning-related parameters, parameter ranges, or combinations thereof. Such parameters or parameter ranges may include sensor capability parameters, processing capability parameters, battery power parameters, RF signaling parameters such as the number and type of antennas, the number and type of subchannels, etc. Such UE positioning-related parameters may further represent a particular type of reference signal used by the BS 108, a manner in which the UE 110 performs reference signal measurements, the type of sensor data fused with the reference signal measurements, etc. Once the candidate UE positioning configurations are selected for training, further in block 804, the training module 408 identifies initial DNN architecture configurations for each of the test BS and the test UE and instructs the test device to implement each of these initial DNN architecture configurations by either providing the test device with an identifier associated with the initial DNN architecture configuration if the test device stores a copy of the candidate initial DNN architecture configuration, or by sending data representing the initial DNN architecture configuration itself to the test device.

[0076] Once a UE positioning configuration is selected and the test device is initialized with a DNN architecture configuration based on the selected UE positioning configuration, in block 806, the training module 408 identifies one or more sets of training data to use in jointly training the DNNs of the DNN chain based on the selected UE positioning configuration and the initial DNN architecture configuration. That is, the one or more training data sets include or represent data that can be provided as input to the corresponding DNNs in offline or online operation and are therefore suitable for training the DNNs. By way of example, this training data may include streams of test positioning (or other) reference signals, test reception indications of the test positioning (or other) reference signals, test parameters or configurations of the test positioning (or other) reference signals, test reference signal measurements, test sensor data matching sensors included in the configuration under test, test UE measurement reports, test UE measurement and sensor reports, test reception indications of the UE measurement reports, test reception indications of the UE measurements and sensor reports, test UE position estimates, etc.

[0077] Once one or more training sets are obtained, the training module 408 begins joint training of the DNNs for the test UE positioning path at block 808. This joint training typically involves initializing the bias weights and coefficients of the various DNNs with initial values ​​that are generally pseudo-randomly selected, and then performing the T Joint training involves inputting a set of training data into an X processing module (e.g., BS position reference signal TX processing module 602), transmitting the resulting output over the air as a transmission to an RX processing module of the test UE device (e.g., UE position reference signal RX processing module 604), analyzing the resulting output, and then updating the DNN architecture configuration based on the analysis. Joint training can further include inputting a set of training data into a TX processing module of the test UE device (e.g., UE position feedback TX processing module 702), transmitting the resulting output over the air as a transmission to an RX processing module of the test BS device (e.g., BS position RX processing module 704), analyzing the resulting output, and then updating the DNN architecture configuration based on the analysis. In another example, the joint training comprises end-to-end joint training including inputting a set of training data to a TX processing module of a test BS device (e.g., BS position reference signal TX processing module 602), transmitting the resulting output over the air as a transmission to an RX processing module of a test UE device (e.g., UE position reference signal RX processing module 604), providing the output of the RX processing module of the test UE device as an input to a TX processing module of the test UE device (e.g., UE position feedback TX processing module 702), transmitting the resulting output over the air as a transmission to the RX processing module of the test BS device (e.g., BS position RX processing module 704), analyzing the resulting output, and then updating the DNN architecture configuration based on the analysis. In at least some embodiments, at least one of the DNN architecture configurations of one or more test devices is trained individually.

[0078] As is frequently used in DNN training, feedback obtained as a result of the actual resulting output of one or more of the BS position reference signal TX processing module 602, the UE position reference signal RX processing module 604, the UE position feedback TX processing module 702, or the BS position RX processing module 704 is used to modify or otherwise refine parameters of one or more DNNs in the UE positioning path, such as via backpropagation. Thus, at block 810, the management component 150 and / or the DNN chain obtains feedback on the transmitted training set. This feedback implementation can take any of a variety of forms or combinations of forms. In at least some embodiments, the feedback includes the training module 408 or other training module determining an error between the actual resulting output and the expected resulting output and backpropagating this error throughout the DNNs in the DNN chain. For example, because processing by the DNN chain effectively provides a form of UE position estimate, objective feedback on the training data set allows some form of measurement of the accuracy of the UE position estimate obtained as output from the DNN chain, e.g., compared to known UE position, known UE orientation, known UE velocity, etc.

[0079] In block 812, the management component 150 or the DNN chain uses the feedback obtained as a result of transmitting the test data set through the DNN chain and, via display or other consumption of the resulting output at the test transmitting device, updates various aspects of one or more DNNs in the UE positioning path, for example, through backpropagation of errors to modify weights, connections, or layers of the corresponding DNNs, or through managed modifications by the management component 150 in response to such feedback. The management component 150 (or other network component) performs the training process of blocks 806-812 on the next training data set selected in the next iteration of block 806, repeating until a certain number of training iterations have been performed or until a certain minimum error rate has been achieved.

[0080] As a result of the joint (or individual) training of the neural networks along the UE positioning path between the test BS device and the test UE device, each neural network has a particular neural network architecture configuration, or DNN architecture configuration, where the implemented neural network is a DNN that characterizes the architecture and parameters of the corresponding DNN, such as the number of hidden layers, the number of nodes in each layer, the connections between each layer, weights, coefficients, and other bias values ​​implemented for each node. Thus, upon completion of the joint or individual training of the DNNs for the UE positioning path in the selected UE positioning configuration, in block 814, the management component 150 (or other network component) distributes some or all of the trained DNN configurations to the BS 108 and the UE 110 in the system 100. Each node stores the resulting DNN configuration of the corresponding DNN as a DNN architecture configuration. In at least one embodiment, the management component 150 (or other network component) can generate the DNN architecture configuration by extracting the architecture and parameters of the corresponding DNN, such as the number of hidden layers, the number of nodes, the connections, coefficients, weights, and other bias values, at the end of the joint training. In other embodiments, the management component 150 stores copies of the paired DNN architecture configurations as candidate neural network architecture configurations 414 in the set 412. The management component 150 (or other network component) then distributes these DNN architecture configurations to the BSs 108 and UEs 110 as needed.

[0081] If there are one or more other candidate UE positioning configurations remaining to be trained, method 800 returns to block 804 for selecting a next candidate UE positioning configuration to be jointly trained, and another iteration of the subprocess of blocks 804-814 is repeated for the next UE positioning configuration selected by training module 408. Otherwise, if the DNNs for the UE positioning path have been jointly trained for all intended UE positioning configurations, method 800 is complete, and system 100 can transition to neural network-assisted RAT-assisted UE positioning, as described below in connection with FIGS.

[0082] As previously mentioned, management component 150 (or other network components) can perform the joint training process using offline test nodes (i.e., while no active communication of control information or user plane data is occurring) or while the actual nodes in the intended transmission path are online (i.e., while active communication of control information or user plane data is occurring). Furthermore, in some embodiments, rather than management component 150 jointly training all DNNs, management component 150 can potentially train or retrain a subset of the DNNs while maintaining other DNNs as static. By way of example, management component 150 can detect that a DNN of a particular device is operating inefficiently or erroneously, e.g., due to a change in the capabilities of the device implementing the DNN or in response to a previously unreported loss of processing capacity, and thus management component 150 can schedule an individual retraining of the device's DNN while maintaining other DNNs of other devices in their current configurations.

[0083] Further, it can be appreciated that there can be a wide variety of devices supporting multiple UE positioning configurations, and that many different nodes can support the same or similar UE positioning configurations. Thus, following joint training of a representative device, without having to repeat joint training for every device incorporated into the UE positioning path, that device can send an indication of its trained DNN architecture configuration in the UE positioning configuration to management component 150, which can then record the DNN architecture configuration. The UE positioning configuration may be stored in the DNN and then transmitted to other devices that support the same or similar UE positioning configuration for implementation in the DNN of the UE positioning path.

[0084] Furthermore, DNN architectural configurations often change over time as the corresponding device operates using the DNN. Thus, as operation progresses, a given device's neural network management module (e.g., neural network management module 216, 314) can be configured to transmit an indication of an updated architectural configuration of one or more DNNs used at that node, for example, by providing updated gradients and related information to management component 150 in response to a trigger. This trigger may be the expiration of a periodic timer, a query from management component 150, a determination that the magnitude of change has exceeded a specified threshold, etc. Management component 150 then incorporates these received DNN updates into the corresponding DNN architectural configuration, thus having an updated DNN architectural configuration available for distribution to nodes in the transmission path as needed.

[0085] 9 and 10 together illustrate an example method 900 for RAT-assisted UE positioning using a jointly trained DNN-based UE positioning path between wireless devices according to some embodiments. For ease of discussion, the method 900 of FIG. 9 is described below with reference to the example UE positioning path 116 of FIGS. 1, 6, and 7. Additionally, the process of the method 900 is described with reference to the example transaction (ladder) diagram 1000 of FIG. 10. The method 900 begins at block 902, where the BS 108 and the UE 110 establish a wireless connection, such as via a 5G NR standalone registration / attachment process in a cellular context or an IEEE 802.11 association process in a wireless local area network (WLAN) context. At block 904, management component 150 obtains capability information from each of BS 108 and UE 110, such as capability information 1002 (FIG. 10) provided by capability management module 320 (FIG. 3) of BS 108 and capability information 1004 (FIG. 10) provided by capability management module 218 (FIG. 2) of UE 110. In at least some embodiments, management component 150 may already be informed of the capabilities of BS 108 if the BS 108 is part of the same infrastructure network, in which case obtaining capability information 1002 at BS 108 may include accessing a local or remote database or other data store for this information. In the case of UE 110, BS 108 may send a capability request to UE 110, and UE 110 responds to this request with capability information 1004, which BS 108 then forwards to management component 150. For example, BS 108 may send a UECapabilityEnquiry RRC message to which UE 110 responds with a UECapabilityInformation RRC message containing CSI-related capability information.

[0086] At block 906, the neural network selection module 410 of the management component 150 selects a pair of UE positioning DNN architecture configurations to be implemented in the BS 108 and the UE 110 to support the UE positioning path 116, e.g., using capability information and other information describing the UE positioning configuration between the BS 108 and the UE 110 (DNN selection 1006 of FIG. 10 ). In at least some embodiments, the neural network selection module 410 employs an algorithm selection process in which the capability information obtained from the BS 108 and the UE 110 and the UE positioning configuration parameters of the UE positioning path 116 are compared with attributes of pairs of candidate neural network architecture configurations 414 in the set 412 to identify suitable pairs of DNN architecture configurations. In other embodiments, the neural network selection module 410 may organize the candidate DNN architecture configurations into one or more LUTs, each entry storing a corresponding pair of DNN architecture configurations and indexed by a corresponding combination of input parameters or parameter ranges, and thus selecting the appropriate neural network architecture configurations. The network selection module 410 can select an appropriate pair of DNN architecture configurations to be used by the BS 108 and the UE 110 by providing the capability and UE positioning configuration parameters identified in block 904 as inputs to one or more LUTs. In at least some embodiments, as indicated by blocks 901, 903, the management component 150 obtains updated capability information from the BS 108 and the UE 110. The management component 150 can then select a different DNN architecture in one or more of the BS 108 or the UE 110 based on the updated capability information.

[0087] Further, in block 906, the management component 150 instructs the BS 108 and the UE 110 to implement their respective DNN architecture configurations from the pair of selected, co-trained DNN architecture configurations. In embodiments in which the BS 108 and the UE 110 each store candidate DNN architecture configurations for potential future use, the management component 150 can transmit a message with an identifier of the DNN architecture configuration to be implemented by the BS 108 and the UE 110. Otherwise, the management component 150 can transmit information representing the DNN architecture configuration, for example, as a layer 1 signal, a layer 2 control element, a layer 3 RRC message, or a combination thereof. For example, referring to FIG. 10 , the management component 150 transmits a DNN configuration message 1008 to the BS 108, the DNN architecture configuration including data representing the selected DNN architecture configuration for the BS 108. In response to receiving this message, the neural network management module 314 of the BS 108 extracts data from the DNN configuration message 1008 to configure one or more of the BS position reference signal TX processing module 602 or the BS positioning RX processing module 704 to implement one or more DNNs having the DNN architecture configuration represented by the extracted data. Similarly, the management component 150 transmits a DNN configuration message 1010 ( FIG. 10 ) to the UE 110 that includes data representing the DNN architecture configuration selected for the UE 110. In response to receiving this message, the neural network management module 216 of the UE 110 extracts data from the DNN configuration message 1010 to configure one or more of the UE position reference signal RX processing module 604 or the UE position feedback TX processing module 702 to implement one or more DNNs having the DNN architecture configuration represented by the extracted data.

[0088] Once the DNN of the UE positioning path 116 is initially configured, the RAT-assisted UE positioning process can begin. Thus, at block 908, the BS position reference signal TX processing module 602 receives reference signal information 122 as input from the BS reference signal management module 316 and generates and configures a corresponding reference signal 1012 ( FIG. 10 ) output from this input. As discussed above in connection with FIG. 6 , the reference signal information 122 includes one or more different types of information, such as BS and / or UE operating characteristics or reference signal parameters, that the DNN of the BS position reference signal TX processing module 602 utilizes as input to generate and configure one or more reference signals 1012. The reference signal information 122 may also include information regarding the UE positioning configuration of the UE positioning path 116, such as the particular beam, antenna, subcarrier, etc. to be used. At block 910, the BS position reference signal TX processing module 602 provides wireless transmission of the reference signal 1012 to the UE 110.

[0089] At block 912, the reference signal 1012 is received and processed by the RF front end 204 of the UE 110, and the reference signal measurement module 220 of the UE 110 performs one or more reference signal measurements 1014 ( FIG. 10 ), such as RSRP, RSTD, OTDoA, UTDoA, TDAV, AoA, AoD, RTT, etc., on the resulting output. At block 914, the UE position reference signal RX processing module 604 of the UE 110 receives the reference signal measurements 1014 and, in some embodiments, processes UE sensor data 1016 ( FIG. 10 ) as input. In at least some embodiments, the UE position reference signal RX processing module 604 receives a reference signal 1012 as input and performs reference signal measurements 1014 compared to receiving reference signal measurements 1014 from the reference signal measurement module 220. From these inputs, the UE position reference signal RX processing module 604 generates corresponding UE measurement and sensor report 1018 ( FIG. 10 ) outputs that fuse the UE sensor data 1016 with the UE reference signal measurements 1014. In at least some embodiments, the UE position reference signal RX processing module 604 does not receive the UE sensor data 1016 as input. In these embodiments, the UE position reference signal RX processing module 604 generates corresponding UE measurement outputs (compared to the corresponding UE measurement and sensor report outputs).

[0090] At block 916, the UE position feedback TX processing module 702 of the UE 110 receives as input the UE measurements and sensor reports 1018 and generates from this input corresponding output signals representing the UE measurements and sensor reports 144 for wireless transmission to the BS 108. The UE 110 may use various messaging mechanisms, such as the RRC protocol, LPP, etc., to configure and transmit wireless communications. At block 918, the output signals representing the UE measurements and sensor reports 1018 are received and processed by the RF front end 304 of the BS 108, which provides the UE measurements and sensor reports 1018 as input to the BS position RX processing module 704 of the BS 108. The BS position RX processing module 704 processes the UE measurements and sensor reports 1018, which, in at least some embodiments, include the UE reference signal measurements 1014 and the UE sensor data 1016, to generate an output representing a position estimate 1020 ( FIG. 10 ) for the UE 110. As discussed above in connection with FIG. 6, in at least some embodiments, the UE position estimate 1020 not only incorporates reference signal measurements 1014 provided by the UE 110, but also incorporates UE sensor data 1016, resulting in a UE position estimate that includes secondary information such as, for example, the local conditions of the UE 110, an indication of the UE's orientation, movement (e.g., rotation, direction, etc.), and / or the like.

[0091] At block 920, the BS position reference signal TX processing module 602 or other TX processing module of the BS 108 optionally generates and transmits to the UE 110 (or other network component) an RF signal 1022 ( FIG. 10 ) configured based on the UE position estimate 1020. At block 922, the neural network management module 216 of the BS 108 or the neural network selection module 410 of the management component 150 optionally adjusts one or more DNNs of the BS position RX processing module 704 based on the UE position estimate 1020 calculated for the current UE 110 and the UE position estimates calculated for one or more other UEs 110. For example, the BS 108 or the management component 150 determines whether the UE position estimate 1020 calculated for the current UE 110 and the UE position estimates calculated for one or more other UEs 110 indicate that the UE 110 occupies the same physical space. If so, because multiple UEs 110 cannot occupy the same physical space, the BS 108 or management component 150 determines that the BS position RX processing module 604 has a positioning error and needs to be refined. The BS 108 or management component 150 can adjust one or more parameters, such as weights, of the BS position RX processing module 704 (or any of the remaining processing modules of the BS 108 or UE 110) to correct the identified positioning error.

[0092] Although the corresponding operational example of method 900 of Figure 9 and ladder diagram 1000 of Figure 10 show an implementation in which BS 108 transmits a reference signal and UE 110 performs reference signal measurements, UE 110 can be similarly configured to transmit a reference signal and BS 108 can be configured to perform reference signal measurements. For example, both Figures 11 and 12 show implementations in which BS 108 transmits a reference signal and UE 110 performs reference signal measurements. 11 illustrates an example method 1100 for RAT-assisted UE positioning using a jointly trained DNN-based UE positioning path between wireless devices, according to an embodiment having a UE 110 configured to receive and a BS 108 configured to perform reference signal measurements. The process of method 1100 is described with reference to example transaction (ladder) diagram 1200 of FIG. 12. Method 1100 begins at block 1102, which may be after block 906 of method 900, such that the DNNs of the BS 108 and the UE 110 have already been initially configured.

[0093] Thus, at block 1102, the UE position reference signal TX processing module 1202 ( FIG. 12 ) receives reference signal information 122 as input and generates and configures from this input a corresponding modulated reference signal 1208 ( FIG. 12 ) output. As discussed above in connection with FIG. 6 , the reference signal information 122 includes one or more different types of information, such as BS and / or UE operating characteristics or reference signal parameters, that the DNN of the UE position reference signal TX processing module 1202 utilizes as input to generate and configure one or more reference signals 1208. At block 1104, the UE position reference signal TX processing module 1202 further receives local UE sensor data 1210 ( FIG. 12 ) from one or more sensors available at the UE 110. Further at block 1104, the UE position reference signal TX processing module 1202 augments the reference signal 1208 with the sensor data 1210 and generates an output representative of the augmented reference signal 1208. In at least some embodiments, the reference signal is an SRS augmented with sensor data 1210. At block 1106, the UE position reference signal TX processing module 1202 provides wireless transmission of the augmented reference signal 1208 to the BS 108.

[0094] At block 1108, a reference signal 1208 is received and processed by the RF front end 304 of the BS 108, and a reference signal measurement module (not shown) of the BS 108 performs one or more reference signal measurements 1212 ( FIG. 12 ), such as RSRP, RSTD, OTDoA, UTDoA, TDAV, AoA, AoD, RTT, etc., on the resulting output. At block 1110, a BS position reference signal RX processing module 1204 (or other processing module) of the BS 108 receives as input the reference signal measurements 1212 transmitted in the reference signal 1208 and the UE sensor data 1210, and generates from these inputs an output representing a position estimate 1214 ( FIG. 12 ) at the UE 110. In at least some embodiments, the BS position reference signal RX processing module 1204 receives as input the reference signal 1208 to calculate the reference signal measurements 1212 compared to receiving the reference signal measurements 1212 as input. In at least some embodiments, the UE position estimate 1214 incorporates both the reference signal measurements 1212 and the UE sensor data 1210, resulting in a UE position estimate that includes secondary information such as, for example, the local conditions of the UE 110, an indication of the UE's orientation, movement (e.g., rotation, direction, etc.), and / or the like.

[0095] At block 1112, the BS TX processing module 1206 of the BS 108 optionally generates and transmits to the UE 110 (or other network component) an RF signal 1216 (FIG. 12) configured based on the UE position estimate 1214. At block 1114, the neural network management module 216 of the BS 108 or the neural network selection module 410 of the management component 150 optionally adjusts one or more DNNs of the BS position RX processing module 704 based on the UE position estimate 1214 calculated for the current UE 110 and UE position estimates calculated for one or more other UEs 110, similar to the process described above with respect to block 922 of FIG. 9.

[0096] In at least some embodiments, certain aspects of the aforementioned technology involve running software. The software may be implemented by one or more processors of a processing system. The software includes one or more sets of executable instructions stored or otherwise tangibly embodied on a non-transitory computer-readable storage medium. The software may include instructions and some data that, when executed by one or more processors, operate the one or more processors to perform one or more aspects of the aforementioned techniques. The non-transitory computer-readable storage medium may include, for example, magnetic or optical disk storage, a solid-state storage device such as flash memory, a cache, a random access memory (RAM), or one or more other non-volatile memory devices. The executable instructions stored on a non-transitory computer-readable storage medium may be source code, assembly language code, object code, or another instruction format that is interpreted or otherwise executable by one or more processors.

[0097] A computer-readable storage medium may include any storage medium, or combination of storage media, that a computer system accesses during use to provide instructions and / or data to the computer system. Such storage media may include, but are not limited to, optical media (such as compact discs (CDs), digital versatile discs (DVDs), Blu-ray discs, etc.), magnetic media (such as floppy disks, magnetic tape, or magnetic hard drives), volatile memory (such as random access memory (RAM) or cache), non-volatile memory (such as read-only memory (ROM) or flash memory), or microelectromechanical systems (MEMS)-based storage media. A computer-readable storage medium may be embedded in a computing system (e.g., system RAM or ROM), permanently attached to a computing system (e.g., a magnetic hard drive), removably attached to a computing system (e.g., an optical disk or universal serial bus (USB)-based flash memory), or coupled to a computer system via a wired or wireless network (e.g., network-access storage (NAS)).

[0098] It should be noted that not all of the activities or elements described above in the general description are required, that some of the specific activities or devices may not be required, and that one or more additional activities may be performed or additional elements may be included in addition to those described. Furthermore, the order in which the activities are listed is not necessarily the order in which they are performed. Also, the concepts have been described with reference to specific embodiments. However, those skilled in the art will recognize that various modifications and changes can be made without departing from the scope of the present disclosure, as set forth in the claims below. Accordingly, the specification and drawings are to be regarded in an illustrative rather than a restrictive sense, and all such modifications are intended to be included within the scope of the present disclosure.

[0099] Benefits, other advantages, and solutions to problems have been described above with regard to specific embodiments. However, benefits, advantages, solutions to problems, and any features by which a benefit, advantage, or solution occurs or becomes more pronounced should not be construed as critical, essential, or essential features of any or all claims. Moreover, the specific embodiments disclosed above are merely exemplary, and the disclosed subject matter may be modified and practiced in different but equivalent manners apparent to those skilled in the art having the benefit of the teachings herein. No limitations are intended to the details of construction or design shown herein, except as set forth in the claims below. It is therefore apparent that the specific embodiments disclosed above may be altered or modified, and all such variations are considered within the scope of the disclosed subject matter. Accordingly, the protection sought herein is as set forth in the following claims.

Claims

1. A computer-implemented method on a first device (108), comprising: receiving reference signal information (122) as input to a transmit neural network (120) of the first device; the transmit neural network (120) generating a first output (138) representing a reference signal based on the reference signal information (122); controlling a radio frequency (RF) antenna interface (304) of the first device (108) to transmit a first RF signal (152) representative of the first output (138) for reception by a second device (110); receiving, in response to transmitting the first RF signal, an input representing one or more RF signals associated with the second device at a receiver neural network of the first device; generating a second output (712) representing a position estimate (130) of the second device (110) based on the input (710) to the receiver neural network (128); A method comprising:

2. Receiving the input (710) representing one or more RF signals (154) associated with the second device (110) comprises: receiving a second RF signal (154) from the second device (110) representing a signal measurement (142) associated with the first RF signal (152); The method of claim 1 , comprising:

3. The method of claim 2 , wherein the second RF signal (154) received from the second device (110) further represents local sensor data (140) generated at the second device (110).

4. The method of any one of claims 1 to 3, wherein the position estimate (130) indicates a position of the second device (110) and an orientation of the second device (110).

5. Producing the first output (138) comprises: generating the first output (138) at the transmit neural network (120) based on a first neural network architecture configuration (324) at the transmit neural network (120); Including, generating the second output (712) includes generating the second output (712) at the receiving neural network (128) based on a second neural network architecture configuration (324) at the receiving neural network (128); The method according to any one of claims 1 to 4.

6. selecting at least one of the first neural network architecture configuration (324) or the second neural network architecture configuration (324) from a plurality of neural network architecture configurations based on one or more capabilities of at least one of the first device (108) or the second device (110); The method of claim 5 further comprising:

7. Selecting the first neural network architecture configuration (324) includes: receiving information from the second device (110) representing one or more capabilities (1004) of the second device (110); using said information to select said first neural network architecture configuration (324); The method of claim 6, comprising:

8. selecting the second neural network architecture configuration (324) from the plurality of neural network architecture configurations based on one or more capabilities (1002, 1004) of at least one of the first device (108) or the second device (110); 8. The method of claim 6 or claim 7, further comprising:

9. receiving a command (1008) from a management infrastructure component (150) to implement at least one of the first neural network architectural configuration (324) in the transmitting neural network (120) or the second neural network architectural configuration (324) in the receiving neural network (128); The method of any one of claims 5 to 8, further comprising:

10. selecting at least one of a third neural network architecture configuration (324) in the transmitting neural network (120) or a fourth neural network architecture configuration (324) in the receiving neural network (128) in response to a change in one or more capabilities (1002, 1004) of at least one of the first device (108) or the second device (110). The method of any one of claims 5 to 9, further comprising:

11. Engaging in joint training of the transmitting neural network (120) and the receiving neural network (128) of the first device (108) with the transmitting neural network (148) and the receiving neural network (134) of the second device (110). The method of any one of claims 1 to 10, further comprising:

12. communicating with a third device (108-2) implementing a transmitting neural network (120-2); configuring the transmit neural network (120-2) of the third device (108-2) to generate an output representing a reference signal (138-2) for reception by the second device (110); The method of any one of claims 1 to 11, further comprising:

13. generating, at the receiver neural network (128) of the first device (108), a third output representing a position estimate (1214) of a third device (110-2) based on one or more RF signals (1208) received from the third device (110-2); determining, at the receiving neural network (120) of the first device (108), that the second output and the third output indicate that the second device (110) and the third device (110-2) occupy the same space; refining one or more parameters of the receiving neural network (120) of the first device (108) in response to the second output and the third output indicating that the second device (110) and the third device (110-2) occupy the same space; The method of any one of claims 1 to 11, further comprising:

14. A computer-implemented method on a first device, comprising: receiving, at a radio frequency (RF) antenna interface (204) of the first device (110), a first RF signal (152) representing a reference signal (138) from a second device (108); providing a representation of the first RF signal (152) as a first input to a receiver neural network (134) of the first device (110); generating a first output (706) representing a measurement report (144) at the first device (110) based on the first input to the receiving neural network (134); A method comprising:

15. receiving, as an input, the first output (706) from the receiving neural network (134) at a transmitting neural network (148) of the first device (110); the transmitting neural network (148) generating a second output representative of the measurement report (144); controlling the RF antenna interface (204) of the first device (110) to transmit a second RF signal (154) representing the second power for reception by the second device (108); 15. The method of claim 14, further comprising:

16. Producing the first output (706) comprises:

16. The method of claim 14 or 15, comprising performing one or more reference signal measurements (142) on the first input representing the reference signal (138), and the measurement report (144) includes at least one of the one or more reference signal measurements (142).

17. 17. The method of claim 16, further comprising providing a representation of sensor data (140) generated by one or more sensors (210) of the first device (110) as a second input to the receiver neural network (134) of the first device (110), wherein the measurement report (144) includes the one or more reference signal measurements (142) fused with the sensor data (140).

18. receiving a command (1010) from a network infrastructure component (150) to implement at least one of a first neural network architecture configuration (224) in the receiving neural network (134) or a second neural network architecture configuration (324) in the transmitting neural network (148); The method of any one of claims 14 to 17, further comprising:

19. In response to a change in one or more capabilities of the first device, sending a message to the network infrastructure component indicating the change in the one or more capabilities; receiving, in response to transmitting the message, from the network infrastructure component, a second neural network architecture configuration for at least one of the receiving neural network or the transmitting neural network; 20. The method of claim 18, further comprising:

20. a radio frequency (RF) antenna interface (204, 304); at least one processor (206, 306) coupled to the RF antenna interface (204, 304); a memory (208, 308) storing executable instructions configured to operate said at least one processor (206, 306) to perform the method of any one of claims 1 to 19; A device (108, 110) comprising:

Images