Adaptive sensing and sensor reconfiguration in perceptual wireless communications

By introducing machine learning models and adaptive sensing technology into the 5G NR system, the problem of rapid changes in wireless link quality caused by LOS path obstruction at high frequencies was solved, enabling real-time prediction and management of beam obstruction and improving communication stability and quality.

CN118975152APending Publication Date: 2024-11-15QUALCOMM INC
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202380031164.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Priority Date
2022-04-06
Filing Date
2023-01-27
Publication Date
2024-11-15

AI Technical Summary

Technical Problem

In 5G NR systems, the quality of wireless links changes rapidly due to the obstruction of LOS paths at high frequencies. Existing technologies struggle to effectively predict and manage beam obstruction, thus affecting communication quality.

Method used

By introducing machine learning (ML) models into user equipment (UE) and network nodes, adaptive sensing and feature extraction are performed using onboard sensors and ML service entities, dynamically adjusting sensor configuration and beam management to predict and avoid beam blocking.

Benefits of technology

It enables real-time prediction and management of beam blocking, improving the stability and quality of wireless communication and adapting to dynamic changes in vehicles and the environment.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118975152A_ABST
    Figure CN118975152A_ABST
Patent Text Reader

Abstract

Aspects are provided that allow a UE to switch between feature data extraction models to provide adaptive feature data of dynamic, potential LOS obstructions for improved beam blocking prediction performance (or other functionality) at a network node. First, the network node receives first feature data from the UE based on a first data extraction model of the UE. The network node sends a message indicating the UE to switch from the first data extraction model of the UE to a second data extraction model based on the status of the UE. In response to the message, the UE determines to switch from the first data extraction model to the second data extraction model, and sends second feature data to the network node based on the second data extraction model. The network node may then determine a beam blocking prediction in response to the second feature data.
Need to check novelty before this filing date? Find Prior Art

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims the benefit of U.S. patent application serial number 17 / 714,946, entitled “ADAPTIVE SENSING AND SENSOR RECONFIGURATION INPERCEPTIVE WIRELESS COMMUNICATIONS” and filed on April 6, 2022, which is assigned to the assignee of the present invention and is incorporated herein by reference in its entirety. Background Art Technical Field

[0003] The present disclosure relates generally to communication systems, and more particularly to adaptive machine learning (ML) and sensor-based inference extraction for dynamic beam interference management.

[0004] introduction

[0005] Wireless communication systems are widely deployed to provide a variety of telecommunication services, such as telephony, video, data, messaging, and broadcast. Typical wireless communication systems may employ multiple access technologies capable of supporting communication with multiple users by sharing available system resources. Examples of such multiple access technologies include code division multiple access (CDMA) systems, time division multiple access (TDMA) systems, frequency division multiple access (FDMA) systems, orthogonal frequency division multiple access (OFDMA) systems, single carrier frequency division multiple access (SC-FDMA) systems, and time division synchronous code division multiple access (TD-SCDMA) systems.

[0006] These multiple access technologies have been adopted in various telecommunication standards to provide a common protocol that enables different wireless devices to communicate at a city, country, region, or even global level. An example telecommunication standard is 5G New Radio (NR). 5G NR is part of the continuous mobile broadband evolution promulgated by the Third Generation Partnership Project (3GPP) to meet new requirements associated with latency, reliability, security, scalability, and other requirements (e.g., in the Internet of Things (IoT)). 5G NR includes services associated with enhanced mobile broadband (eMBB), massive machine type communications (mMTC), and ultra-reliable low latency communications (URLLC). Certain aspects of 5G NR may be based on the 4G Long Term Evolution (LTE) standard. Further improvements to 5GNR technology are needed. In addition, these improvements may also be applicable to other multiple access technologies and telecommunication standards that employ these technologies. Summary of the invention

[0007] The present invention is defined by the claims. Embodiments and aspects that do not fall within the scope of the claims are merely examples for explaining the present invention.

[0008] A simplified summary of one or more aspects is presented below in order to provide a basic understanding of these aspects. This summary is not an extensive overview of all contemplated aspects, and is neither intended to identify key or important elements of all aspects, nor to describe the scope of any or all aspects. Its sole purpose is to present some concepts of one or more aspects in a simplified form as a prelude to a more detailed description presented later.

[0009] In one aspect of the present disclosure, a method, a computer-readable medium, and a device for wireless communication are provided. The device may be a user equipment (UE). The device includes a processor; a memory coupled to the processor; and instructions stored in the memory and operable when executed by the processor to cause the device to: receive a message indicating that the device switches from a current ML-based feature data extraction model of the device to one of a plurality of ML-based feature data extraction models based on a state of the device, wherein each of the plurality of ML-based feature data extraction models provides ML-based feature data for predicting beam blocking between the device and a network entity; determine to switch from the current ML-based feature data extraction model to the one of the ML-based feature data extraction models based at least in part on the message; and send ML-based feature data for predicting beam blocking based on one of the ML-based feature data extraction models based on the ML-based feature data extraction models in response to the switch.

[0010] In one aspect of the present disclosure, a device is provided. The device may be a user equipment (UE). The UE may include a component for receiving a message instructing the device to switch from a current machine learning (ML)-based feature data extraction model of the device to one of a plurality of ML-based feature data extraction models based on a state of the device, wherein each of the plurality of ML-based feature data extraction models provides ML-based feature data for predicting beam blocking between the device and a network entity; a component for determining to switch from the current ML-based feature data extraction model to one of the ML-based feature data extraction models based on the message at least in part; and a component for sending ML-based feature data for predicting beam blocking based on one of the ML-based feature data extraction models based on the ML-based feature data extraction models in response to the switch.

[0011] In one aspect of the present disclosure, a method, a computer-readable medium, and an apparatus for wireless communication are provided. The apparatus may be a network node. The network node includes a processor; a memory coupled to the processor; and instructions stored in the memory and operable when executed by the processor to cause the network node to: receive first ML-based feature data from a UE based on a first ML-based feature data extraction model of the UE; send a message indicating that the UE switches from the first ML-based feature data extraction model of the UE to a second ML-based feature data extraction model based on a state of the UE; receive second ML-based feature data from the UE based on the second ML-based feature data extraction model of the UE; and determine a beam blocking prediction in response to the second ML-based feature data.

[0012] In one aspect of the present disclosure, an apparatus is provided. The apparatus may be a network node. The network node includes a component for receiving first ML-based feature data from a UE based on a first ML-based feature data extraction model of the UE; a component for sending a message instructing the UE to switch from the first ML-based feature data extraction model of the UE to a second ML-based feature data extraction model based on a state of the UE; a component for receiving second ML-based feature data from the UE based on the second ML-based feature data extraction model of the UE; and a component for determining a beam blocking prediction in response to the second ML-based feature data.

[0013] To achieve the foregoing and related ends, one or more aspects include the features fully described below and particularly pointed out in the claims. The following description and the accompanying drawings set forth in detail some illustrative features of one or more aspects. However, these features are merely indicative of some of the various ways in which the principles of the various aspects may be employed, and this specification is intended to include all such aspects and their equivalents. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 is a diagram illustrating an example of a wireless communication system and an access network.

[0015] Figure 2A is a diagram illustrating an example of a first frame according to various aspects of the present disclosure.

[0016] Figure 2B is a diagram illustrating an example of DL channels within a subframe according to various aspects of the present disclosure.

[0017] Figure 2C is a diagram illustrating an example of a second frame according to various aspects of the present disclosure.

[0018] Figure 2Dis a diagram illustrating an example of UL channels within a subframe according to various aspects of the present disclosure.

[0019] Figure 3 is a diagram illustrating an example of a base station and a user equipment (UE) in an access network.

[0020] Figure 4 A diagram illustrating an example decomposed base station architecture is shown.

[0021] Figure 5 is a conceptual diagram of an example open radio access network (O-RAN) architecture.

[0022] Figure 6 is a flowchart of the example training and inference model.

[0023] Figure 7 An overhead view of the coverage area for wireless communications extracted using adaptive ML training and inference for beam prediction.

[0024] Figure 8 ML service entities at the ML server and UE, respectively, are shown using signaling procedures and parameters for adaptive sensing and feature extraction as well as reconfiguring the sensors of the UE.

[0025] Fig. 9 is an example flow chart of adaptive model training and inference functionality for predicting beam blocking using a session based on data collected from sensors and ML models of a UE.

[0026] Fig.10 is a conceptual diagram of an example in which the training or inference of a feature extraction model at a UE may be separated from the training or inference of a prediction model at an ML serving entity of an ML server or base station.

[0027] Fig.11 is a flow chart of an example communication flow between a UE and an ML serving entity of an ML server or base station, where the ML serving entity may request adaptive sensing and feature extraction for prediction tasks to better support separate ML training, inference, and / or performance optimization.

[0028] Fig.12 is a conceptual diagram illustrating an example of an ML service entity at a UE including a plurality of feature extraction models.

[0029] Fig.13 is a conceptual diagram illustrating a method for adaptive sensing and feature extraction, wherein an ML service entity at an ML server or a base station includes a plurality of ML models.

[0030] Fig.14is a conceptual diagram of an example in which the training or inference of a feature extraction model at a UE may be combined or associated with the training or inference of a prediction model at an ML serving entity of an ML server or base station.

[0031] Fig.15 is a flow chart of an example communication flow between a UE and an ML serving entity of an ML server or base station, where the ML serving entity may request adaptive sensing and feature extraction for prediction tasks to better support joint ML training, inference, and / or performance optimization.

[0032] Fig.16 is a flow chart of an example communication flow between a UE and an ML service entity of an ML server or base station, wherein the ML service entity or the UE may trigger sensor configuration or reconfiguration for a prediction task to better support ML training, inference, and / or performance optimization.

[0033] Fig.17 is a flow chart of a method of performing wireless communication at a UE.

[0034] Fig.18 is a flow chart of a method of wireless communication at a network node.

[0035] Fig.19 is a diagram illustrating an example of a hardware implementation for an example apparatus.

[0036] Fig. 20 is a diagram illustrating another example of a hardware implementation for another example apparatus. DETAILED DESCRIPTION

[0037] The specific embodiments described below in conjunction with the accompanying drawings are intended as descriptions of various configurations and are not intended to represent the only configurations by which the concepts described herein can be practiced. In order to provide a thorough understanding of the various concepts, the specific embodiments include specific details. However, it is obvious to those skilled in the art that these concepts can be practiced without these specific details. In some cases, well-known structures and components are shown in block diagram form to avoid blurring these concepts.

[0038] With fifth generation (5G) wireless technology and beyond, wireless networks may operate in substantially higher frequency bands, such as 28 gigahertz (GHz) ("FR2"), 60 GHz ("FR4"), to the terahertz (THz) bands in the range of 100 GHz and above. Due to the high attenuation and diffraction losses inherent in these frequency bands, obstruction of the line of sight (LOS) path may significantly degrade the quality of the wireless link. Obstruction may occur frequently, and if the LOS path is blocked by a moving obstacle such as a vehicle, a pedestrian, the received power at the user device may drop significantly.

[0039] To overcome these rapid changes in link quality at such high frequencies caused by LOS blocking, vehicles can be equipped with UEs and onboard sensors (e.g., RADAR, LIDAR, cameras, etc.) to provide the radio network with information about moving obstacles that may eventually degrade signal quality by causing beam blocking. The sensing information can be utilized to provide the radio network information of the communication environment and the moving obstacles that may block the LOS beam path.

[0040] In the case where a vehicle equipped with sensors (also referred to as an "autonomous" vehicle) enters a coverage area that includes an ML service entity, a situation in which the problems inherent in the above-described method may arise. The coverage area may include moving objects (vehicles, pedestrians, etc.) and fixed objects (buildings), each of which may affect the LOS. The ML service entity may reside within a network entity in the coverage area, such as a base station, such as a Node B, an evolved Node B, a new radio base station, or a 5G Node B (gNB). Alternatively, the network entity itself may be an ML service entity, such as an ML service entity residing within a base station. The ML service entity may reside within a base station, in an ML server co-located with the base station, in an ML server located near the base station, or in an ML server located elsewhere (such as in a cloud or edge server). Sensor information received from the UE of the entering autonomous vehicle and possibly from other vehicles in the coverage area may assist the base station working with the ML server or ML service entity.

[0041] Using the sensing information provided by the on-board sensors of the vehicles and their ML models, the ML service entity can assist the base station to fully understand the environment in which the ML service entity is located and actively manage the beam to improve the radio link quality. The ML service entity can complete this beam management by performing training tasks to obtain models such as beam blocking prediction models, inference tasks to make predictions of blocked beams based on the trained models, and performance optimization tasks to improve the models. However, due to the mobility of the vehicles equipped with sensors and the dynamic nature of objects that may cause beams to be blocked, dynamic and adaptive sensing is required in this case. Accordingly, it may also be necessary to adaptively and frequently reconfigure the sensors in this case.

[0042] Therefore, various aspects of the present disclosure provide signaling procedures and parameters, through which the ML service entity of the UE and the ML service entity at the ML server or base station can utilize adaptive sensing and feature extraction and on-board reconfigurable sensors to better serve the training, inference and performance optimization tasks of the ML service entity at the ML server or base station. For example, the UE can support ML services through equipped sensors (such as radars or cameras). For example, the UE can also support ML functions through one or more neural networks (NNs) for extracting features from sensor data that may have been collected from the vehicle RADAR or camera. In one aspect, the UE may include an ML control or management function within its ML service entity, which is configured to control and exchange messages for adaptive sensing and feature extraction. The ML control or management function can also be configured to configure or reconfigure sensors through requests or instructions. The ML service entity including these functions can also reside on the vehicle UE. On the other hand, the ML service entity at the vehicle UE can reside in a layer above the 5G protocol stack of the UE modem ( Figure 8 ), such as in the application layer. Similarly, the ML service entity at the ML server or base station may reside in a layer above the protocol stack of the radio access network (RAN). Thus, the ML service entity at the vehicle UE may communicate directly with the ML server or base station to provide adaptive or dynamic beam blocking prediction services. In other configurations, the ML service entity at the UE may communicate with the ML service entity at the base station via an RRC connection.

[0043] In other aspects, an ML server or an ML service entity at a base station that performs a centralized beam blocking prediction service at an ML server may include one or more ML engines that can dynamically predict beam blocking. The ML engine may make these predictions by aggregating sensory data or features received from multiple UEs / vehicles. Thus, the ML engine may proactively direct the base station to adjust beam operations.

[0044] As described above, the ML service entity may reside in a base station or at an ML server that is co-located with the base station, located near the base station, or located elsewhere (such as in a cloud or edge server). As also described, the ML service entity may include processes for adaptive sensing and feature extraction. In one example, the control or management function of the ML service entity of the base station or ML server may request the UE to switch between different ML models of the UE based on training or inference needs, and the control or management function of the ML service entity of the UE may adjust its ML model accordingly. For example, if the state of the UE in the dynamic environment of the ML server or base station indicates that the inference from the UE is more accurate (longer inference time) or the inference from the UE is faster (shorter inference time), the ML service entity at the base station or ML server may instruct the UE to switch to a more complex or less complex feature extraction model, respectively. In the case of beam blocking prediction, a more accurate inference in response to UE model switching can result in a more accurate beam blocking prediction, which may be desirable in an environment where the UE is moving slowly, in an environment with a large number of dynamic objects, or in a similar UE state. Alternatively, faster inference in response to UE model switching may result in faster beam blockage prediction, which may be desirable in environments where the UE is moving quickly, in an environment with few dynamic objects, or similar UE states.

[0045] In addition, in other examples, the control or management function of the ML service entity of the base station or ML server may apply multiple beam management models, such as beamforming and beam tracking, and adaptively instruct the ML service entity of the UE to provide different training or inference data for these models accordingly. In other processes, the control or management function of the ML service entity of the base station or ML server may adaptively request the UE to communicate their confidence for each feature extraction or inference, or apply a model capable of communicating such confidence levels. In any of these processes, the control or management function of the ML service entity of the base station or ML server may aggregate the data received from various UEs into its ML model, and communicate the aggregated performance characteristics of its model (such as back-propagated gradients) to the various UEs to adjust their own feature extraction models as part of a joint system for ML training, inference and / or performance optimization. Alternatively, the ML service entity of the base station or ML server may avoid transmitting such information to maintain separation between its own model and the local feature extraction model of each UE.

[0046] In a further example, the ML service entity of the base station or ML server and the ML service entity of the UE may include a process for adaptively configuring or reconfiguring UE sensors. These processes may occur during ML service discovery and session establishment, ML model training, ML model inference or feature extraction, and / or ML model performance optimization. In one example, the control or management function of the ML service entity of the base station or ML server may adaptively configure or reconfigure UE sensors based on ML service training requirements, ML service inference requirements, ML service performance requirements, network traffic load, and / or the number of UEs in the area. In another example, the control or management function of the ML service entity of the UE may adaptively configure or reconfigure UE sensors based on vehicle sensor settings and configuration, sensor availability, sensor selection, position change, speed change, direction change, radio link quality with the gNB or ML server, UE vehicle ADAS (Advanced Driver Assistance System), and / or UE sensor occlusion.

[0047] Thus, the ML service entity of the base station or ML server can adaptively and dynamically instruct different vehicles to provide different data, extracted features, and model results to predict beam blocking. For example, to improve training, inference, and / or performance optimization, the ML service entity of the base station or ML server can adaptively instruct the UE to switch between feature extraction models, configure or reconfigure the sensor parameters, model parameters (within a given model), or object tracker parameters of the UE, reselect the sensor type of the UE and the task specification to be completed by the UE, perform a forward pass in its model based on aggregated UE data, provide aggregated feedback of gradients or other performance characteristics to the UE for the UE to adaptively adjust its sensing and / or feature extraction, or perform a combination of these aspects. In addition, the ML service entity of the base station or ML server can adaptively trigger sensor reconfiguration at the UE based on training, inference, or performance optimization criteria, system load, number of UEs in an area, or similar factors, and / or confirm the sensor reconfiguration of the UE adaptively triggered by the ML service entity of the UE based on sensor availability, sensor selection, position change, speed change, direction change, radio link quality, etc. In response to these adaptive instructions, the base station, via the ML serving entity of the base station or ML server, may dynamically divide and schedule beam transmissions between entities in a manner that prevents or reduces LOS interference.

[0048] In addition, it should be understood that although the examples of aspects described throughout the present disclosure specifically relate to beam blocking prediction or predicted beam blocking, the disclosed aspects are not limited to application to beam blocking prediction. Instead, in other examples, the disclosed aspects can be similarly applied to beam prediction, beam management, scheduling, load balancing, or other network functions. For example, in response to sensor data or extracted features from a UE, an ML service entity can adaptively instruct the UE to switch between feature extraction models, trigger sensor reconfiguration, etc., to not only predict and prevent LOS beam blocking as previously described, but also or alternatively predict the best beam for communicating with the UE, perform beam management (beamforming training or refinement), optimally schedule resources to the UE, balance current traffic loads, etc.

[0049] Several aspects of telecommunication systems will now be presented with reference to various apparatus and methods. These apparatus and methods will be described in the following detailed description and illustrated in the accompanying drawings by various blocks, components, circuits, processes, algorithms, etc. (collectively referred to as "elements"). These elements may be implemented using electronic hardware, computer software, or any combination thereof. Whether such elements are implemented as hardware or software depends on the specific application and the design constraints imposed on the overall system.

[0050] For example, an element, or any part of an element, or any combination of elements, can be implemented as a "processing system" including one or more processors. Examples of processors include microprocessors, microcontrollers, graphics processing units (GPUs), central processing units (CPUs), application processors, digital signal processors (DSPs), reduced instruction set computing (RISC) processors, systems on chip (SoCs), baseband processors, field programmable gate arrays (FPGAs), programmable logic devices (PLDs), state machines, gating logic, discrete hardware circuits, and other suitable hardware configured to perform the various functions described in this disclosure. One or more processors in a processing system can execute software. Whether referred to as software, firmware, middleware, microcode, hardware description language, or other names, software should be broadly interpreted as meaning instructions, instruction sets, codes, code segments, program codes, programs, subroutines, software components, applications, software applications, software packages, routines, subroutines, objects, executable files, execution threads, processes, functions, etc.

[0051] Therefore, in one or more example embodiments, the described functions can be implemented with hardware, software or any combination thereof. If implemented with software, the functions can be stored or encoded on a computer-readable medium as one or more instructions or codes. Computer-readable media include computer storage media. Storage media can be any available medium that can be accessed by a computer. By way of example and not limitation, such computer-readable media can include a random access memory (RAM), a read-only memory (ROM), an electrically erasable programmable ROM (EEPROM), an optical disk storage device, a magnetic disk storage device, other magnetic storage devices, a combination of computer-readable media of the above type, or any other medium that can be used to store computer executable code in the form of instructions or data structures that can be accessed by a computer.

[0052] Figure 1 1 is a diagram illustrating an example of a wireless communication system and access network 100. The wireless communication system (also referred to as a wireless wide area network (WWAN)) includes a base station 102, a user equipment (UE) 104, an evolved packet core (EPC) 160, and another core network 190 (e.g., a 5G core (5GC)). The base station 102 may include a macro cell (a high-power cellular base station) and / or a small cell (a low-power cellular base station). A macro cell includes a base station. A small cell includes a femto cell, a pico cell, and a micro cell.

[0053] The base station 102 configured for 4G Long Term Evolution (LTE) (collectively referred to as Evolved Universal Mobile Telecommunications System (UMTS) Terrestrial Radio Access Network (E-UTRAN)) can interface with the EPC 160 via a first backhaul link 132 (e.g., an S1 interface). The base station 102 configured for 5G New Radio (NR) (collectively referred to as Next Generation RAN (NG-RAN)) can interface with the core network 190 via a second backhaul link 184. Among other functions, the base station 102 can also perform one or more of the following functions: delivery of user data, radio channel encryption and decryption, integrity protection, header compression, mobility control functions (e.g., handover, dual connectivity), inter-cell interference coordination, connection establishment and release, load balancing, distribution of non-access stratum (NAS) messages, NAS node selection, synchronization, radio access network (RAN) sharing, multimedia broadcast multicast service (MBMS), subscriber and equipment tracking, RAN information management (RIM), paging, positioning, and delivery of warning messages. The base stations 102 may communicate with each other directly or indirectly (eg, via the EPC 160 or the core network 190) via the third backhaul link 134 (eg, an X2 interface). The first backhaul link 132, the second backhaul link 184, and the third backhaul link 134 may be wired or wireless.

[0054] Base station 102 can communicate wirelessly with UE 104. Each of base stations 102 can provide communication coverage for a corresponding geographic coverage area 110. There may be overlapping geographic coverage areas 110. For example, a small cell 102' can have a coverage area 110' that overlaps with the coverage area 110 of one or more macro base stations 102. A network including both small cells and macro cells can be referred to as a heterogeneous network. A heterogeneous network can also include a home evolved Node B (eNB) (HeNB), which can provide services to a restricted group called a closed subscriber group (CSG). The communication link 120 between base station 102 and UE 104 can include an uplink (UL) (also called a reverse link) transmission from UE 104 to base station 102 and / or a downlink (DL) (also called a forward link) transmission from base station 102 to UE 104. The communication link 120 can use multiple input multiple output (MIMO) antenna technology, including spatial multiplexing, beamforming and / or transmit diversity. The communication link can be through one or more carriers. For each carrier allocated in the carrier aggregation for up to a total of up to Yx MHz (x component carriers) for transmission in each direction, the base station 102 / UE 104 can use a spectrum with a bandwidth of up to Y megahertz (MHz) (e.g., 5 MHz, 10 MHz, 15 MHz, 20 MHz, 100 MHz, 400 MHz, etc.). The carriers may or may not be adjacent to each other. The allocation of carriers may be asymmetric with respect to DL and UL (e.g., more or fewer carriers may be allocated for DL ​​than for UL). The component carriers may include a primary component carrier and one or more secondary component carriers. The primary component carrier may be referred to as a primary cell (PCell) and the secondary component carrier may be referred to as a secondary cell (SCell).

[0055] Some UEs 104 may communicate with each other using a device-to-device (D2D) communication link 158. The D2D communication link 158 may use DL / UL WWAN spectrum. The D2D communication link 158 may use one or more sidelink channels, such as a physical sidelink broadcast channel (PSBCH), a physical sidelink discovery channel (PSDCH), a physical sidelink shared channel (PSSCH), and a physical sidelink control channel (PSCCH). The D2D communication may be through various wireless D2D communication systems, such as, for example, WiMedia, Bluetooth, ZigBee, Wi-Fi based on the Institute of Electrical and Electronics Engineers (IEEE) 802.11 standard, LTE, or NR.

[0056] The wireless communication system may also include a Wi-Fi access point (AP) 150 that communicates with a Wi-Fi station (STA) 152 via a communication link 154 in, for example, a 5 gigahertz (GHz) unlicensed spectrum. When communicating in the unlicensed spectrum, the STA 152 / AP 150 may perform a clear channel assessment (CCA) to determine whether a channel is available prior to communication.

[0057] The small cell 102' may operate in licensed and / or unlicensed spectrum. When operating in an unlicensed spectrum, the small cell 102' may employ NR and use the same unlicensed spectrum (e.g., 5 GHz, etc.) as used by the Wi-Fi AP 150. The small cell 102' employing NR in the unlicensed spectrum may improve the coverage of the access network and / or increase the capacity of the access network.

[0058] The electromagnetic spectrum is typically subdivided into various categories, bands, channels, etc. based on frequency / wavelength. In 5G NR, two initial operating bands have been identified as frequency range designations FR1 (410 MHz–7.125 GHz) and FR2 (24.25 GHz–52.6 GHz). Frequencies between FR1 and FR2 are typically referred to as mid-band frequencies. Although a portion of FR1 is greater than 6 GHz, FR1 is typically (interchangeably) referred to as the “sub-6 GHz” band in various documents and articles. A similar naming issue sometimes occurs with respect to FR2, which is typically (interchangeably) referred to as the “millimeter wave” band in documents and articles, despite being different from the extremely high frequency (EHF) band (30 GHz–300 GHz) identified as the “millimeter wave” band by the International Telecommunication Union (ITU).

[0059] In view of the above aspects, unless otherwise specifically stated, it should be understood that if the term "below 6 GHz" or the like is used herein, it can broadly represent frequencies that can be less than 6 GHz, can be within FR1, or can include mid-band frequencies. In addition, unless otherwise specifically stated, it should be understood that if the term "millimeter wave" or the like is used herein, it can broadly represent frequencies that can include mid-band frequencies, can be within FR2, or can be within the EHF band.

[0060] The base station 102, whether a small cell 102' or a large cell (e.g., a macro base station), may include and / or be referred to as an eNB, a gNodeB (gNB), or another type of base station. Some base stations, such as gNB 180, may operate in the traditional sub-6 GHz spectrum, in millimeter wave frequencies, and / or near millimeter wave frequencies to communicate with the UE 104. When the gNB 180 operates in millimeter wave or near millimeter wave frequencies, the gNB 180 may be referred to as a millimeter wave base station. The millimeter wave base station 180 may utilize beamforming 182 with the UE 104 to compensate for path loss and short distance. The base station 180 and the UE 104 may each include multiple antennas, such as antenna elements, antenna panels, and / or antenna arrays to facilitate beamforming.

[0061] The base station 180 may transmit a beamformed signal in one or more transmit directions 182' to the UE 104. The UE 104 may receive the beamformed signal from the base station 180 in one or more receive directions 182". The UE 104 may also transmit a beamformed signal to the base station 180 in one or more transmit directions. The base station 180 may receive the beamformed signal from the UE 104 in one or more receive directions. The base station 180 / UE 104 may perform beam training to determine the best receive direction and the best transmit direction for each of the base station 180 / UE 104. The transmit direction and the receive direction of the base station 180 may be the same or different. The transmit direction and the receive direction of the UE 104 may be the same or different.

[0062] The EPC 160 may include a mobility management entity (MME) 162, other MMEs 164, a serving gateway 166, an MBMS gateway 168, a broadcast multicast service center (BM-SC) 170, and a packet data network (PDN) gateway 172. The MME 162 may communicate with a home subscriber server (HSS) 174. The MME 162 is a control node that handles signaling between the UE 104 and the EPC 160. Typically, the MME 162 provides bearer and connection management. All user Internet Protocol (IP) packets are passed through the serving gateway 166, which itself is connected to the PDN gateway 172. The PDN gateway 172 provides UE IP address allocation and other functions. The PDN gateway 172 and the BM-SC 170 are connected to IP services 176. The IP services 176 may include the Internet, an intranet, an IP multimedia subsystem (IMS), a PS streaming service, and / or other IP services. The BM-SC 170 may provide functions for MBMS user service provision and delivery. The BM-SC 170 may serve as an entry point for content provider MBMS transmissions, may be used to authorize and initiate MBMS bearer services in a public land mobile network (PLMN), and may be used to schedule MBMS transmissions. The MBMS gateway 168 may be used to distribute MBMS services to base stations 102 belonging to a multicast broadcast single frequency network (MBSFN) area of ​​a broadcast specific service, and may be responsible for session management (start / stop) and for collecting eMBMS-related billing information.

[0063] The core network 190 may include an access and mobility management function (AMF) 192, other AMFs 193, a session management function (SMF) 194, and a user plane function (UPF) 195. AMF 192 may communicate with unified data management (UDM) 196. AMF 192 is a control node that handles signaling between UE 104 and the core network 190. Typically, AMF 192 provides quality of service (QoS) flow and session management. All user IP packets are delivered through UPF 195. UPF 195 provides UE IP address allocation and other functions. UPF 195 is connected to IP services 197. IP services 197 may include the Internet, intranet, IMS, packet switching (PS) streaming media services, and / or other IP services.

[0064] Base stations may include and / or be referred to as gNBs, Node Bs, eNBs, access points, base transceiver stations, radio base stations, radio transceivers, transceiver functions, basic service sets (BSSs), extended service sets (ESSs), transmit receive points (TRPs), or some other suitable terminology. Base stations 102 provide access points to EPC 160 or core network 190 for UEs 104. Examples of UEs 104 include cellular phones, smart phones, session initiation protocol (SIP) phones, laptops, personal digital assistants (PDAs), satellite radios, global positioning systems, multimedia devices, video devices, digital audio players (e.g., MP3 players), cameras, game consoles, tablet devices, smart devices, wearable devices, vehicles, electric meters, gas pumps, large or small kitchen appliances, healthcare devices, implants, sensors / actuators, displays, or any other similarly functional devices. Some of UEs 104 may be referred to as IoT devices (e.g., parking meters, gas pumps, toasters, vehicles, heart monitors, etc.). UE 104 may also be referred to as a station, a mobile station, a user station, a mobile unit, a user unit, a wireless unit, a remote unit, a mobile device, a wireless device, a wireless communication device, a remote device, a mobile user station, an access terminal, a mobile terminal, a wireless terminal, a remote terminal, a handset, a user agent, a mobile client, a client, or some other suitable terminology.

[0065] Reference again Figure 1 In some aspects, the UE 104 may include an adaptive ML model component 198 configured to receive a configuration instructing the UE to switch between multiple ML-based feature data extraction models of the UE based on the state of the UE; determine to switch from a first ML-based feature data extraction model to a second ML-based feature data extraction model in response to the configuration; and send ML-based feature data for beam blocking prediction to a network node based on the second ML-based feature data extraction model. In this case, the UE 104 may be located in a vehicle equipped with sensors and an ML inference model for feature extraction, but the UE may be another vehicle in the form of a car, or a pedestrian, etc. The network node may be, for example, an ML service entity at the base station 180 or an ML service entity in an ML server co-located with the base station 180 or located near the base station, the base station itself, the ML server itself, a component of a decomposed base station, or some other network entity. The UE may also include a local ML service entity containing an ML model for extracting features from sensor data. This local ML service entity at the UE is different from the ML service entity at the base station, co-located with the base station, or located near the base station, and the ML service entity can predict beam blocking based on aggregated sensor data or inferences from multiple UEs.

[0066] Still refer to Figure 1 , in some ways, Figure 1 The base station 180 in the embodiment may include an adaptive configuration component 199 configured to receive first ML-based feature data from the UE based on a first ML-based feature data extraction model of the UE; send a configuration indicating that the UE switches from the first ML-based feature data extraction model of the UE to a second ML-based feature data extraction model based on the state of the UE; receive second ML-based feature data from the UE based on the second ML-based feature data extraction model of the UE; and determine a beam blocking prediction in response to the second ML-based feature data. For example, the base station may determine the beam blocking prediction by obtaining one or more predicted blocked beams from an ML service entity or ML inference host that the base station can interface with. The ML service entity or the ML inference host may be physically located at the base station 180, or may be located in an ML server that is co-located with the base station 180 or located near the base station. The ML service entity or the base station or ML server interfaced with the ML service entity may exchange signals with the UE 104 to perform adaptive sensing and feature extraction as well as sensor configuration and reconfiguration. For example, the ML service entity or base station / ML server may engage in signaling procedures and parameters with the UE 104 to collect sensor data and inferences, request the UE to adjust its ML models and sensors, and obtain updated sensor data for beam blocking prediction at or from the ML service entity. For example, prediction and inference-based techniques may be used to determine the likelihood of beam blocking, where the ML service entity uses this information to suggest alternative beam selection paths, e.g., to avoid predicted line-of-sight blocking.

[0067] Although the present disclosure may focus on 5G NR, the concepts and aspects described herein may be applicable to other similar areas such as LTE, Advanced LTE (LTE-A), Code Division Multiple Access (CDMA), Global System for Mobile Communications (GSM), and / or other wireless / radio access technologies.

[0068] Additionally or alternatively, the concepts and various aspects described herein may be particularly applicable to one or more specific areas, such as for use in an open radio access network (O-RAN) architecture with a RAN intelligent controller (RIC), as described in more detail below.

[0069] In some aspects, the term "receive" and its variations (e.g., "receive" and / or "received", as well as other examples) may alternatively be referred to as "obtain" or its corresponding variations (e.g., "obtain" and / or "obtained", as well as other examples). Similarly, the term "send" and its variations (e.g., "send" and / or "sent", as well as other examples) may alternatively be referred to as "provide" or its corresponding variations (e.g., "provide" and / or "provided", as well as other examples), "generate" or its corresponding variations (e.g., "generate" and / or "generated", as well as other examples) and / or "output" or its corresponding variations (e.g., "output" and / or "outputted", as well as other examples).

[0070] Figure 2A FIG200 is a diagram illustrating an example of a first subframe within a 5G NR frame structure. Figure 2B is a diagram 230 illustrating an example of DL channels within a 5G NR subframe. Figure 2C FIG250 is a diagram illustrating an example of a second subframe within a 5G NR frame structure. Figure 2D FIG280 is a diagram illustrating an example of an UL channel within a 5G NR subframe. The 5G NR frame structure may be frequency division duplex (FDD) (wherein, for a particular set of subcarriers (carrier system bandwidth), a subframe within the subcarrier set is dedicated to either DL or UL), or may be time division duplex (TDD) (wherein, for a particular set of subcarriers (carrier system bandwidth), a subframe within the subcarrier set is dedicated to both DL and UL). Figure 2A , Figure 2C In the example provided, the 5G NR frame structure is assumed to be TDD, where subframe 4 is configured with slot format 28 (most of which are DL), where D is DL, U is UL, and F is flexible between DL / UL, and subframe 3 is configured with slot format 34 (most of which are UL). Although subframes 3 and 4 are shown as having slot formats 34 and 28, respectively, any particular subframe may be configured with any of the various available slot formats 0-61. Slot formats 0 and 1 are DL and UL, respectively. Other slot formats 2-61 include a mixture of DL, UL and flexible symbols. The slot format is configured for the UE by a received slot format indicator (SFI) (dynamically configured by DL control information (DCI) or semi-statically / statically configured by radio resource control (RRC) signaling). Note that the following description also applies to the 5GNR frame structure as TDD.

[0071] Other wireless communication technologies may have different frame structures and / or different channels. For example, a 10 millisecond (ms) frame may be divided into 10 equally sized subframes (1 ms). Each subframe may include one or more time slots. A subframe may also include a microslot, which may include 7, 4, or 2 symbols. Each time slot may contain 7 or 14 symbols, depending on the time slot configuration. For time slot configuration 0, each time slot may include 14 symbols, and for time slot configuration 1, each time slot may include 7 symbols. The symbols on the DL may be cyclic prefix (CP) orthogonal frequency division multiplexing (OFDM) (CP-OFDM) symbols. The symbols on the UL may be CP-OFDM symbols (for high throughput scenarios) or discrete Fourier transform (DFT) spread OFDM (DFT-s-OFDM) symbols (also known as single carrier frequency division multiple access (SC-FDMA) symbols) (for power-limited scenarios; limited to single stream transmission). The number of time slots in a subframe is based on the time slot configuration and parameter set. For slot configuration 0, different parameter sets μ0 to 4 allow 1, 2, 4, 8, and 16 slots per subframe, respectively. For slot configuration 1, different parameter sets 0 to 2 allow 2, 4, and 8 slots per subframe, respectively. Accordingly, for slot configuration 0 and parameter set μ, there are 14 symbols per slot and 2 per subframe. μ time slots. The subcarrier spacing and symbol length / duration are functions of the parameter set. The subcarrier spacing can be equal to 2 μ *15 kilohertz (kHz), where μ is parameter set 0 to 4. Therefore, the subcarrier spacing for parameter set μ=0 is 15 kHz, and the subcarrier spacing for parameter set μ=4 is 240 kHz. The symbol length / duration is inversely related to the subcarrier spacing. FIG. 2A to FIG. 2D An example is provided for slot configuration 0 with 14 symbols per slot and parameter set μ=2 with 4 slots per subframe. The slot duration is 0.25 ms, the subcarrier spacing is 60 kHz, and the symbol duration is approximately 16.67 μs. Within a frame set, there may be one or more different bandwidth parts (BWPs) that are frequency-division multiplexed (see Figure 2B ). Each BWP can have a specific set of parameters.

[0072] A resource grid may be used to represent a frame structure. Each slot includes a resource block (RB) (also referred to as a physical RB (PRB)) extending over 12 consecutive subcarriers. The resource grid is divided into a plurality of resource elements (REs). The number of bits carried by each RE depends on the modulation scheme.

[0073] like Figure 2A As illustrated in FIG. 1 , some of the REs carry reference (pilot) signals (RS) for the UE. The RS may include a demodulation RS (DM-RS) for channel estimation at the UE (indicated as R for a particular configuration).x , where 100x is the port number, but other DM-RS configurations are possible) and a channel state information reference signal (CSI-RS). RS may also include beam measurement RS (BRS), beam refinement RS (BRRS) and phase tracking RS (PT-RS).

[0074] Figure 2B Examples of various DL channels within a subframe of a frame are illustrated. The physical downlink control channel (PDCCH) carries DCI within one or more control channel elements (CCEs), each CCE including nine RE groups (REGs), and each REG includes four consecutive REs in an OFDM symbol. The PDCCH within a BWP may be referred to as a control resource set (CORESET). Additional BWPs may be located at higher and / or lower frequencies on the channel bandwidth. The primary synchronization signal (PSS) may be within symbol 2 of a specific subframe of a frame. The PSS is used by UE 104 to determine the subframe / symbol time and physical layer identification. The secondary synchronization signal (SSS) may be within symbol 4 of a specific subframe of a frame. The SSS is used by the UE to determine the physical layer cell identification group number and the radio frame timing. Based on the physical layer identification and the physical layer cell identification group number, the UE may determine the physical cell identifier (PCI). Based on the PCI, the UE may determine the position of the aforementioned DM-RS. The physical broadcast channel (PBCH) carrying the master information block (MIB) can be logically grouped with the PSS and SSS to form a synchronization signal (SS) / PBCH block (also called an SS block (SSB)). The MIB provides the number of RBs in the system bandwidth and the system frame number (SFN). The physical downlink shared channel (PDSCH) carries user data, broadcast system information not sent over the PBCH (such as the system information block (SIB)), and paging messages.

[0075] like Figure 2C As illustrated, some of the REs carry DM-RS (indicated as R for a particular configuration, but other DM-RS configurations are possible) for channel estimation at the base station. The UE may send a DM-RS for a physical uplink control channel (PUCCH) and a DM-RS for a physical uplink shared channel (PUSCH). The PUSCH DM-RS may be sent in the first or first two symbols of the PUSCH. Depending on whether a short PUCCH or a long PUCCH is sent and depending on the specific PUCCH format used, the PUCCH DM-RS can be sent in different configurations. The UE may send a sounding reference signal (SRS). The SRS may be sent in the last symbol of a subframe. The SRS may have a comb structure, and the UE may send the SRS on one of the combs. The SRS may be used by the base station for channel quality estimation to achieve frequency-dependent scheduling on the UL.

[0076] Figure 2DExamples of various UL channels within a subframe of a frame are illustrated. The PUCCH may be located at a position as indicated in one configuration. The PUCCH carries uplink control information (UCI), such as scheduling requests, channel quality indicator (CQI), precoding matrix indicator (PMI), rank indicator (RI), and hybrid automatic repeat request (HARQ) acknowledgement (ACK) / negative acknowledgement (NACK) feedback. The PUSCH carries data and may be used to carry, among other things, a buffer status report (BSR), a power headroom report (PHR), and / or UCI.

[0077] Figure 3 3 is a block diagram of a base station 310 in an access network communicating with a UE 350. In the DL, IP packets from the EPC 160 may be provided to a controller / processor 375. The controller / processor 375 implements layer 3 and layer 2 functionality. Layer 3 includes a radio resource control (RRC) layer, and layer 2 includes a service data adaptation protocol (SDAP) layer, a packet data convergence protocol (PDCP) layer, a radio link control (RLC) layer, and a medium access control (MAC) layer. The controller / processor 375 provides RRC layer functionality associated with broadcasting of system information (e.g., MIB, SIB), RRC connection control (e.g., RRC connection paging, RRC connection establishment, RRC connection modification, and RRC connection release), inter-radio access technology (RAT) mobility, and measurement configuration for UE measurement reporting; PDCP layer functionality associated with header compression / decompression, security (encryption, decryption, integrity protection, integrity verification), and handover support functions; RLC layer functionality associated with delivery of upper layer packet data units (PDUs), error correction through ARQ, concatenation, segmentation, and reassembly of RLC service data units (SDUs), re-segmentation of RLC data PDUs, and reordering of RLC data PDUs; and MAC layer functionality associated with mapping between logical channels and transport channels, multiplexing of MAC SDUs onto transport blocks (TBs), demultiplexing of MAC SDUs from TBs, scheduling information reporting, error correction through HARQ, priority handling, and logical channel prioritization.

[0078] The transmit (TX) processor 316 and receive (RX) processor 370 implement layer 1 functionality associated with various signal processing functions. Layer 1 (which includes the physical (PHY) layer) may include error detection on the transmission channel, forward error correction (FEC) decoding / decoding of the transmission channel, interleaving, rate matching, mapping to the physical channel, modulation / demodulation of the physical channel, and MIMO antenna processing. The TX processor 316 processes the mapping to the signal constellation based on various modulation schemes (e.g., binary phase shift keying (BPSK), quadrature phase shift keying (QPSK), M-order phase shift keying (M-PSK), M-order quadrature amplitude modulation (M-QAM)). The decoded and modulated symbols may then be divided into parallel streams. Subsequently, each stream may be mapped to an OFDM subcarrier, each stream may be multiplexed with a reference signal (e.g., a pilot) in the time domain and / or frequency domain, and each stream may then be combined using an inverse fast Fourier transform (IFFT) to generate a physical channel carrying a time domain OFDM symbol stream. The OFDM stream is spatially pre-coded to generate multiple spatial streams. Channel estimates from a channel estimator 374 may be used to determine the coding and modulation schemes, as well as for spatial processing. Channel estimates may be derived from reference signals and / or channel condition feedback sent by the UE 350. Each spatial stream may then be provided to a different antenna 320 via a separate transmitter 318TX. Each transmitter 318TX may modulate an RF carrier with a corresponding spatial stream for transmission.

[0079] At the UE 350, each receiver 354RX receives a signal through its corresponding antenna 352. Each receiver 354RX recovers the information modulated onto the RF carrier and provides the information to a receive (RX) processor 356. The TX processor 368 and the RX processor 356 implement layer 1 functionality associated with various signal processing functions. The RX processor 356 can perform spatial processing on the information to recover any spatial stream destined for the UE 350. If multiple spatial streams are destined for the UE 350, they can be combined into a single OFDM symbol stream by the RX processor 356. The RX processor 356 then converts the OFDM symbol stream from the time domain to the frequency domain using a fast Fourier transform (FFT). The frequency domain signal includes a separate OFDM symbol stream for each subcarrier of the OFDM signal. The symbols on each subcarrier and the reference signal are recovered and demodulated by determining the signal constellation point most likely to be sent by the base station 310. These soft decisions can be based on channel estimates calculated by the channel estimator 358. The soft decisions are then decoded and deinterleaved to recover the data and control signals originally sent on the physical channel by base station 310. The data and control signals are then provided to controller / processor 359, which implements layer 3 and layer 2 functionality.

[0080] The controller / processor 359 may be associated with a memory 360 that stores program codes and data. The memory 360 may be referred to as a computer readable medium. In the UL, the controller / processor 359 provides demultiplexing between transport and logical channels, packet reassembly, decryption, header decompression, and control signal processing to recover IP packets from the EPC 160. The controller / processor 359 is also responsible for error detection using an ACK and / or NACK protocol to support HARQ operations.

[0081] Similar to the functionality described in conjunction with DL transmissions performed by the base station 310, the controller / processor 359 provides RRC layer functionality associated with system information (e.g., MIB, SIB) acquisition, RRC connection, and measurement reporting; PDCP layer functionality associated with header compression / decompression and security (encryption, decryption, integrity protection, integrity verification); RLC layer functionality associated with delivery of upper layer PDUs, error correction through ARQ, concatenation, segmentation and reassembly of RLC SDUs, resegmentation of RLC data PDUs, and reordering of RLC data PDUs; and MAC layer functionality associated with mapping between logical channels and transport channels, multiplexing of MAC SDUs onto TBs, demultiplexing of MAC SDUs from TBs, scheduling information reporting, error correction through HARQ, priority handling, and logical channel prioritization.

[0082] Channel estimates derived by the channel estimator 358 from a reference signal or feedback transmitted by the base station 310 may be used by the TX processor 368 to select appropriate coding and modulation schemes and to facilitate spatial processing. The spatial streams generated by the TX processor 368 may be provided to different antennas 352 via separate transmitters 354TX. Each transmitter 354TX may modulate an RF carrier with a corresponding spatial stream for transmission.

[0083] UL transmissions are processed at the base station 310 in a manner similar to that described in conjunction with the receiver functionality at the UE 350. Each receiver 318RX receives a signal through its respective antenna 320. Each receiver 318RX recovers information modulated onto an RF carrier and provides the information to a RX processor 370.

[0084] The controller / processor 375 may be associated with a memory 376 that stores program codes and data. The memory 376 may be referred to as a computer readable medium. In the UL, the controller / processor 375 provides demultiplexing between transport channels and logical channels, packet reassembly, decryption, header decompression, control signal processing to recover IP packets from the UE 350. The IP packets from the controller / processor 375 may be provided to the EPC 160. The controller / processor 375 is also responsible for error detection using ACK and / or NACK protocols to support HARQ operations.

[0085] At least one of the TX processor 368, the RX processor 356, and the controller / processor 359 may be configured to perform operations related to Figure 1 198. At least one of the TX processor 316, the RX processor 370, and the controller / processor 375 may be configured to perform operations related to Figure 1 The adaptive configuration component 199 combines various aspects.

[0086] With the advent of new wireless technologies and higher transmit beam frequencies, users can enjoy many of the attendant benefits of these technologies, such as faster data rates, artificial intelligence, and more sophisticated machine learning models for performing various tasks. Such technologies, such as 5G and the latest Wi-Fi standards, can be used in conjunction with different architectures, such as O-RAN. In decentralized base stations where base station functionality may be physically distributed, the ML service entity that performs beam interference prediction may be located at the ML server, or at a near real-time RAN intelligent controller (RIC) or at a different network node dictated by the details of the specific architecture.

[0087] In addition to these enhanced benefits, high frequencies also bring new challenges to the exemplary coverage area served by the base station, which may be located in a traffic congested area in an urban area, for example. The faster network speeds and higher frequencies of 5G operating at 28 GHz to 100 GHz or higher and the increased number of beams are more likely to cause LOS blocking, which, if not addressed, can severely degrade the performance of the system. These potential blocking issues may be exacerbated by the higher attenuation and diffraction losses inherent in these higher frequencies. For these reasons, it is important to establish an effective set of protocols to predict such blocking caused by moving obstacles such as pedestrians, vehicles or other objects, and redirect communications in real time or near real time to prevent such blocking.

[0088] It should be noted that the term "UE" in this disclosure may generally refer to a UE equipped in a vehicle, as can generally be seen from the context. For the same reason, the use of the term "vehicle" may also cover the UE and / or physical sensors equipped within the UE. However, the present disclosure is not limited in this regard, as the UE herein may also refer to any UE, whether carried by a user, integrated in a car, truck or train, or otherwise.

[0089] As a starting point to overcome these expected rapid changes in link quality of communication systems operating at these higher frequencies due in part to LOS path obstructions, manufacturers can equip UE-based vehicles with one or more onboard sensors to provide rapid radio network information to base stations. These sensors can include one or more cameras, radio detection and ranging systems (RADAR), and light detection and ranging systems (LIDAR), among others. The sensors can be coupled to the UE in the vehicle to send sensed information about the communication environment in the relevant coverage area, excluding moving obstacles that potentially block the LOS path and degrade communication quality.

[0090] In one aspect, relevant network components can employ perceptual wireless communications. For example, upon receiving sensing information provided by vehicle sensors, the radio network can employ ML models to detect or predict expected blockages and proactively initiate beam management and, if necessary, handover procedures.

[0091] While various aspects may involve multiple vehicles communicating with a network and aggregating that information, for simplicity, in some configurations, the present disclosure relates to, for example, related communications between a vehicle and a base station, rather than multiple vehicles equipped with sensors and ML capabilities. Reference to a single UE-based vehicle is for simplicity and to avoid unduly obscuring the concepts herein. However, those skilled in the art will appreciate upon reading this disclosure that a coverage area may involve communications with multiple UEs, in a vehicle, and otherwise.

[0092] Thus, in one aspect, the goal of this document (such as in the case of mm wavelength signaling) is to collect sensing information from each equipped UE in the coverage area and utilize one or more ML models to predict beam blocking and optimal beams. Aspects of the present disclosure are particularly directed to addressing how an ML service entity at an ML server or base station can perform discovery of these UE-based vehicles that support sensor-based ML functions, and addressing how an ML service session between the ML service entity and the vehicle-based UE can be efficiently established if such ML service discovery can be achieved, so that the ML service entity can collect relevant sensing information for ML training, inference, and performance optimization. Other aspects of the present disclosure are also discussed herein.

[0093] In addition to performing other functions, the ML server or ML service entity at the base station may be primarily responsible for mediating UE / ML server communications and processing sensing information, extracting features, etc., to ultimately perform beam prediction using dynamic and adaptive ML training and inference. For example, the ML service entity may include one or more ML models to predict or infer beam blocking based on received sensing information, or may perform training of one or more ML models for predicting blocking. The ML service entity may reside in the base station. In other configurations, the ML service entity may reside in an ML server that is co-located with the base station or located near the base station.

[0094] In some cases, other configurations involving alternative network deployments may affect the physical or virtual location of the ML service entity. One example of such a configuration includes a disaggregated network architecture, in which the ML service entity may be physically or logically deployed in a network node separate from the network nodes of a disaggregated base station. For example, a base station may include multiple units or network nodes, such as a central or centralized unit (CU), a distributed unit (DU), a radio unit (RU), etc., and the ML service entity may be physically or logically separated from one or more of these network nodes.

[0095] More generally, the deployment of a communication system (such as a 5G new radio (NR) system) can be arranged with various components or components in a variety of ways. In a 5G NR system or network, a network node, a network entity, a mobility element of a network, a radio access network (RAN) node, a core network node, a network element or network equipment such as a base station (BS), or one or more units (or one or more components) performing base station functionality can be implemented in an aggregated or decomposed architecture. For example, a base station (BS) (such as a node B (NB), an evolved NB (eNB), an NR BS, a 5G NB, an access point (AP), a transmit receive point (TRP) or a cell, etc.) can be implemented as an aggregated base station (also referred to as an independent BS or a monolithic BS) or a decomposed base station.

[0096] The clustered base station can be configured to utilize a radio protocol stack physically or logically integrated within a single RAN node. The decomposed base station can be configured to utilize a protocol stack physically or logically distributed between two or more units (such as one or more CUs, one or more DUs, or one or more RUs). In some aspects, the CU can be implemented within a RAN node, and one or more DUs can be co-located with the CU, or alternatively, can be geographically or virtually distributed in one or more other RAN nodes. The DU can be implemented to communicate with one or more RUs. Each of the CU, DU, and RU can also be implemented as a virtual unit, i.e., a virtual central unit (VCU), a virtual distributed unit (VDU), or a virtual radio unit (VRU).

[0097] Base station type operations or network designs may take into account the aggregated nature of base station functionality. For example, a disaggregated base station may be utilized in an integrated access backhaul (IAB) network, an open radio access network (O-RAN (network configurations such as those initiated by the O-RAN Alliance)), or a virtualized radio access network (vRAN, also known as a cloud radio access network (C-RAN)). Decomposition may include distributing functionality across two or more units at various physical locations, as well as virtually distributing functionality of at least one unit, which may enable flexibility in network design. Individual units of a disaggregated base station or disaggregated RAN architecture may be configured for wired or wireless communication with at least one other unit.

[0098] Figure 4 A diagram illustrating an example decomposed base station 400 architecture is illustrated. The decomposed base station 400 architecture may include one or more CUs 410 that may communicate directly with a core network 420 via a backhaul link, or indirectly with a core network 420 through one or more decomposed base station units, such as a near real-time RIC 425 via an E2 link, or a non-real-time RIC 415 associated with a service management and orchestration (SMO) framework 405, or both. The CU 410 may communicate with one or more DUs 430 via respective midhaul links, such as an F1 interface. The DU 430 may communicate with one or more RUs 440 via respective fronthaul links. The RU 440 may communicate with respective UEs 104 via one or more radio frequency (RF) access links. In some implementations, a UE 104 may be served simultaneously by multiple RUs 440.

[0099] Each of the units (i.e., CU 410, DU 430, RU 440, and near-RT RIC 425, non-RT RIC 415, and SMO framework 405) may include or be coupled to one or more interfaces configured to receive or send signals, data, or information (collectively, signals) via a wired or wireless transmission medium. Each of the units or an associated processor or controller that provides instructions to the communication interface of these units may be configured to communicate with one or more of the other units via a transmission medium. For example, these units may include a wired interface configured to receive or send signals to one or more of the other units on a wired transmission medium. In addition, these units may include a wireless interface that may include a receiver, a transmitter, or a transceiver (such as a radio frequency (RF) transceiver) that is configured to receive or send signals or both to one or more of the other units on a wireless transmission medium.

[0100] In some aspects, CU 410 may host higher layer control functions. Such control functions may include radio resource control (RRC), packet data convergence protocol (PDCP), service data adaptation protocol (SDAP), etc. Each control function may be implemented as having an interface configured to communicate signals with other control functions hosted by CU 410. CU 410 may be configured to handle user plane functionality (i.e., central unit-user plane (CU-UP)), control plane functionality (i.e., central unit-control plane (CU-CP)), or a combination thereof. In some specific implementations, CU 410 may be logically split into one or more CU-UP units and one or more CU-CP units. When implemented in an O-RAN configuration, the CU-UP unit may communicate bidirectionally with the CU-CP unit via an interface (such as an E1 interface). As needed, CU 410 may be implemented to communicate with DU 430 for network control and signaling.

[0101] DU 430 may correspond to a logical unit that includes one or more base station functions for controlling the operation of one or more RUs 440. In some aspects, DU 430 may host one or more of a radio link control (RLC) layer, a medium access control (MAC) layer, and one or more high physical (PHY) layers (such as modules for forward error correction (FEC) encoding and decoding, scrambling, modulation and demodulation, etc.) at least in part according to a functional split such as those defined by the 3rd Generation Partnership Project (3GPP). In some aspects, DU 430 may further host one or more low PHY layers. Each layer (or module) may be implemented with an interface that is configured to communicate signals with other layers (and modules) hosted by DU 430 or with control functions hosted by CU 410.

[0102] The lower layer functionality may be implemented by one or more RUs 440. In some deployments, a RU 440 controlled by a DU 430 may correspond to a logical node that hosts RF processing functions or low PHY layer functions (such as performing Fast Fourier Transform (FFT), Inverse FFT (iFFT), digital beamforming, physical random access channel (PRACH) extraction and filtering, etc.), or both, based at least in part on a functional split (such as a lower layer functional split). In this architecture, the RU 440 may be implemented to handle over-the-air (OTA) communications with one or more UEs 120. In some implementations, real-time and non-real-time aspects of control plane and user plane communications with the RU 440 may be controlled by the corresponding DU 430. In some scenarios, this configuration may enable the implementation of the DU 430 and the CU 410 in a cloud-based RAN architecture (such as a vRAN architecture).

[0103] The SMO framework 405 may be configured to support RAN deployment and provisioning of non-virtualized network elements and virtualized network elements. For non-virtualized network elements, the SMO framework 405 may be configured to support the deployment of dedicated physical resources for RAN coverage requirements, which may be managed via an operation and maintenance interface (such as an O1 interface). For virtualized network elements, the SMO framework 405 may be configured to interact with a cloud computing platform (such as an open cloud (O-Cloud) 490) to perform network element lifecycle management (such as instantiating virtualized network elements) via a cloud computing platform interface (such as an O2 interface). Such virtualized network elements may include, but are not limited to, CU 410, DU 430, RU 440, and near-RT RIC 425. In some specific implementations, the SMO framework 405 may communicate with hardware aspects of the 4G RAN (such as an open eNB (O-eNB) 411) via the O1 interface. Additionally, in some specific implementations, the SMO framework 405 may communicate directly with one or more RUs 440 via the O1 interface. The SMO framework 405 may also include a non-RT RIC 415 configured to support the functionality of the SMO framework 405 .

[0104] The non-RT RIC 415 may be configured to include logic functions that implement non-real-time control and optimization of RAN elements and resources, artificial intelligence / machine learning (AI / ML) workflows including model training and updating, or policy-based guidance of applications / features in the near-RT RIC 425. The non-RT RIC 415 may be coupled to or communicate with the near-RT RIC 425 (such as via an A1 interface). The near-RT RIC 425 may be configured to include logic functions that implement near-real-time control and optimization of RAN elements and resources via data collection and actions through an interface (such as via an E2 interface) that connects one or more CUs 410, one or more DUs 430, or both, and the O-eNB with the near-RT RIC 425.

[0105] In some implementations, in order to generate an AI / ML model to be deployed in the near-RT RIC 425, the non-RT RIC 415 may receive parameters or external enrichment information from an external server. Such information may be utilized by the near-RT RIC 425 and may be received from a non-network data source or from a network function at the SMO framework 405 or the non-RT RIC 415. In some examples, the non-RT RIC 415 or the near-RT RIC 425 may be configured to tune RAN behavior or performance. For example, the non-RT RIC 415 may monitor long-term trends and patterns of performance and employ AI / ML models to perform corrective actions through the SMO framework 405 (such as via reconfiguration of O1) or via the creation of RAN management policies (such as A1 policies).

[0106] Figure 5 500. O-RAN 500 advantageously provides an open architecture to which different network operators can link to provide a systematic, interoperable network. A conventional O-RAN architecture can be configured to include certain key functional elements. These include a SMO framework 505 (which may include a non-RT RIC 556, as well as Figure 4 ), O-DU software component 562, multi-RAT CU protocol stack (which in turn may include O-RAN CU-CP (O-CU-CP) 560 and O-RAN CU-UP (O-CU-UP) 561), near-RT RIC 525, O-RAN eNB 570, infrastructure management framework 558, and infrastructure commercial off-the-shelf (COTS) / white box / peripheral hardware and virtualization layer 568. Return to reference Figure 4 , the SMO framework 505 may correspond to the SMO framework 405, the non-RT RIC 556 may correspond to the non-RT RIC 415, the O-DU software component 562 may correspond to the DU 430, the O-CU-CP 560 and the O-CU-UP 561 may correspond to the functionality of the CU 410, the near-RT RIC 525 may correspond to the near-RT RIC425, the O-RAN eNB 570 may correspond to the O-eNB 411, and the infrastructure management framework 558 may be included in the O-Cloud 490.

[0107] For simplicity and to avoid unduly obscuring the present disclosure, Figure 5The architecture omits various inputs and outputs, including signals to and from O-CU-CP 560 and O-CU-UP 561. Non-RT components 556 are capable of processing data offline or quickly, but typically greater than a threshold, such as one millisecond. Thus, non-RT components 556 may handle tasks that do not have an immediate need for a response but may have a time tolerance that makes non-RT components 556 a natural choice for processing data.

[0108] One or more of these components may also interact with the radio unit (RU) hardware 564. For example, the O-DU component 562 may communicate with the O-RU component 564 via an open fronthaul interface 590. Components such as the non-RT RIC 556 and the near-RT RIC 525 may interact with the O-RU hardware 564 to help the O-RU 564 operate more efficiently and optimize the O-RU 564 in real time as part of the RAN cluster to deliver a better network experience to end users. Both the non-RT RIC 556 and the near-RT RIC 525 may be used in conjunction with service discovery and service session processes due to their ability to process priority data at high speed or in the background.

[0109] As reference Figure 4 and Figure 5 As discussed, the non-RT RIC 415, 556 includes several functions that are potentially relevant to aspects of the present invention. The non-RT RIC 415, 556 includes functions such as configuration management, device management, fault management, performance management, and lifecycle management of all network elements. The non-RT RIC 415, 556 can use data analytics, artificial intelligence (AI), and ML training and inference to determine RAN optimization actions for which it can utilize the service management and orchestration (SMO) framework 405, 505, including data collection and provisioning services for O-RAN nodes.

[0110] The near-RT RIC 425, 525 can utilize embedded processors or intelligent code for load balancing, RB management, interference detection and mitigation for per-UE control, and other functions that are expected to be handled in a prioritized manner in order to successfully use ML training / inference models. The near-RT RIC 425, 525 can provide quality of service (QoS) management, connection management, and seamless handover control. The near-RT RIC 425, 525 can also utilize the near-real-time status of the underlying network and can feed RAN data to train AI / ML models. The modified model can then be provided to the near-RT RIC to facilitate high-quality radio resource management for subscribers.

[0111] In some configurations, the near-RT RIC 425, 525 performs similar beam prediction management functions as the non-RT RIC 415, 556 for data that does not require near-RT priority. More commonly, due to the nature of its time priority, the near-RT RIC 425, 525 performs different ML models and beam interference predictions for different participants (such as, O-CU-CP 560, O-CU-UP 561, O-DU 562, and O-RU 564). The last four components are functions with the base station, and these four elements show the decomposition of the elements in the architecture. In addition, in this configuration, the near-RT RIC 425, 525 is co-located with the gNB because it supports loop delays in inferred operations that are faster than 1 second.

[0112] As mentioned, the non-RT RIC 415, 556 can support inference operations with latency slower than 1ms and can be located near the gNB, such as near a cloud or edge server. In short, the near-RT RIC 425, 525 or non-RT RIC can act as an inference master in the beam prediction architecture, and in a decomposed base station, the four participants 560, 561, 562, and 564 are part of the gNB application.

[0113] In short, about reference Figure 4 and Figure 5 Different contemplated network configurations and server-based architectures are described, including converged and disaggregated base stations, O-RAN, SMO (including non-RT RIC), near-RT RIC, or other network frameworks or modifications thereof, and the principles of the present disclosure are intended to cover any one or more of these specific implementations.

[0114] Figure 6600 is a flow chart of an example training and inference model. The data source 612 may include training and inference data collected from network entities. The model training host 621 and the model inference host 614 may each be included in the near RT RIC 425, 525 and / or the non-RT RIC 415, 556, or alternatively synonymous therewith. Model deployment or updates from the model training host 621 may be fed back to the model inference host 614, and model performance feedback may be provided to the model training host 621. The training and inference outputs from the model inference host 614 may be provided to the action party 616. The action party 616 may be any entity within the 3GPP network. For example, if the action party is a gNB, the action subjects 617a and 617b may update the models, which in turn may be provided back to the data source 612 as performance feedback 618. Other action subjects 617a, 617b may include energy saving, load balancing, mobility management, coverage optimization, etc. Actions that do not require near real-time processing may be processed via the non-RT RIC 415, 556. As previously described, actions related to model building and inference training based on extracted sensor information (e.g., oriented bounding box (OBB) for estimating object position, scale, and orientation based on the YOLO algorithm) and inference model are performed by the ML service entity at the ML server or base station.

[0115] Figure 7700 is a top view of a coverage area 700 of wireless communications including use of adaptive ML training and inference extraction for beam prediction. A vehicle 704a (which may be referred to as an "autonomous" vehicle) may enter a coverage area, such as a cell of a gNB 702. The vehicle / UE 704a may be equipped with sensors, such as cameras, RADAR, LIDAR, or other sensing equipment. When the vehicle enters the area, the gNB 702 may transmit a transmission 732 to the UE 704a requesting sensing information. The sensing information may include, for example, raw sensor data or inferred data. The raw sensor data may include, for example, RADAR or LIDAR point clouds, camera pixels, etc., which are obtained from sensors of the vehicle / UE. The inferred data may include, for example, directional bounding boxes (OBBs) or other extracted features, which the vehicle / UE 704a may derive as output from an ML model in the UE's ML service entity 705 based on the raw sensor data. The ML model used for extraction may be, for example, a neural network, such as a multi-layer perceptron (MLP), a convolutional neural network (CNN), or a recurrent neural network (RNN). The vehicle 704a may respond to the transmission 732 by transmitting sensing information that may be used as training data or inferred data for beam blocking prediction. For example, training data is data that the ML service entity 749 of the gNB may use to obtain or train a beam blocking prediction model, and inferred data is data that the ML service entity 749 of the gNB may use to predict beam blocking using the trained model. For example, the ML service entity 749 of the gNB may determine based on the inferred data that there is a feature in the LOS of the beam to the vehicle / UE 704a. For example, the feature may be a wall or a pedestrian.

[0116] The ML service entity 749 may be located in the gNB 702, or co-located with the gNB or near the gNB in ​​the ML server, and includes a model training host and a model inference host. The model training host includes a training model, such as a neural network, which typically determines and deploys weights for the inference model at the model inference host. The inference model may be, for example, a neural network that performs beam blocking prediction based on inference data received from the UE. For example, the inference model at the ML service entity 749 may be a beam blocking prediction model. The inference host may provide outputs from the inference model to detect blocked beams. The outputs from the ML service entity 749 may be provided to an action party, such as a base station gNB 702. The ML service entity 749 associated with the gNB 702 may also analyze the data and create gradients, and provide the gradients back to the model training host so that new weights may be provided and the training model may be updated (and optionally, based on performance feedback, the inference model may be updated). Based on this performance optimization, new actions may also be performed at the gNB 702, such as changing beams in response to information provided by one or more vehicles.

[0117] The coverage area 700 may include static objects (such as buildings) and moving objects (such as cars, buses, trucks, pedestrians, etc.). The quality of the radio link between the UE and the gNB may be affected by both moving objects and fixed objects. For example, the gNB 702 may currently schedule a directional transmit beam 707 to a pedestrian UE 704d, which has a receive beam 709.

[0118] Still refer to Figure 7 , the ML service entity 749 may query the sensing information based on the sensors of the vehicle UEs 704a, 704b, and 704c. The ML service entity 749 may instruct one or more of these vehicles to adjust the direction or orientation of the sensors, such as the field of view (FoV) 764a, 764b, 764c, in order to obtain more accurate feedback. The type of the UE's sensor may change. The ML service entity 749 may make these determinations based on the obstacles it finds in the coverage area, including buildings 722, 724, 726, and 730, and pedestrians 751 and 704d. Based on the exchange of exchanged training and inference data and the resulting model, the ML service entity 749 may change the pattern of beam prediction to reduce the chance of signal interference. For example, the ML service entity 749 may predict that the truck 711 may enter the LOS of the beam 707 to the pedestrian 704d and cause beam blocking, and the gNB may change to a different beam accordingly.

[0119] It should be noted that the above Figures 5 to 7Example environments are shown as some possible implementations to which aspects of the present disclosure may be applied. These illustrations also highlight the differences between a base station monolithic architecture and a disaggregated architecture. Although the network architecture may be different in these cases, the principles of the present disclosure are intended to be applied to different situations with equal force. Other network types are also possible and may be equally suitable for applying the principles herein.

[0120] Figure 8 A conceptual diagram 800 of an example of an ML service entity 830 and an ML service entity 840 at a UE, respectively, interfacing with a gNB 804 (or ML server), and applying signaling procedures and parameters for adaptive sensing and feature extraction and sensor reconfiguration. The ML service entity 840 at the UE generally assists in collecting sensing information and communicates with the ML service entity 830 at the gNB or ML server. The ML service entity 840 at the UE includes a sensor and one or more ML functions, an example of which is shown in an enlarged form as a sensor and ML function 810. In this configuration, the ML service entity 840 is located on top of the protocol stack 841 of the UE modem. At the sensor and ML function block 810 is a sensor coverage information component 810a, which provides UE sensor parameters such as field of view, resolution, update rate, etc. Also shown in this example is a radar cloud 810b system and a camera 810c, where the ML service uses RADAR and camera and respectively combines a neural network NN a and NN b Object detection or inference models (e.g., based on the You Only Look Once (Yolo) algorithm). a and NN b Features such as directional bounding boxes (OBBs) are extracted from the RADAR point cloud or camera pixels sensed by the UE, respectively. Although the ML model Yolo is specifically referenced for the neural network in this example, it should be understood that the neural network is not limited to this architecture and can be based on other frameworks including but not limited to MobileNet and EfficientNet in other examples.

[0121] In order to make near real-time predictions about potential beam interference and performance degradation, there is a need in the art to establish mechanisms for adaptive sensing and sensor reconfiguration. Due to the mobility of sensor-equipped vehicles and the dynamic nature of detectable objects, sensing and feature extraction should be adaptive to better serve the training, inference, and performance optimization tasks of the ML service entity. In one case, the ML service entity 830 can be configured to control the sensors of the UE and the model at the ML service entity 840 of the UE by exchanging messages for adaptive sensing and feature extraction. In another case, the ML service entity 830 can be configured to configure or reconfigure the sensors of the UE through a configuration or reconfiguration request, or to receive and confirm a sensor configuration or reconfiguration request from the UE to adapt to the dynamic environment of the UE. Adaptive sensing and sensor reconfiguration can also allow the ML service entity 830 to improve its training, inference, and performance optimization tasks to make more accurate beam blocking predictions.

[0122] Return to reference Figure 8 , the ML service entity 840 at the UE includes a control / management component 842, which includes an adaptive sensing function and a sensor reconfiguration function. The ML service entity 830 at the ML server or base station also includes a corresponding control / management component 867, which includes an adaptive sensing function and a sensor reconfiguration function. These components together form one or more data streams related to the beam blocking prediction service. Therefore, in this regard, adaptive sensing is configured by communication from the adaptive sensing function of the control / management component 867 of the ML service entity 830. In addition, on the other hand, the sensor reconfiguration can be indicated by a request or other communication from the control / management component 842, 867 of the ML service entity 830 or the sensor reconfiguration function of the ML service 840 at the UE.

[0123] exist Figure 8 In the aspects shown and as will become more apparent later, the communication between the entities does not occur at lower layers, but may in some examples occur at layers above the protocol stack of the UE modem on the UE side, and the ML service entity 830 similarly resides in layers above the RAN (gNB at 804, including an exemplary hierarchy of one or more CUs, DUs, and RUs in this deployment) on the gNB side. Still referring to Figure 8 , the ML service entity 830 also includes an ML engine 805 (or multiple such engines). An enlarged version of the ML engine 805 is shown in the upper right of the figure. The ML engine 805 can use data from multiple UEs in the coverage area to perform beam blocking prediction, provide feedback to the UE, and perform beam management functions to make the required changes when necessary.

[0124] In the upper right of the enlarged ML engine 805, extracted features 811.1-811.N from multiple UEs 1-N can be received and aggregated into N channel input 822. In general, the aggregated features provide a set of features or inferences at a certain moment in time. In a sense, they provide a basis for making beam blockage predictions. The shape of the features can change over time as vehicles and pedestrians move and other dynamic events occur.

[0125] In addition, the UE may provide aggregated sensing coverage data 855 (e.g., a combination of various UE sensor parameters) and location information 856, such as the UE's transmit and receive locations, angle of descent (AoD), etc. In addition to the features from the aggregator (N channel input 822), the sensing coverage aggregate data 855 and location information 856 are provided to an inference model including one or more neural networks (NNs) 826 for beam blocking prediction. The output of the neural network 826 with the aggregated data includes feature predictions 828, such as predicted beam blocking, potential changes to beam / TX spatial pre-decoders, potential changes to Tx FD / TD pre-decoders, etc. This information can be used to modify the pre-decoders and change communications to avoid or mitigate the occurrence of beam blocking.

[0126] Therefore, from Figure 8 It is apparent that in order to achieve the desired results of adaptive sensing and sensor reconfiguration, an initial communication link is required to initiate service discovery of sensors and ML models, as well as subsequent session data where the desired parameters can be obtained. Adaptive sensing and sensor reconfiguration can be completed after service discovery and session establishment have been completed, as further described below.

[0127] Fig. 9 900 is an example flow chart of adaptive ML training and inference functionality for predicting beam blocking using a session based on data collected from sensors and ML models of a UE. In one aspect, the diagram 900 includes a sensor data collector 932, a model training (non-RT) component 980, a RAN 993, and an end user 104a. The RAN 993 may include a model inference (near RT) component 995 and an actor 991, such as a gNB 102 or other network node. In this example, the model training (non-RT) component 980 may be a component of or act as a non-RT RIC 415, 556, and the model inference (near RT) component 995 may be a component of or act as a near RT RIC 425, 525. The sensor data may include a UE-based vehicle 104 equipped with sensors 930 and one or more ML models 935 for object detection or feature extraction.

[0128] At 901, raw sensor data may be provided to an ML model 935 where object detection and feature extraction may be performed. At 902, non-RT training data is submitted from the UE 104 to the non-RT model training component 980 and stored / processed at the data management module 937. The data is thereafter sent in sequence to the training module 939 where predictions may be made on the non-RT data. Additionally, at or about the same time (at 902), training data such as non-RT beam information may be sent from the actor 991 (gNB or network node (CU, DU, etc.) in a decomposed base station) to the data management module 937. The training data from the gNB / network node may be used together with the training data from the UE 104 to make predictions at the training module 939.

[0129] Thereafter, at 903, the training component 939 of the non-RT module 980 sends model deployment or update data based on these predictions to the near RT model inference component 995. At 904, the near RT inference data from the UE 104 is sent to the near RT model inference component 995 and provided to the data management unit 999. Similarly, at 904, the inference data including beam blocking information (in near RT) is passed from the actor 991 (e.g., a gNB or one or more network nodes in a decomposed configuration) to the ML model 997 for prediction. At the near RT model inference unit 995, the inference data at 999 may be provided to the ML prediction model 997 for beam blocking prediction.

[0130] At 905, the beam blocking prediction is provided to the action party 991. At 906, the action determined to be responsive to the prediction can be forwarded to the various end users 104a (including the UE 104). At 907, the end user who receives the action data can provide feedback to the action party 991. At the same time, the action party 991 can provide feedback to the near RT unit 995 for performance monitoring. If necessary, the near RT unit 995 forwards the model performance feedback to the non-RT training model 980 at 908 for use in the non-RT training component 939.

[0131] It is worth noting that Figure 6 Unlike the example where the data source 612 collects training or inference data internally, in this configuration, the data is collected externally from the sensor. Thus, the sensor-equipped vehicle UE 104 acts as a data collector, rather than having a data storage for storing internally collected data. In this example, Fig. 9 The action party 991 in the O-RAN may be, for example, a gNB, and the subject of the action may be to instruct the UE 104 to change the beam or use a different sensor, etc. On the other hand, the action party may be a network node in the O-RAN—CU, DU, or some combination thereof.

[0132] Fig.10 1000 is a conceptual diagram of an example in which the training or feature extraction model at the UE may be separated from the training or inference of the prediction model at the ML service entity of the ML server or base station. UE 1 receives sensory data 1002 into a feature extraction model 1004, which extracts features 1006 from the sensory data. UE 2 may similarly receive sensory data 1008 into a feature extraction model 1010, which also extracts features 1012 from the sensory data. UE 1 and UE 2 may each perform back propagation to update the weights of the models 1004 and 1010 and improve feature extraction performance. After extracting the features, UE 1 and UE 2 may send these output features to the ML service entity 1014 to serve as input to the beam blocking prediction model 1016 for line of sight interference. The ML service entity 1014 may aggregate the features 1006, 1012 from the UEs, for example, by a summation or concatenation process, and pass these aggregated features forward through the beam blocking prediction model 1016. Therefore, the beam blocking prediction model can predict whether beam blocking occurs based on the aggregated sensing data from the UE. Similarly, the ML service entity 1014 can perform back propagation to update the weights of the model 1004 and the model 1010 for feature extraction and improve the beam blocking prediction performance. Fig.10 As illustrated in , the training or extraction performed at the UE and the training or inference performed at the ML service entity 1014 are separated or decoupled because there is no exchange of training / inference parameters (such as gradients) between the devices.

[0133] Fig.11 is a flow diagram of an example communication flow 1100 between a UE 1102 and an ML service entity 1104 (e.g., typically within or part of an ML server or base station (gNB, etc.), as indicated in the figure), wherein the ML service entity may request adaptive sensing and feature extraction for prediction tasks (e.g., line of sight obstruction prediction) to better support separate ML training, inference, and / or performance optimization, such as Fig.10As illustrated in . Adaptive sensing and feature extraction may occur after performing an ML service discovery process and establishing an ML service session between the UE and the ML service entity. For example, at 1, the UE 1102 may transmit a registration request to the ML service entity 1104. At 2, the ML service entity 1104 may confirm the request and inquire about the sensor information and ML model information of the UE. Then at 3, the UE 1102 may provide its sensor and model information, and the ML service entity may confirm the sensor and model information at 4. After the discovery process, at 5, the ML service entity 1104 may provide an ML service request to the UE 1102. In response to the ML service request, at 6, the UE 1102 may transmit an ML session request to the ML service entity 1104, and the ML service entity may confirm the session request at 7. After the session establishment process, training, inference, or performance optimization of the ML model at the ML service entity may begin.

[0134] In one approach, at 8.1, the UE 1102 may provide training or inference data (extract features) to the ML model at the ML service entity 1104, as previously described. However, the UE may also include multiple ML models that each perform a feature extraction task (e.g., detecting OBB). Although multiple ML models may each perform the same task, they may have different tradeoffs between inference time and model accuracy. For example, for the task of OBB detection, the UE may include multiple models with their own computational speed, detection accuracy, and output format. Some of these models may be more complex and therefore achieve higher accuracy, but often require longer inference times, while other models may be less complex and therefore achieve lower accuracy, but require shorter extraction times.

[0135] For example, Fig.12 is a conceptual diagram illustrating an example 1200 of an ML service entity 840 at a UE including a plurality of feature extraction models 1202. For example, in response to a request from Fig.111206 , the UE may perform model selection 1204 between any one (or more) of the i models configured at the UE for extracting the OBB 1206 . Each of the models 1202 may have a different computation time and detection accuracy. For example, some of the models 1202 may be more complex than other models (e.g., have a longer computation timing, include different amounts of data collection, etc.), and some of these models 1202 may have different performance than other models (e.g., have lower accuracy, more false alarms or misdetections of OBBs, etc.). The different complexity and performance of these models 1202 may also be due to different characteristics of these models. For example, some models 1202 may be independent architecture frameworks, such as MLP, CNN, or RNN, while other models 1202 may be a combination of the aforementioned architecture frameworks. In addition, the models 1202 may have different numbers of layers, kernel sizes, activation functions used in different layers, weights of different layers, etc.

[0136] Return to reference Fig.11 , the ML service entity 1104 may request the UE 1102 to perform the training or inference according to the training or inference needs of the ML service entity 1104. Fig.12 , which may depend on the state of the UE 1102. The state of the UE 1102 may include, for example, the mobility state of the UE (e.g., the speed of the UE, the moving direction of the UE, or the position of the UE relative to other UEs, pedestrians, or buildings, etc.), the number of UEs in the UE area or location, the computing or data processing capabilities of the UE, the amount of uplink traffic sharing the bandwidth of the UE or the uplink traffic load of the network including the UE, etc. For example, if the UE 1102 moves slowly, is in an environment with a small number of dynamic objects (e.g., moving vehicles or pedestrians), etc., the training or inference needs may be different from the case where the UE 1102 moves quickly, is in an environment with a large number of dynamic objects, etc. In the latter case, the ML service entity 1104 may determine that it is necessary to receive the extracted features from the UE 1102 at a faster rate than in the previous case, although the accuracy needs to be traded off in order to predict the beam blocking in the fast and dense environment of the UE 1102 more quickly. On the other hand, in the former case, if the ML service entity 1104 determines that the beam blocking prediction it currently makes is not accurate enough, the ML service entity 1104 may determine that it needs to receive extracted features from the UE 1102 more accurately than in the latter case, although a trade-off in speed is required in order to more accurately predict the beam blocking of the UE in a slow and sparsely populated environment.

[0137] Thus, in response to receiving training or extraction features (e.g., OBB 1206) from UE 1102 at 8.1, at 8.2, ML service entity 1104 may determine and request UE 1102 to use a slower but more accurate model (e.g., EfficientNet) or a faster but less accurate model (e.g., SSD MobileNet). At 8.3, ML service entity 1104 may transmit a request to that effect. For example, if during the discovery or session establishment process (described above) Fig.11 If the ML service entity 1104 is notified of the various model parameters of the model 1202 of the UE during steps 1-7 in the request, the ML service entity 1104 may request the UE 1102 to switch to the model explicitly indicated in the request based on the UE state, as previously described. Alternatively, the ML service entity 1104 may request the UE 1102 to switch to a different model based on the UE's own determination based on its UE state, similar to how the ML service entity 1104 makes a determination as previously described.

[0138] In response to the request, at 8.4, the ML service entity 840 ( Figure 8 1206. The UE 1102 may then select a different model from its model 1202 that outputs the OBB 1206. Thereafter, the UE 1102 may provide an acknowledgement of the request to the ML service entity 1104 at 8.5 and send new sensing data and extracted features (e.g., OBB 1206 from the new model) after the adaptation at 8.6. Thus, the ML service entity 1104 may input this new data into its beam blocking prediction model to make a more accurate or faster beam blocking prediction (depending on the selected UE model). The session may continue until the ML service entity 1104 finally transmits a session termination notification at 9 and the UE confirms the session termination at 10.

[0139] Fig.13 1300 is a conceptual diagram illustrating another method for adaptive sensing and feature extraction, in which an ML service entity 1302 at an ML server or gNB includes multiple ML models. In this method, an ML service entity 1304 of a UE may provide training or extracted features to the ML model at the ML service entity 1302, as previously described in Fig.11as described at 8.1. However, the ML service entity 1302 here may include multiple ML models that perform different functions. For example, one of the models may be a beamforming model 1306 that predicts K best Tx beams for beam scanning, and another of the models may be a beam tracking or beam refinement model 1308 that predicts K best refined Tx beams from a Tx / Rx beam pair including the K best Tx beams. The ML service entity 1302 may apply these models to perform beam management more efficiently. For example, based on the sensing information provided by the UE and the state of the UE, the beamforming model 1306 may predict a certain angular area in which many UEs may be positioned, and the base station may accordingly perform beam scanning more frequently for beam training in the area.

[0140] Although Fig.13 The example describes that the models 1306, 1308 of the ML service entity are related to beam management, but it should be understood that in other examples, the models of the ML service entity may not be limited to beam management. For example, one or more of the models 1306, 1308 of the ML service entity may respond to sensor data 1310 or extract features 1312 from the UE to predict LOS beam blocking between the base station and the UE, predict the best beam for communicating with the UE, perform beam management (beamforming training or refinement), optimally schedule resources to the UE, balance current traffic load, etc. In addition, any one of the ML service entity models may have a relationship or association with any other one of the ML service entity models, or be triggered in response to data from any other one of the ML service entity models. For example, as Fig.13 As illustrated in the example of , the beam tracking model can be triggered to predict the best refined beam in response to the beam pair predicted by the beamforming model.

[0141] In addition, the sensor data 1310 or extracted features 1312 used as training or inference data 1314 for one ML model of the ML service entity 1302 may be different from the sensor data 1310 or extracted features 1312 used as training or inference data 1316 for another ML model of the ML service entity 1302. For example, if the ML service entity 1302 is currently applying the beamforming model 1306 to make beamforming predictions, the ML service entity 1302 may request the UE to switch to the beamforming model 1306 it is currently applying. Fig.121302 and provide training or inference data 1314. The training or inference data 1314 may be, for example, an OBB indicating potential UEs in an area within the field of view of the UE's sensor. Alternatively, if the ML service entity 1302 is currently applying the beam refinement model 1308 to make beam refinement predictions, the ML service entity 1302 may request the UE to switch to its beam refinement model 1308. Fig.12 1314 and provide different training or inference data 1316 accordingly. The training or inference data 1316 may be, for example, an OBB indicating a potential UE having a finer sensor resolution than the sensor resolution of the training or inference data 1314, or even the same OBB as in the training or inference data 1314 but at a faster rate. Therefore, it should be understood that the model 1202 at the UE may be configured not only to output the same extracted features with different computation speeds or detection accuracies as previously described, but may alternatively be configured to output different extracted features for input to a different ML model of the ML service entity 1302. For example, the ML service entity 1302 may instruct the UE not only to switch to a different model in its model 1202, but also to reconfigure one or more of its sensors (e.g., field of view, etc.) to provide different sensor data to the new model 1202, thereby extracting different OBBs accordingly.

[0142] In addition, the ML service entity 1302 may request the UE to switch to its ML model based on the performance of the ML service entity 1302. Fig.12 For example, if the ML service entity 1302 determines that the performance (prediction accuracy) of the beamforming model 1306 or the beam refinement model 1308 is unacceptably low or below a certain standard based on beam measurements collected by the base station or reported by the end user, the ML service entity 1302 may retrain its model or switch to a different model. Alternatively or additionally, the ML service entity 1302 may request the UE to switch to a different model in its model 1202 based on the UE state to provide a slower or more accurate OBB. For example, if the UE is moving slowly or is in an environment with a small number of dynamic objects, the ML service entity may request the UE to switch to a slower but more accurate model, thereby providing the ML service entity 1302 with slower but more accurate extracted features. In turn, the ML service entity 1302 may input this more accurate sensing information in its model and more accurately predict beam blocking, optimal beam, etc., thereby obtaining better model performance.

[0143] Return to reference Fig.11At 8.2, the ML service entity 1104 may determine and request the UE 1102 to provide different sensor data or extracted features (from different ML models of the UE) for different ML models of the ML service entity 1104. For example, when the ML service entity 1104 uses the ML beamforming model to perform beamforming based on the training or inference data provided by one model of the UE 1102, the ML service entity 1104 may determine at 8.2 to use the ML beam refinement model to perform beam refinement. Therefore, at 8.3, the ML service entity 1104 may request the UE 1102 to switch to a different model to provide different data to the ML service entity 1104, thereby serving as input to the ML beam refinement model. In another example, when the ML service entity 1104 performs beamforming using an ML beamforming model based on training or inference data provided by one model of the UE 1102, or performs beam refinement using an ML beam refinement model based on different training or inference data provided by another model of the UE 1102, the ML service entity 1104 may determine at 8.2 that the performance of the beamforming or beam refinement model is insufficient. Therefore, at 8.3, the ML service entity 1104 may request the UE 1102 to switch to a different model to provide more accurate data to the ML service entity 1104, thereby improving the accuracy of its ML model and thus improving the performance of its ML model. In response to the request, at 8.4, the ML service entity 1104 at the UE 1102 may adjust its sensing and extraction models accordingly, provide a confirmation of the request at 8.5, and send new sensing data and extracted features after the adjustment at 8.6. Therefore, the ML service entity 1104 can input this new data into its beamforming or beam refinement model to make a more accurate prediction (depending on the selected UE model). Due to the higher accuracy, potential beam blockages that would otherwise be missed from an inaccurate prediction can be avoided.

[0144] In another method for adaptive sensing and feature extraction, the UE may be configured to communicate the confidence level of each feature extraction. For example, the UE 1102 similarly provides training or inference data to the ML model at the ML service entity 1104, as previously described in Fig.11 8.1 or 8.6, but also includes a confidence level indicating the accuracy associated with the provided data. Alternatively (or additionally), at 8.2, the ML service entity 1104 may determine the confidence level based on the data provided by the UE 1102, from a message including the data itself, or from the ML service entity's own interpretation of the confidence level.

[0145] In one example, if the field of view of a UE's sensor is severely obstructed, the point cloud represented by the data may not be fully informative or misleading, and thus the UE or ML server / gNB may classify the data with a low confidence level. The UE or ML server / gNB may detect such an occlusion of the field of view of a vehicle sensor and thus determine a confidence level associated with the data from the point cloud included in the field of view based on previous moments when other data associated with that sensor was provided. Fig.11 In response to determining the low accuracy of the data accordingly, at 8.3, the ML service entity 1104 may request the UE 1102 to adapt its feature extraction to these occlusions. For example, at 8.4, the UE 1102 may reconfigure its sensor field of view to avoid occlusions, or the UE may switch to a more accurate model, resulting in a higher confidence level for the feature data.

[0146] In another example, the ML service entity 1104 may determine and request the UE 1102 to apply a feature extraction model that infers one or more confidence levels. Fig.11 , the ML service entity 1104 may determine at 8.2 to provide the UE 1102 with a switch at 8.3 Fig.12 The UE 1102 may request one of its multiple models 1202 that has the capability to infer its OBB 1206 or other extracted features with a confidence level. The confidence level may be, for example, a flag or bit indicating whether the OBB was accurately detected to bind the object. In another example, the model 1202 may infer a distribution curve for the dimensions of the OBB, and the confidence level may indicate the accuracy of the inferred dimensions of the OBB based on the distribution curve. In another example, the model 1202 may infer a bounding box along with its dimensions and orientation while allowing a larger error margin along the dimensions and directions that carry the largest uncertainty measure, and the confidence level may indicate the error margin for these dimensions or orientations. In response to receiving the request, at 8.4, the UE 1102 may determine to switch to one of its ML models that has the requested confidence level capability, and adjust its sensing and feature extraction accordingly, as previously described.

[0147] Fig.14 is a conceptual diagram of an example 1400 in which training or feature extraction at a UE may be combined or associated with training or inference of a prediction model at an ML serving entity of an ML server or base station. Fig.10In the example of , UE 1 receives sensory data 1402 into a local feature extraction model 1404, which extracts features 1406 from the sensory data. UE 2 may similarly receive sensory data 1408 into a local feature extraction model 1410, which also extracts features 1412 from the sensory data. UE 1 and UE 2 may also each perform back propagation to update the weights of models 1404, 1410 and improve feature extraction performance. After extracting features 1406, 1412 in the forward pass, UE 1 and UE 2 may send these output features to an ML service entity 1414 to serve as input to a beam blocking prediction model 1416 for line of sight interference. The ML service entity 1414 may aggregate features from the UEs, for example, by a summation or cascade process, and pass these aggregated features forward through the beam blocking prediction model. Therefore, the beam blocking prediction model 1416 may predict whether beam blocking occurs based on the aggregated sensory data from the UEs. Similarly, the ML serving entity 1414 may perform back-propagation to update the weights of the model and improve beam blocking prediction performance.

[0148] In addition, if Fig.14 As exemplified in and with Fig.10 Unlike the example of , the ML service entity 1414 can broadcast aggregated gradient information 1418 (or other aggregated back-propagation output from the beam blocking prediction model 1416) to the UEs. After receiving the aggregated performance characteristics from the ML service entity 1414, each UE can apply these characteristics to their own back-propagation in their local feature extraction models 1404, 1410 and adaptively adjust their sensing and feature extraction accordingly. For example, each UE can update the weights of its feature extraction model based not only on its local gradients but also on the aggregated gradients 1418 received from the ML server or gNB.

[0149] Fig.15 is a flow diagram of an example communication flow 1500 between a UE 1502 and an ML service entity 1504, where the ML service entity may request adaptive sensing and feature extraction for prediction tasks (e.g., line of sight obstruction prediction) to better support joint ML training, inference, and / or performance optimization, such as Fig.14 Similar to Fig.11, adaptive sensing and feature extraction may occur after performing an ML service discovery process and establishing an ML service session between the UE 1502 and the ML service entity 1504. For example, at 1, the UE 1502 may transmit a registration request to the ML service entity 1504. At 2, the ML service entity 1504 may confirm the request and inquire about the sensor information and ML model information of the UE. Then at 3, the UE 1502 may provide its sensor and model information, and the ML service entity 1504 may confirm the sensor and model information at 4. After the discovery process, at 5, the ML service entity 1504 may provide an ML service request to the UE 1502. In response to the ML service request, at 6, the UE 1502 may transmit an ML session request to the ML service entity 1504, and the ML service entity may confirm the session request at 7. After the session establishment process, the training, inference, or performance optimization of the ML model at the ML service entity 1504 may begin.

[0150] Initially, at 8.1, UE 1502 may train its local feature extraction model, and at 8.2, UE 1502 may provide sensor data or inferred data to the beam blocking prediction model at ML service entity 1504, as previously described in Fig.14 . Then, at 8.3, the ML service entity 1504 aggregates the data or features received from the UE 1502 with the data or features received from other UEs, and inputs the aggregated data into its beam blocking prediction model for forward pass. The ML service entity 1504 may also calculate gradients and back-propagate these features through the beam blocking prediction model to update the model weights and improve the beam blocking performance. Then, at 8.4, the ML service entity 1504 broadcasts these aggregated features or back-propagation outputs to the UE 1502. Upon receiving this information, at 8.5, the UE 1502 may apply these aggregated gradients to its own back-propagation through its local feature extraction model, and adaptively adjust the model accordingly. Thereafter, at 8.6, the UE 1502 may provide the adjusted sensor data or inferred data from the adjustment model to the ML service entity 1504 accordingly. Therefore, the ML service entity 1504 may input this new data into its beam blocking prediction model to continue beam blocking prediction. The session may continue until the ML service entity 1504 eventually transmits a session termination notification at 9 and the UE 1502 confirms the session termination at 10 .

[0151] In addition to exchanging control messages for adaptive sensing and feature extraction with the UE, the ML server or ML service entity of the base station may exchange adaptive sensor configuration or reconfiguration messages. These messages may include parameters for configuring or reconfiguring the sensors of the UE. Figure 8, these messages may be exchanged between the control or management function of the ML service entity 840 of the UE and the control or management function at the ML service entity 830 of the gNB or an ML server (e.g., a cloud or edge server) co-located with or near the gNB. In addition, messages may be exchanged during one or more processes such as ML service discovery and ML session establishment, model training, model inference, and / or model performance optimization.

[0152] In one approach, the gNB / ML server may initiate or trigger adaptive sensor configuration or reconfiguration by transmitting a configuration or reconfiguration message to the UE that includes parameters configured based on one or more factors. These factors may be related to ML service training requirements, ML service inference requirements, ML service performance requirements, network traffic load, and the number of UEs in the area. In one example, with respect to ML service training, the UE may have an unobstructed or obstructed field of view. If the UE has an unobstructed field of view, the gNB / ML server may configure the UE to combine all RADAR point clouds from its various sensors and send a joint feature map to the gNB / ML server. If the UE has an obstructed field of view, such as a front RADAR from which it can supplement the left window radar of a neighboring vehicle UE, the gNB / ML server may configure the UE to return a feature map from the front RADAR point cloud alone or perform some other supplementary configuration. In another example, with respect to ML service inference, the UE may have an occlusion in the previously declared field of view. For example, if a UE that previously had a nearly unobstructed RADAR has an occlusion (or near occlusion) of the RADAR, the gNB / ML server may instruct the UE not to use the obstructed RADAR and separately instruct neighboring vehicle UEs to start using their unobstructed RADAR. In another example, with respect to ML performance requirements, performance degradation of the beam obstruction prediction model may occur due to poor resolution of the data or high mobility of the UE. Therefore, the gNB / ML server may configure the UE sensors to adjust accordingly, thereby increasing sensor resolution or accounting for UE mobility. In another example, if the gNB / ML server determines that the network traffic load is high, the gNB / ML server may configure the UE to reduce its RADAR measurement update rate, or reduce the resolution and frame rate of its camera. In another example, if the number of UEs in the UE area is high, the gNB / ML server may configure the UE to similarly reduce its RADAR measurement update rate.

[0153] In another method, the UE may initiate or trigger adaptive sensor configuration or reconfiguration by transmitting a message to the gNB / ML server including a request or notification of a configuration update or reconfiguration of preferred or updated parameters based on one or more factors. These factors may be related to vehicle sensor setup and configuration, sensor availability, sensor selection, location change, speed change, direction change, radio link quality with the gNB / ML server, UE vehicle advanced driver assistance system (ADAS), or UE sensor occlusion. In one example related to vehicle sensor setup and configuration, the UE may determine to change the FoV, range, or measurement update rate of its sensor in order to obtain information about the area of ​​interest. In another example related to sensor availability, after determining that a sensor such as a rear RADAR is not actively used for ADAS tasks, the UE may determine to reconfigure the sensor to meet the sensing needs of the ML server / gNB. In another example related to sensor selection, the UE may determine to select one or more specific sensors based on the sensing needs of the ML server or gNB, such as only its front RADAR or only its front-mounted camera. In another example related to a change in position, the UE may determine to change its FoV and range in response to determining that the UE is on a highway or at an intersection. In another example related to a change in speed, the UE may determine to change its measurement update rate or change its FoV in response to determining an acceleration or deceleration. In another example related to a change in direction, the UE may determine to use a different set of sensors for beam blocking prediction in response to determining a change in its direction. In another example related to the quality of the radio link to the gNB / ML server, the UE may determine to change its data rate or latency associated with transmitting its data to the gNB / ML server, for example, by reducing the RADAR measurement update rate or the resolution or frame rate of the camera when the connection is poor, and increasing the RADAR measurement update rate or the resolution or frame rate of the camera when the connection is strong. In another example related to vehicular ADAS of a UE, a UE may determine to reconfigure the FoV of, for example, a RADAR that was previously configured to meet the training, inference, or ML performance optimization needs of a gNB / ML server by pre-overriding the previous setting of the FoV and transmitting a reconfiguration update message including the reconfigured parameters for its RADAR to the gNB / ML server to meet the ADAS needs of a vehicle. In another example related to UE sensor occlusion, a UE may determine to adjust its sensing due to an occlusion in a previously declared FoV, for example, if a UE that previously had a nearly unobstructed RADAR has an occlusion (or near occlusion) of the RADAR, the UE may pre-reconfigure the FoV or switch to another RADAR with an unobstructed field of view, and transmit a reconfiguration update message including the reconfigured parameters to the gNB / ML server accordingly.

[0154] Fig.16 16 is a flow chart of an example communication flow 1600 between a UE 1602 and an ML service entity 1604, wherein the ML service entity or the UE may trigger sensor configuration or reconfiguration for a prediction task (e.g., line of sight obstruction prediction) to better support ML training, inference, and / or performance optimization. For example, the UE may include a configurable sensor, such as a RADAR sensor or a camera, which includes reconfigurable parameters including, but not limited to, FoV, orientation range, resolution or image resolution, update rate or frame rate, etc. The ML service entity 1604 at the ML server or gNB may configure the sensor configuration or reconfiguration for a prediction task (e.g., line of sight obstruction prediction) to better support ML training, inference, and / or performance optimization. Fig.16 The configuration or reconfiguration of these parameters of the UE's sensors is managed by the process illustrated in FIG.

[0155] Initially, sensor configuration may occur during the ML service discovery or ML session establishment phase. During ML service discovery, initially at 0, the ML server may transmit an ML service announcement to the UE 1602, or the gNB may transmit system information with ML learning capabilities to the UE 1602. At 1, the UE 1602 may transmit an ML service registration request to the ML server or a registration request to the gNB. At 2, the ML server may transmit an ML subscription request to the UE 1602, or the gNB may transmit an ML capability survey to the UE 1602. Then at 3, the UE 1602 may transmit an ML subscription response to the ML server or a UE ML capability information message to the gNB, which indicates its ML subscription or capability information (e.g., its onboard sensor configuration, supported ML models, reconfigurable parameters, etc.). For example, the UE 1602 may indicate a list of RADAR sensor / detector reconfigurable parameters, such as FoV, orientation, range, resolution, update rate, etc., as well as other RADAR sensor parameters including, but not limited to, sensor identification (e.g., the number of RADAR sensors and associated IDs), sensor mounting on the vehicle (e.g., location and mounting rotation angles [roll, pitch, yaw] relative to the center of the autonomous vehicle), detector configuration (e.g., angular field of view, range limits (minimum and maximum detection range), range rate limits (minimum and maximum range rate), detection probability, false alarm rate, range resolution, angular center band frequency, etc.), and measurement resolution and bias (e.g., azimuth, elevation, range, range rate resolution, etc.). Similarly, the UE 1602 may indicate a list of camera sensor / detector reconfigurable parameters such as FoV, image resolution, frame rate, etc., and other camera sensor parameters including, but not limited to, sensor identification (e.g., number of camera sensors and associated IDs), sensor mounting within the vehicle (e.g., location and mounting rotation angles [roll, pitch, yaw] relative to the center of the autonomous vehicle), detector configuration (e.g., camera image size, camera focal length, optical center, radial and tangential distortion coefficients, etc.). Thereafter, at 4, the ML server may transmit an ML service registration complete message, or the gNB may transmit a registration complete message.

[0156] After completing the session discovery and indicating the reconfigurable sensor parameters, at 5, the UE 1602 and the ML server / gNB may establish a session between the devices for training, inference, or performance optimization. Thereafter, the gNB / ML server or the UE may trigger adaptive sensor reconfiguration. In one example where the gNB / ML server triggers adaptive sensor reconfiguration during the training, inference, or performance optimization process, at 6, the gNB / ML server determines the reconfiguration parameters based on the criteria or factors previously described, and then at 7, the gNB / ML server transmits a sensor reconfiguration request to the UE indicating that the sensor parameters are to be reconfigured. These factors may be related to, for example, ML service training requirements, ML service inference requirements, ML service performance requirements, network traffic load, or the number of UEs in the area. In response to the request, at 8, the UE 1602 may reconfigure its onboard sensors accordingly, and at 9, the UE may provide a confirmation or completion message to the ML server or gNB.

[0157] In another example where the UE 1602 triggers adaptive sensor reconfiguration, at 10, the UE may determine reconfiguration parameters for its sensors based on the various criteria or factors previously described, and at 11, the UE may transmit a sensor reconfiguration request to the gNB / ML server indicating that the sensor parameters are to be reconfigured. These factors may be related, for example, to vehicle sensor setup and configuration, sensor availability, sensor selection, location change, speed change, direction change, radio link quality with the gNB / ML server, UE vehicle advanced driver assistance system (ADAS), or UE sensor occlusion. If, at 12, the gNB / ML server responds with an indication that the reconfiguration is allowed, at 13, the UE reconfigures its onboard sensors accordingly, and at 14, the UE may provide a confirmation or completion message to the ML server or gNB. Alternatively, if, at 12, the gNB / ML server responds with an indication that the reconfiguration is rejected, the UE 1602 may avoid performing steps 13 and 14. Alternatively, if at 12 the gNB / ML server responds with an indication of a modified reconfiguration, at 13 the UE may reconfigure its onboard sensors based on the modification, and at 14 the UE may provide a confirmation or completion message to the ML server or gNB.

[0158] Fig.171700 is a flow chart of a method of wireless communication. The method may be performed by a UE (e.g., UE 104, 104a, 350, 704a, 704b, 704c, 704d, 1102, 1502, 1602; device 1902). Optional aspects are illustrated by dashed lines. The method allows the UE to switch between its ML-based feature data extraction models (and optionally reconfigure its sensors) in response to adaptive configuration from a base station to provide feature data of dynamic or moving potential LOS obstacles for improving beam blocking prediction performance at a network entity or node. The network entity or node may be, for example, a converged or decomposed base station, an ML server co-located with or near a converged or decomposed base station, a component of a decomposed base station (e.g., a near-RT RIC, a non-RT RIC, a CU, a DU or a RU, or other decomposed base station components), an ML service entity in a converged base station, an ML service entity in a component of a decomposed base station, an ML service entity in an ML server co-located with or near such a base station, or other network entities or nodes.

[0159] At 1702, the UE may receive a message instructing the UE to switch from a current ML-based feature data extraction model of the UE to one of a plurality of ML-based feature data extraction models based on the state of the UE. For example, 1702 may be performed by the configuration receiving component 1940. Each of the plurality of ML-based feature data extraction models provides ML-based feature data for predicting beam blocking between the UE and a network entity or node. For example, referring to Fig.11 and Fig.12, UE 1102 (e.g., Rx processor 356 of UE 350 or ML service of UE) may receive a request from a network node (e.g., ML service entity 1104 of a base station or ML server) at 8.3 for UE to switch from one of its models 1202 to another of its models 1202 for extracting OBB 1206 or other feature data from point cloud, image or other sensor data. The feature data may in turn be applied to predict beam blocking between UE and network node. The request may be received based on the state of UE. The state of UE may include at least one of the following: mobility state of the device, number of UEs in the area where UE is located, data processing capability of UE, amount of uplink traffic sharing bandwidth of UE, or uplink traffic load of network including UE. For example, the state of UE may be mobility state of UE (e.g., speed of UE, direction of movement of UE or position of UE relative to other UE, pedestrian or building, etc.), number of UEs in area or location of UE, computing or data processing capability of UE, amount of uplink traffic sharing bandwidth of UE, or uplink traffic load of network including UE, etc. For example, if the UE initially applied one of its feature data extraction models at a particular UE speed, location, network traffic load, etc., but later the UE increased its speed, changed its location, experienced an increase in uplink traffic, or some other UE state change occurred, then the request may instruct the UE to switch from that model to a different model that may output OBB or other feature data at a faster rate (although with a trade-off in accuracy) so that the ML service entity 1104 may predict beam blocking more quickly.

[0160] In one example, the message at 1702 may indicate the performance of at least one ML model in a plurality of ML models for beam management. Figures 11 to 13 , if the ML service entity 1104 determines that the performance (prediction accuracy) of the beamforming model 1306 or the beam refinement model 1308 is unacceptably low or below a certain standard based on beam measurements collected by the base station or reported by the end user, the ML service entity can request (in the configuration / request at 8.3) that the UE switch to a different model in its model 1202 to provide a slower or more accurate OBB 1206. In turn, the ML service entity 1104 can input this more accurate sensing information in its model and more accurately predict beam blockage, best beam, etc., thereby obtaining better model performance.

[0161] In another example, the message also includes an instruction for the UE to send different ML-based feature data for different ML models for beam management. Figures 11 to 13If the ML service entity 1104 is currently applying the beamforming model 1306 to make beamforming predictions, the ML service entity may request (in the message at 8.3) that the UE 1102 switch to its Fig.12 13.3. The ML service entity 1104 may provide one of the models 1202 in the ML service entity 1104 and provide training or inference data 1314. The training or inference data 1314 may be, for example, an OBB indicating potential UEs in the area within the field of view of the UE's sensor. Alternatively, if the ML service entity 1104 is currently applying the beam refinement model 1308 to make beam refinement predictions, the ML service entity may request (in the message at 8.3) that the UE 1102 switch to its beam refinement model 1308 in the ML service entity 1104. Fig.12 1202 and accordingly provide different training or inference data 1316. The training or inference data 1316 may be, for example, an OBB indicating a potential UE having a finer sensor resolution than the sensor resolution of the training or inference data 1314, or even the same OBB as in the training or inference data 1314 but at a faster rate.

[0162] At 1704, the UE may determine to switch from the current ML-based feature data extraction model to one of the ML-based feature data extraction models in response to the message at 1702. For example, 1704 may be performed by the model switching determination component 1942. For example, referring to Fig.11 and Fig.12 At 8.4, UE 1102 (e.g., controller / processor 359 of UE 350 or ML service at UE) may determine to adjust its sensing and extraction model based on the message received at 8.3. For example, if ML service entity 1104 requests the UE to switch to a model explicitly indicated in the request based on the UE state at 8.3, as previously described, UE 1102 may perform model selection 1204 to select a different model (e.g., model 1202) of the OBB 1206 whose output is as explicitly indicated in the request. Fig.12 1-i in ). Alternatively, if the ML service entity 1104 requests the UE to switch to a different model according to the UE's own determination at 8.3 (without an explicit indication of the model), the UE may determine which model to select based on the UE state. For example, if the request at 8.3 indicates that the ML service entity requests the UE to provide extracted features at a faster rate, the UE may determine to select a less complex model in its model 1202 during model selection 1204, and if the ML service entity requests the UE to provide more accurate extracted features, the UE may determine to select a more complex model in its model 1202 during model selection 1204.

[0163] In one example, the ML-based feature data extraction models may include different computation speeds and different detection accuracies. For example, some of the models 1202 may be more complex than other models (e.g., having longer computation timing, including different amounts of data collection, etc.), and some of these models 1202 may have different performance than other models (e.g., having lower accuracy, more false alarms or misdetection of OBB, etc.). The different complexity and performance of these models 1202 may also be due to the different characteristics of these models. For example, some models 1202 may be independent architecture frameworks, such as MLP, CNN, or RNN, while other models 1202 may be a combination of the aforementioned architecture frameworks. In addition, the models 1202 may have different numbers of layers, kernel sizes, activation functions used in different layers, weights of different layers, etc.

[0164] In one example, the determination at 1704 to switch to a second ML-based feature data extraction model is independent of the aggregate performance characteristics of the ML model for beam blocking prediction of the network node. In this example, the ML model for beam blocking prediction may include an aggregation of ML-based feature data input from multiple UEs including the UE. Alternatively, in another example, at 1706, the UE may receive the aggregate performance characteristics of the ML model for beam blocking prediction. For example, 1706 may be performed by the performance characteristic receiving component 1944. In this alternative example, the ML model for beam blocking prediction may similarly include an aggregation of ML-based feature data input from multiple UEs including the UE. However, in this example, the determination at 1704 to switch to the second ML-based feature data extraction model in the ML-based feature data extraction model may be based on the aggregate performance characteristics.

[0165] For example, Fig.10 and Fig.14 The ML service entities 1014, 1414 are illustrated to include beam blocking prediction models 1016, 1416 (see also Figure 8 and Fig. 9 826, 939 or 997), the beam blocking prediction model receives aggregated extracted features 1006, 1012, 1406, 1412 from multiple UEs (UE 1 and UE 2 in this example), predicts whether beam blocking occurs based on the aggregated UE sensing information, and performs back propagation including identifying aggregate performance characteristics (such as gradient 1418) to update the model weights and improve beam blocking prediction performance. Fig.10 In the example of , the training or inference performed at the UE and the ML service entity can be separated or decoupled because there is no exchange of gradients 1418 between the devices. Therefore, in this separate environment, reference Figure 11 to Figure 12, UE 1102 (e.g., controller / processor 359 of UE 350, or ML service at the UE) may determine to switch to one of its models 1202 regardless of gradient 1418 or other aggregate performance characteristics of the beam blocking prediction model. For example, if the UE determines to switch during model selection 1204 (when Fig.11 8.4) to a high accuracy model in model 1202, the high accuracy may be generated by the updated model weights based on the local gradient of the feature extraction model obtained by the UE rather than the aggregate gradient 1418 of the beam blocking prediction model obtained by the ML service entity. Therefore, the UE may determine to switch to this model because it has a higher accuracy and is independent of the aggregate gradient.

[0166] In contrast, Fig.14 In the example of , the ML service entity 1414 can broadcast aggregated gradient information 1418 (or other aggregated back-propagation output from the beam blocking prediction model 1416) to the UE. Upon receiving these aggregated performance characteristics (e.g., by the RX processor 356 of the UE 350 or by the UE's ML service), the UE can update the weights of its feature extraction model based not only on its local gradients but also on the aggregated gradients 1418 received from the ML server or gNB. Thus, with reference to Fig.12 , Fig.14 and Fig.15 , if UE 1502 determines to switch during model selection 1204 (when Fig.15 8.5 of ) to a high accuracy model in model 1202, the high accuracy here can be generated not only by the updated model weights based on the local gradient of the feature extraction model obtained by the UE, but also by the updated model weights based on the aggregated gradient 1418 of the beam blocking prediction model obtained by the ML service entity. Fig.15 After receiving the aggregate gradient at 8.4, the UE (eg, the controller / processor 359 of the UE 350, or an ML service at the UE) may determine to switch to that model due to its higher accuracy resulting from the aggregate gradient.

[0167] At 1708, the UE may send ML-based feature data for predicting beam blocking based on one of the ML-based feature data extraction models in response to the switch. For example, 1708 may be performed by feature data sending component 1946. For example, referring to Figures 9 to 16, the UE 104, 104a, 1102, 1502, 1602 (e.g., the TX processor 368 of the UE 350 or the ML service of the UE) can send training data 902 to the ML server or ML service entity at the base station for training at the non-RT RIC, send inference data 904 to the ML service entity for inference at the near-RT RIC (e.g., training or inference data 1314, 1316), send OBB 1206 or other features 1006, 1012, 1406, 1412 to the ML service entity 1014, 1104, 1302, 1414, 1504, 1604 or to the base station or ML server interfaced with the ML service entity. For example, Fig.11 or Fig.15 The ML-based feature data sent at 8.6 of the embodiment may be based on the model 1202 to which the UE determines to switch at 1704. For example, Fig.11 or Fig.15 The OBB 1206 sent at 8.6 of 1204 may be output from the new model after the model switching. In addition, the ML-based feature data may be used for beam blocking prediction, or the beam blocking prediction may be based on the new model to which the UE switches. For example, referring to Figure 7 ,exist Fig.11 or Fig.16 The OBB 1206 sent from UE 704a at 8.6 of FIG. 8 may identify potential UEs or other objects (e.g., truck 711) that the ML service entity 749 may predict to cause LOS blocking of a beam (e.g., beam 707). Similarly, model switching may result in faster or more accurate OBB 1206 from UE 704a, which in turn may allow the ML service entity 749 to better identify potential causes of LOS beam blocking in a dynamic or mobile environment of the UE.

[0168] In one example, at 1710, the UE can send a confidence level associated with the ML-based feature data. For example, 1710 can be performed by a confidence level sending component 1948. For example, referring to Fig.11 , when UE 1102 (e.g., TX processor 368 of UE 350 or ML service of UE) provides training or inference data to ML service entity 1104 (as previously described at 8.1 or 8.6), the UE may also include a confidence level indicating the accuracy associated with the provided data. For example, referring to Fig.12, the confidence level may be, for example, a flag or bit indicating whether a corresponding one of the OBBs 1206 output from the model 1202 is accurately detected to bind the object. In another example, the model 1202 may infer a distribution curve of the dimensions of the OBB, and the confidence level may indicate the accuracy of the inferred dimensions of the OBB based on the distribution curve. In another example, the model 1202 may infer a bounding box along with its size and orientation while allowing a larger error margin along the dimensions and directions carrying the largest uncertainty measure, and the confidence level may indicate the error margin for these dimensions or directions. In another example, the confidence level may reflect a combination of any one or more of the accuracy indicators described above.

[0169] In one example, at 1704, it is determined that switching to one of the ML-based feature data extraction models can derive a confidence level based on a second one of the ML-based feature data extraction models. Fig.11 , the ML service entity 1104 may request the UE 1102 to switch to Fig.12 The UE 1102 may request a request for one of its multiple models 1202 that has the ability to infer a confidence level for its OBB 1206 or other extracted features. The confidence level may reflect a combination of any one or more of the accuracy metrics mentioned above at 1710. In response to receiving the request, at 8.4, the UE 1102 may determine to switch to one of its ML models 1202 that has the requested confidence level capability, and adjust its sensing and feature extraction accordingly, as previously described. For example, the UE may switch to an ML model explicitly indicated in the request because it has the ability to infer the confidence level, or if no explicit indication is given, the UE may switch to any ML model it deems appropriate (e.g., one that produces a faster OBB or a more accurate OBB depending on the UE state), as long as that model is also capable of inferring a confidence level for the OBB.

[0170] In one example, the message at 1702 may also include an instruction for the UE to reconfigure a sensor of the UE, and the ML-based feature data may be further based on the sensor. Fig.16, the UE 1602 may include a configurable sensor, such as a RADAR sensor, a LIDAR sensor, or a camera, which includes reconfigurable parameters such as, but not limited to, FoV, orientation range, resolution or image resolution, update rate or frame rate, etc., and the ML service entity 1604 at the ML server or base station may manage the configuration or reconfiguration of these parameters of the UE's sensor. For example, the UE 1602 may receive a sensor reconfiguration configuration / request at 7 (which may indicate the UE sensor parameters that the gNB / ML server requests the UE to reconfigure) or receive a sensor reconfiguration response at 12 (which may confirm the UE sensor parameters that the UE previously requested to reconfigure (at 11)) from the ML service entity 1604. Fig.16 A sensor reconfiguration request at 7 or Fig.16 The sensor reconfiguration response at 12 may be combined in the Fig.11 Alternatively, the request may be received in a different configuration message. Fig.12 As a result of sensor reconfiguration, the sensor data input into the model 1202 may change, thereby producing a different OBB 1206 .

[0171] In a variation of the example related to sensor reconfiguration, the message may be received in response to a criterion for sensor reconfiguration being met. Fig.16, if the ML service entity 1604 determines that a certain criterion triggering a sensor reconfiguration request has been met, the UE 1602 may receive a sensor reconfiguration request at 7. The criterion may be related to, for example, service training requirements, ML service inference requirements, ML service performance requirements, network traffic load, and the number of UEs in the area. In one example, with respect to ML service training, the UE may have an unobstructed or obstructed field of view. If the UE has an unobstructed field of view, the gNB / ML server may configure the UE to combine all RADAR point clouds from its various sensors and send a joint feature map to the gNB / ML server. If the UE has an obstructed field of view, such as a front RADAR from which it can supplement the left window radar of a neighboring vehicle UE, the gNB / ML server may configure the UE to return a feature map from the front RADAR point cloud alone or perform some other supplementary configuration. In another example, with respect to ML service inference, the UE may have an occlusion in the previously declared field of view. For example, if a UE that previously had a nearly unobstructed RADAR has an occlusion (or near occlusion) of the RADAR, the gNB / ML server may instruct the UE not to use the obstructed RADAR and separately instruct neighboring vehicle UEs to start using their unobstructed RADAR. In another example, with respect to ML performance requirements, performance degradation of the beam obstruction prediction model may occur due to poor resolution of the data or high mobility of the UE. Therefore, the gNB / ML server may configure the UE sensors to adjust accordingly, thereby increasing sensor resolution or accounting for UE mobility. In another example, if the gNB / ML server determines that the network traffic load is high, the gNB / ML server may configure the UE to reduce its RADAR measurement update rate, or reduce the resolution and frame rate of its camera. In another example, if the number of UEs in the UE area is high, the gNB / ML server may configure the UE to similarly reduce its RADAR measurement update rate.

[0172] In another variation of the example related to sensor reconfiguration, at 1712, the UE may determine that a criterion for sensor reconfiguration is met, and at 1714, the UE may reconfigure at least one of the FoV, range, measurement update rate, resolution, or frame rate of the sensor in response to meeting the criterion. For example, 1712 may be performed by the criterion determination component 1950, and 1714 may be performed by the sensor reconfiguration component 1952. For example, referring to Fig.16, the UE 1602 (e.g., the controller / processor 359 of the UE 350 or the ML service of the UE) may determine that a certain criterion has been met, which triggers the UE to transmit a sensor reconfiguration request to the ML server or ML service entity of the base station at 11. The criterion may be related to, for example, vehicle sensor setup and configuration, sensor availability, sensor selection, position change, speed change, direction change, radio link quality with the gNB / ML server, UE vehicle advanced driver assistance system (ADAS), or UE sensor occlusion. In one example related to vehicle sensor setup and configuration, the UE may determine to change the FoV, range, or measurement update rate of its sensors in order to obtain information about the area of ​​interest. In another example related to sensor availability, after determining that a sensor such as a rear RADAR is not actively used for ADAS tasks, the UE may determine to reconfigure the sensor to meet the sensing needs of the ML server / gNB. In another example related to sensor selection, the UE may determine to select one or more specific sensors, such as only its front RADAR or only its front-mounted camera, based on the sensing needs of the ML server or gNB. In another example related to a change in position, the UE may determine to change its FoV and range in response to determining that the UE is on a highway or at an intersection. In another example related to a change in speed, the UE may determine to change its measurement update rate or change its FoV in response to determining an acceleration or deceleration. In another example related to a change in direction, the UE may determine to use a different set of sensors for beam blocking prediction in response to determining a change in its direction. In another example related to the quality of the radio link to the gNB / ML server, the UE may determine to change its data rate or latency associated with transmitting its data to the gNB / ML server, for example, by reducing the RADAR measurement update rate or the resolution or frame rate of the camera when the connection is poor, and increasing the RADAR measurement update rate or the resolution or frame rate of the camera when the connection is strong. In another example related to vehicular ADAS of a UE, a UE may determine to reconfigure the FoV of, for example, a RADAR that was previously configured to meet the training, inference, or ML performance optimization needs of a gNB / ML server by pre-overriding the previous setting of the FoV and transmitting a reconfiguration update message including the reconfigured parameters for its RADAR to the gNB / ML server to meet the ADAS needs of a vehicle. In another example related to UE sensor occlusion, a UE may determine to adjust its sensing due to an occlusion in a previously declared FoV, for example, if a UE that previously had a nearly unobstructed RADAR has an occlusion (or near occlusion) of the RADAR, the UE may pre-reconfigure the FoV or switch to another RADAR with an unobstructed field of view, and transmit a reconfiguration update message including the reconfigured parameters to the gNB / ML server accordingly.In any one or more of these examples, after confirming the sensor reconfiguration request at 12, the UE (e.g., the controller / processor 359 of the UE 350 or the ML service of the UE) may reconfigure its sensors at 13. For example, the UE may update its RADAR, LIDAR, or camera parameters, such as previously described in the aforementioned examples.

[0173] Fig.18 1800 is a flow chart of a method of wireless communication. The method may be performed by a network node. The network node may be, for example, a converged or decomposed base station, an ML server co-located with or near a converged or decomposed base station (e.g., base station 102 / 180, 310; device 2002), a component of a decomposed base station (e.g., near-RT RIC 425, 525; non-RT RIC 415, 556; CU 410, 560, 561; DU 430, 562; RU 440, 564; or other decomposed base station components) or an ML service entity 749, 830, 1014, 1104, 1302, 1414, 1504, 1604 in a converged base station, a component of a converged base station, or an ML server co-located with or near a base station. The method allows a network node to adaptively configure a UE to switch between its ML-based feature data extraction models (and optionally reconfigure its sensors) in order to provide feature data of dynamic or moving potential LOS obstacles for improving beam blocking prediction performance at the network node.

[0174] At 1802, the network node may receive first ML-based feature data from the UE based on a first ML-based feature data extraction model of the UE. For example, 1802 may be performed by feature data receiving component 2040. For example, a base station, an ML server, a near-RT RIC, a non-RT RIC, a CU, a DU, a RU, or other network node may initially receive extracted features from the UE. For example, referring to Figures 9 to 16 ,exist Fig.11 8.1 or Fig.15 8.2, ML service entities 1014, 1104, 1302, 1414, 1504, 1604 (or Figure 3The Rx processor 370 of the network node or base station 310 of the UE 104, 104a, 1102, 1502, 1602 may receive the training data 902 for training at the non-RTRIC or the inference data 904 for inference at the near-RT RIC (e.g., the training or inference data 1314, 1316), the OBB 1206, or the features 1006, 1012, 1406, 1412. Alternatively, a base station or an ML server interfaced with the ML service entity, or a component of a decomposed base station interfaced with the ML service entity may receive the data. The data may be received by, for example, Fig.11 or Fig.15 The ML-based feature data received by the ML service entity at 8.6 may be based on one of the models 1202 for feature extraction available at the UE. Fig.11 8.1 or Fig.15 The OBB 1206 received at 8.2 of 1204 may be output from the model 1202. In addition, the ML-based feature data may be used for beam blocking prediction, or the beam blocking prediction may be based on the UE's model. For example, referring to Figure 7 ,exist Fig.11 or Fig.15 The OBB 1206 received from the UE 704a at 8.1 or 8.2 of FIG. 1204 may identify potential UEs or other objects (eg, the truck 711) that the ML service entity 749 may predict may cause LOS blocking of a beam (eg, the beam 707).

[0175] At 1804, the network node sends a message instructing the UE to switch from the first ML-based feature data extraction model of the UE to the second ML-based feature data extraction model based on the state of the UE. For example, 1804 can be performed by the configuration sending component 2042. For example, referring to Fig.11 and Fig.12 , the ML service entity 1104 of the base station or ML server, the base station or ML server itself (e.g., Figure 3A network node or TX processor 316 of a base station 310 of a network node or ... For example, if the UE initially applied one of its feature data extraction models at a particular UE speed, location, network traffic load, etc., but later the UE increased its speed, changed its location, experienced an increase in uplink traffic, or some other UE state change occurred, then the request may instruct the UE to switch from that model to a different model that may output OBB or other feature data at a faster rate (although with a trade-off in accuracy) so that the ML service entity 1104 may predict beam blocking more quickly.

[0176] In one example, the network node may further include a plurality of ML models for beam management, and the message may be based on the performance of at least one of the ML models for beam management. Figures 11 to 13 , the ML service entity 1104, 1302 may include a beamforming model 1306 and a beam tracking model 1308. If the ML service entity 1104, 1302 determines that the performance (prediction accuracy) of the beamforming model 1306 or the beam refinement model 1308 is unacceptably low or below a certain standard based on beam measurements collected by the base station or reported by the end user, the ML service entity, the base station, the ML server, a component of the base station, or other network node may request (in a message / request at 8.3) that the UE switch to a different model in its model 1202 to provide the ML service entity with a slower or more accurate OBB 1206. In turn, the ML service entity 1104 may input this more accurate sensing information in its model and more accurately predict beam blocking, optimal beam, etc., thereby obtaining better model performance.

[0177] In another example, the message may also include an instruction for the UE to send different ML-based feature data of different ML models for beam management of the network node. Figures 11 to 13, if the ML service entity 1104 is currently applying the beamforming model 1306 to make beamforming predictions, the ML service entity, the base station, the ML server, a component of the base station, or other network node may request (in the configuration at 8.3) that the UE 1102 switch to the beamforming model 1306 it is currently applying. Fig.12 1202 in the model 1204 and provide training or inference data 1314. The training or inference data 1314 may be, for example, an OBB indicating potential UEs in the area within the field of view of the UE's sensor. Alternatively, if the ML service entity 1104 is currently applying the beam refinement model 1308 to make beam refinement predictions, the ML service entity, the base station, the ML server, a component of the base station, or other network node may request (in the configuration at 8.3) that the UE 1102 switch to its Fig.12 1202 and accordingly provide different training or inference data 1316. The training or inference data 1316 may be, for example, an OBB indicating a potential UE having a finer sensor resolution than the sensor resolution of the training or inference data 1314, or even the same OBB as in the training or inference data 1314 but at a faster rate.

[0178] In one example, the first ML-based feature data extraction model and the second ML-based feature data extraction model may include different computation speeds and different detection accuracies. For example, some of the models 1202 may be more complex than other models (e.g., having longer computation timing, including different amounts of data collection, etc.), and some of these models 1202 may have different performance than other models (e.g., having lower accuracy, more false alarms or misdetection of OBB, etc.). The different complexity and performance of these models 1202 may also be due to the different characteristics of these models. For example, some models 1202 may be independent architecture frameworks, such as MLP, CNN, or RNN, while other models 1202 may be a combination of the aforementioned architecture frameworks. In addition, the models 1202 may have different numbers of layers, kernel sizes, activation functions used in different layers, weights of different layers, etc.

[0179] At 1806, the network node may receive second ML-based feature data from the UE based on the second ML-based feature data extraction model of the UE. For example, 1806 may be performed by the feature data receiving component 2040. For example, referring to Figures 9 to 16 ,exist Fig.11 or Fig.15At 8.6 of , the ML service entity 1014, 1104, 1302, 1414, 1504, 1604 (or the RX process 370 of the network node or base station 310) may receive training data 902 for training at a non-RT RIC or inference data 904 for inference at a near-RT RIC (e.g., training or inference data 1314, 1316), OBB 1206, or features 1006, 1012, 1406, 1412 from the UE 104, 104a, 1102, 1502, 1602. Alternatively, a base station or ML server interfaced with the ML service entity, or a component of a decomposed base station interfaced with the ML service entity may receive the data. The data may be received by, for example, Fig.11 or Fig.15 The ML-based feature data received by the ML service entity at 8.6 may be a new model in the model 1202 available at the UE for feature extraction to which the UE is switched based on the configuration instruction at 1804. For example, Fig.11 or Fig.15 The OBB 1206 received at 8.6 of UE may be output from the new model 1202 after the model switch. Similarly, the ML-based feature data may be used for beam blocking prediction, or the beam blocking prediction may be based on the new model of the UE. Figure 7 ,exist Fig.11 or Fig.15 8.6 The OBB 1206 received from UE 704a may be better than Fig.11 8.1 or Fig.15 8.2 may be faster or more accurate, and may therefore better identify potential UEs or other objects (eg, truck 711) that the ML service entity 749 may predict is causing LOS blocking of a beam (eg, beam 707).

[0180] At 1808, the network node can determine a beam blocking prediction in response to the second ML-based feature data. For example, 1808 can be performed by the prediction determination component 2044. For example, referring to Figure 8 , the ML service entity 830 may include a beam blocking prediction model 1016, 1416 (e.g., a neural network 826) that Fig.11 or Fig.158.6 of the UE to predict whether there is beam blocking. For example, the model may receive at least an aggregation of OBB 811.1-811.N (including OBB 1206 of the new model from the UE after the model switch) as input, forward pass these inputs together with the model weights to one or more activation functions of the neural network layer, and output a prediction in response to the forward pass. After obtaining the model prediction / output, the ML service entity at the ML server or base station may interpret the presence of LOS beam blocking and communicate the prediction to the ML server or base station to perform beam management to resolve the LOS blocking. Therefore, if the network node is an ML service entity, the network node may determine the beam blocking prediction by obtaining the model prediction / output from the model and / or interpreting the model prediction / output as LOS beam blocking. Similarly, if the network node is an entity interfaced with an ML service entity, such as a converged or disaggregated base station, an ML server co-located with or located near a base station, or a component of a disaggregated base station (e.g., a controller / processor 375, or a CU, DU, RU, etc. of a network node or base station 310), the network node can determine the beam blocking prediction by obtaining the transmitted model prediction and / or interpreted LOS beam blocking from the ML service entity.

[0181] Fig.191900 is a diagram illustrating an example of a hardware implementation for an apparatus 1902. The apparatus 1902 is a UE and includes a cellular baseband processor 1904 (also referred to as a modem) coupled to a cellular RF transceiver 1922 and one or more subscriber identity modules (SIM) cards 1920, an application processor 1906 coupled to a secure digital (SD) card 1908 and a screen 1910, a Bluetooth module 1912, a wireless local area network (WLAN) module 1914, a global positioning system (GPS) module 1916, and a power source 1918. The cellular baseband processor 1904 communicates with the UE 104 and / or the BS 102 / 180 via the cellular RF transceiver 1922. The cellular baseband processor 1904 may include a computer-readable medium / memory. The computer-readable medium / memory may be non-transitory. The cellular baseband processor 1904 is responsible for general processing, including executing software stored on the computer-readable medium / memory. The software, when executed by the cellular baseband processor 1904, causes the cellular baseband processor 1904 to perform the various functions described above. The computer-readable medium / memory may also be used to store data manipulated by the cellular baseband processor 1904 when executing the software. The cellular baseband processor 1904 also includes a receiving component 1930, a communication manager 1932, and a sending component 1934. The communication manager 1932 includes one or more of the illustrated components. The components within the communication manager 1932 may be stored in a computer-readable medium / memory and / or configured as hardware within the cellular baseband processor 1904. The cellular baseband processor 1904 may be a component of the UE 350 and may include a memory 360 and / or at least one of a TX processor 368, an RX processor 356, and a controller / processor 359. In one configuration, the device 1902 may be a modem chip and include only the baseband processor 1904, and in another configuration, the device 1902 may be the entire UE (e.g., see Figure 3 350), and includes the aforementioned additional modules of device 1902.

[0182] The communication manager 1932 includes a configuration receiving component 1940 configured to receive a message instructing the device to switch from a current machine learning (ML)-based feature data extraction model of the device to one of a plurality of ML-based feature data extraction models based on the state of the device, the plurality of ML-based feature data extraction models each providing ML-based feature data for predicting beam blocking between the device and a network entity, for example, as described in conjunction with 1702. The message may indicate the performance of at least one ML model of the plurality of ML models for beam management. The message may also include an instruction for the device to send ML-based feature data of a different model of the ML models for beam management. The communication manager 1932 also includes a model switching determination component 1942 that receives input in the form of a message from the configuration receiving component 1940 and is configured to determine, at least in part based on the message, to switch from the current ML-based feature data extraction model to one of the ML-based feature data extraction models, for example, as described in conjunction with 1704. The ML-based feature data extraction models may include different calculation speeds and different detection accuracies. The communication manager 1932 also includes a feature data sending component 1946, which receives input in the form of one of the ML-based feature data extraction models from the model switching determination component 1942 and is configured to send ML-based feature data for predicting beam blocking based on one of the ML-based feature data extraction models in response to the switching, for example, as described in conjunction with 1708.

[0183] In one example, the model switching determination component 1942 may be further configured to determine switching to one of the ML-based feature data extraction models independently of an aggregate performance characteristic of an ML model for beam blocking prediction of a network node, wherein the ML model for beam blocking prediction includes an aggregation of input ML-based feature data from a plurality of UEs including the device. In another example, the communication manager 1932 may further include a performance characteristic receiving component 1944, which is configured to receive an aggregate performance characteristic of an ML model for beam blocking prediction, wherein the ML model for beam blocking prediction includes an aggregation of input ML-based feature data from a plurality of UEs including the device, for example, as described in conjunction with 1706. The model switching determination component 1942 in this other example may receive input in the form of an aggregate performance characteristic from the performance characteristic receiving component 1944, and may be further configured to determine switching to one of the ML-based feature data extraction models based on the aggregate performance characteristic.

[0184] The communication manager 1932 may also include a confidence level sending component 1948 that receives input in the form of ML-based feature data from the feature data sending component 1946 and is configured to send a confidence level associated with the ML-based feature data, e.g., as described in conjunction with 1710. In one example, the model switching determination component 1942 may receive input in the form of a confidence level from the confidence level sending component 1948 and may be further configured to determine to switch to one of the ML-based feature data extraction models based on the ability of one of the ML-based feature data extraction models to derive a confidence level.

[0185] The message received by the configuration receiving component 1940 may also include an instruction for the device to reconfigure a sensor of the device, and the ML-based feature data may be further based on the sensor. In one example, the message may be received in response to a criterion for sensor reconfiguration being met. In another example, the communication manager 1932 may also include a criterion determination component 1950, which is configured to determine that the criterion for sensor reconfiguration is met, for example, as described in conjunction with 1712, and the communication manager 1932 may also include a sensor reconfiguration component 1952, which may receive input in the form of a criterion determination from the criterion determination component 1950 and may be configured to reconfigure at least one of the sensor's FoV, range, measurement update rate, resolution, or frame rate in response to meeting the criterion, for example, as described in conjunction with 1714.

[0186] The apparatus may include executing Fig.17 Each of the boxes in the algorithm in the preceding flowchart is an additional component. Therefore, Fig.17 Each block in the aforementioned flow chart of can be performed by a component, and the apparatus may include one or more of those components. These components may be one or more hardware components that are specifically configured to perform the process / algorithm, implemented by a processor configured to perform the process / algorithm, stored in a computer-readable medium for implementation by a processor, or some combination thereof.

[0187] In one configuration, the device 1902 (and specifically, the cellular baseband processor 1904) includes a component for receiving a message instructing the device to switch from a current machine learning (ML)-based feature data extraction model of the device to one of a plurality of ML-based feature data extraction models based on a state of the device, the plurality of ML-based feature data extraction models each providing ML-based feature data for predicting beam blocking between the device and a network entity; a component for determining to switch from the current ML-based feature data extraction model to one of the ML-based feature data extraction models based at least in part on the message; and a component for sending ML-based feature data for predicting beam blocking based on one of the ML-based feature data extraction models in response to the switch.

[0188] In one configuration, the ML-based feature data extraction model may include different computation speeds and different detection accuracies.

[0189] In one configuration, the message may indicate the performance of at least one ML model of the plurality of ML models for beam management. In one configuration, the message may also include instructions for the device to send ML-based feature data for different models of the ML models for beam management.

[0190] In one configuration, the state of the device may include at least one of: the mobility state of the device, the number of UEs in the area where the device is located, the data processing capability of the device, the amount of uplink traffic sharing the bandwidth of the device, or the uplink traffic load of the network including the device.

[0191] In one configuration, determining to switch to a second one of the ML-based feature data extraction models may be independent of aggregate performance characteristics of an ML model for beam blocking prediction of a network node, the ML model for beam blocking prediction comprising an aggregation of ML-based feature data input from a plurality of UEs comprising the device.

[0192] In one configuration, the component for receiving may be further configured to receive an aggregate performance characteristic of an ML model for beam blocking prediction, the ML model for beam blocking prediction comprising an aggregation of ML-based feature data of inputs from a plurality of UEs comprising the device, wherein determining to switch to one of the ML-based feature data extraction models is further based on the aggregate performance characteristic.

[0193] In one configuration, the means for sending may be further configured to send a confidence level associated with the ML-based feature data. In one configuration, determining to switch to one of the ML-based feature data extraction models may be based on an ability of the one of the ML-based feature data extraction models to derive a confidence level.

[0194] In one configuration, the message may also include an instruction for the device to reconfigure a sensor of the device, and the ML-based feature data may be further based on the sensor. In one configuration, the configuration may be received in response to a criterion for sensor reconfiguration being met. In one configuration, the means for determining may be further configured to determine that the criterion for sensor reconfiguration is met, and the device 1902 (and specifically, the cellular baseband processor 1904) may include means for reconfiguring at least one of the FoV, range, measurement update rate, resolution, or frame rate of the sensor in response to the criterion being met.

[0195] The aforementioned means may be one or more of the aforementioned components of the device 1902 configured to perform the functions recited by the aforementioned means. As described above, the device 1902 may include a TX processor 368, an RX processor 356, and a controller / processor 359. Therefore, in one configuration, the aforementioned means may be a TX processor 368, an RX processor 356, and a controller / processor 359 configured to perform the functions recited by the aforementioned means.

[0196] Fig. 20 2000 is an example of a hardware implementation for an apparatus 2002. The apparatus 2002 is a BS and includes a baseband unit 2004. The baseband unit 2004 can communicate with the UE 104 via a cellular RF transceiver. The baseband unit 2004 may include a computer-readable medium / memory. The baseband unit 2004 is responsible for general processing, including executing software stored on a computer-readable medium / memory. The software, when executed by the baseband unit 2004, causes the baseband unit 2004 to perform the various functions described above. The computer-readable medium / memory may also be used to store data manipulated by the baseband unit 2004 when executing the software. The baseband unit 2004 also includes a receiving component 2030, a communication manager 2032, and a sending component 2034. The communication manager 2032 includes one or more of the illustrated components. The components within the communication manager 2032 may be stored in a computer-readable medium / memory and / or configured as hardware within the baseband unit 2004. The baseband unit 2004 may be a component of the BS 310 and may include a memory 376 and / or at least one of the TX processor 316 , the RX processor 370 , and the controller / processor 375 .

[0197] The communication manager 2032 includes a feature data receiving component 2040 configured to receive first ML-based feature data from the UE based on a first ML-based feature data extraction model of the UE, for example, as described in conjunction with 1802. The communication manager 2032 also includes a configuration sending component 2042 that receives input in the form of the first ML-based feature data from the feature data receiving component 2040 and is configured to send a message instructing the UE to switch from the first ML-based feature data extraction model of the UE to a second ML-based feature data extraction model based on the state of the UE, for example, as described in conjunction with 1804. The network node may also include multiple ML models for beam management, and the message may be based on the performance of at least one of the ML models for beam management. In addition, the message may include an instruction for the UE to send different ML-based feature data of different ML models for beam management of the network node. The feature data receiving component 2040 may be further configured to receive second ML-based feature data from the UE based on the second ML-based feature data extraction model of the UE, for example, as described in conjunction with 1806. The communication manager 2032 also includes a prediction determination component 2044 that receives input in the form of second ML-based feature data from the feature data receiving component 2040 and is configured to determine a beam blockage prediction in response to the second ML-based feature data, e.g., as described in conjunction with 1808. The first ML-based feature data extraction model and the second ML-based feature data extraction model may include different computation speeds and different detection accuracies.

[0198] The apparatus may include executing Fig.18 Each of the boxes in the algorithm in the preceding flowchart is an additional component. Therefore, Fig.18 Each block in the above flowchart may be performed by a component, and the apparatus may include one or more of these components. These components may be one or more hardware components that are specifically configured to perform the process / algorithm, implemented by a processor configured to perform the process / algorithm, stored in a computer-readable medium for implementation by a processor, or some combination thereof.

[0199] In one configuration, the apparatus 2002 (and specifically, the baseband unit 2004) includes a component for receiving first ML-based feature data from a UE based on a first ML-based feature data extraction model of the UE; a component for sending a message instructing the UE to switch from the first ML-based feature data extraction model of the UE to a second ML-based feature data extraction model based on a state of the UE; wherein the component for receiving is further configured to receive second ML-based feature data from the UE based on the second ML-based feature data extraction model of the UE; and a component for determining a beam blocking prediction in response to the second ML-based feature data.

[0200] In one configuration, the first ML-based feature data extraction model and the second ML-based feature data extraction model may include different computation speeds and different detection accuracies.

[0201] In one configuration, the network node may also include multiple ML models for beam management, and the message may be based on the performance of at least one of the ML models for beam management. In one configuration, the message may also include instructions for the UE to send different ML-based feature data for different ML models for beam management of the network node.

[0202] The aforementioned means may be one or more of the aforementioned components of the apparatus 2002 configured to perform the functions recited by the aforementioned means. As described above, the apparatus 2002 may include a TX processor 316, an RX processor 370, and a controller / processor 375. Therefore, in one configuration, the aforementioned means may be a TX processor 316, an RX processor 370, and a controller / processor 375 configured to perform the functions recited by the aforementioned means.

[0203] Thus, aspects of the present disclosure allow a network node to configure a UE to switch between UE feature data extraction models and optionally reconfigure UE sensors to provide adaptive feature data of dynamic or moving potential LOS obstacles for improving beam blocking prediction performance, beam management adaptation, scheduling, load balancing, or other functions at the network node. Some of the feature data extraction models may be more complex than other models (e.g., have longer calculation timing, include different amounts of data collection, etc.), and some of these models may have different performance than other models (e.g., have lower accuracy, more false alarms or misdetection of OBB, etc.). Thus, depending on the state of the UE in the case where the UE is present in a mobile or dynamic environment, the network node may configure the UE to switch between different types of models (e.g., faster or more accurate models) to adaptively improve the performance of its ML-based beam blocking prediction or other inference. The network node may also include multiple ML models serving different functions (such as beamforming or beam refinement), and model switching may result in different feature data, which the network node may apply respectively to improve the performance of its different ML models. The UE may also provide a confidence level with the associated feature data to assist the network node in determining the accuracy of the received feature data, and the network node may further optimize its model performance by instructing the UE to switch to a model capable of inferring such confidence level or to switch to a more accurate model if the confidence level is below a given threshold. To simplify the implementation, the training or inference performed at the UE and the network node may also be separated or decoupled, or associated together in a joint system (i.e., by reporting aggregate performance characteristics to the UE by the network node) for improving performance optimization of the UE feature data extraction model based on feature data from other UEs.

[0204] It should be understood that the specific order or hierarchy of the blocks in the disclosed process / flowchart is only an illustration of the exemplary method. It should be understood that the specific order or hierarchy of the blocks in the process / flowchart can be rearranged based on design preferences. Further, some blocks can be combined or omitted. The attached method claims provide the elements of various blocks in a sample order, but are not meant to be limited to the specific order or hierarchy provided.

[0205] The foregoing description is provided to enable any person skilled in the art to practice the various aspects described herein. Various modifications to these aspects will be apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects. Therefore, the claims are not intended to be limited to the aspects shown herein, but to conform to the full scope consistent with the language claims, wherein unless otherwise specified, the elements mentioned in the singular are not intended to represent "one and only one", but "one or more". Terms such as "if", "when ......" and "while ......" should be interpreted as "under the conditions of ......", rather than meaning an immediate time relationship or reaction. That is, these phrases, such as "when ......", do not mean an immediate action in response to the occurrence of an action or during the occurrence of an action, but only mean that if the conditions are met, the action will occur, but no specific or immediate time limit is required for the occurrence of the action. The word "exemplary" is used herein to mean "used as an example, instance or illustration". Any aspect described as "exemplary" herein is not necessarily interpreted as being preferred or having advantages over other aspects. Unless otherwise specified, the term "some" refers to one or more. Combinations such as “at least one of A, B, or C,” “one or more of A, B, or C,” “at least one of A, B and C,” “one or more of A, B, and C,” and “A, B, C, or any combination thereof” include any combination of A, B, and / or C, and may include multiple A, multiple B, or multiple C. Specifically, combinations such as “at least one of A, B, or C,” “one or more of A, B, or C,” “at least one of A, B, and C,” “one or more of A, B, and C,” and “A, B, C, or any combination thereof” may be A only, B only, C only, A and B, A and C, B and C, or A and B and C, wherein any such combination may include one or more members of A, B, or C. All structural and functional equivalents to the elements of the various aspects described in this disclosure that are known or later come to be known to those of ordinary skill in the art are expressly incorporated herein by reference and are intended to be encompassed by the claims. Furthermore, nothing disclosed herein is intended to be disclosed to the public, regardless of whether such disclosure is explicitly stated in the claims. Words such as "module," "mechanism," "element," "device," etc. cannot replace the word "component." Therefore, no claim element will be understood as a functional component unless the element is explicitly stated using the phrase "component for..."

[0206] The following examples are merely illustrative and may be combined with aspects of other embodiments or teachings described herein, but are not limited thereto.

[0207] Embodiment 1 is a device, comprising: a processor; a memory coupled to the processor; and instructions, the instructions being stored in the memory and operable when executed by the processor to cause the device to: receive a message instructing the device to switch from a current machine learning (ML)-based feature data extraction model of the device to one of a plurality of ML-based feature data extraction models based on a state of the device, wherein each of the plurality of ML-based feature data extraction models provides ML-based feature data for predicting beam blocking between the device and a network entity; determine to switch from the current ML-based feature data extraction model to the one of the ML-based feature data extraction models based at least in part on the message; and send the ML-based feature data for predicting the beam blocking based on the one of the ML-based feature data extraction models in response to the switch.

[0208] Embodiment 2 is a device according to embodiment 1, wherein the ML-based feature data extraction model includes different calculation speeds and different detection accuracies.

[0209] Embodiment 3 is an apparatus according to embodiment 1 or 2, wherein the message indicates the performance of at least one ML model among a plurality of ML models for beam management.

[0210] Embodiment 4 is an apparatus according to any one of embodiments 1 to 3, wherein the message further includes instructions for the apparatus to send different ML-based feature data for different ML models for beam management.

[0211] Embodiment 5 is a device according to any one of Embodiments 1 to 4, wherein the state of the device includes at least one of the following: the mobility state of the device, the number of user equipment (UE) in the area where the device is located, the data processing capability of the device, the uplink traffic volume sharing the bandwidth of the device, or the uplink traffic load of the network including the device.

[0212] Embodiment 6 is an apparatus according to any one of embodiments 1 to 5, wherein the instructions, when executed by the processor, further cause the apparatus to: receive aggregate performance characteristics of an ML model for beam blocking prediction, the ML model for beam blocking prediction comprising an aggregation of ML-based feature data input from multiple UEs including the apparatus, wherein the determination to switch to the one of the ML-based feature data extraction models is further based on the aggregate performance characteristics.

[0213] Embodiment 7 is the apparatus of embodiments 1 to 6, wherein the instructions, when executed by the processor, further cause the apparatus to: send a confidence level associated with the ML-based feature data.

[0214] Embodiment 8 is an apparatus according to embodiment 7, wherein the determination to switch to said one of the ML-based feature data extraction models is based on the ability of said one of the ML-based feature data extraction models to derive the confidence level.

[0215] Embodiment 9 is the apparatus of any one of embodiments 1 to 8, wherein the message further includes an instruction for the apparatus to reconfigure a sensor of the apparatus, and the ML-based feature data is further based on the sensor.

[0216] Embodiment 10 is the apparatus of embodiment 9, wherein the message is received in response to criteria for sensor reconfiguration being met.

[0217] Embodiment 11 is a device according to embodiment 9 or 10, wherein the instructions, when executed by the processor, further cause the device to: determine that criteria for sensor reconfiguration are met; and reconfigure at least one of the field of view (FoV), range, measurement update rate, resolution, or frame rate of the sensor in response to meeting the criteria.

[0218] Embodiment 12 is a method for performing wireless communications at a user equipment (UE), comprising: receiving a message instructing the UE to switch from a current machine learning (ML)-based feature data extraction model of the UE to one of a plurality of ML-based feature data extraction models based on a state of the UE, wherein each of the plurality of ML-based feature data extraction models provides ML-based feature data for predicting beam blocking between the UE and a network entity; determining to switch from the current ML-based feature data extraction model to the one of the ML-based feature data extraction models based at least in part on the message; and sending the ML-based feature data for predicting the beam blocking based on the one of the ML-based feature data extraction models in response to the switch.

[0219] Embodiment 13 is a method according to embodiment 12, wherein the ML-based feature data extraction model includes different calculation speeds and different detection accuracies.

[0220] Embodiment 14 is a method according to embodiment 12 or 13, wherein the message indicates the performance of at least one ML model among a plurality of ML models for beam management.

[0221] Embodiment 15 is a method according to any one of embodiments 12 to 14, wherein the message further includes an instruction for the UE to send different ML-based feature data for different ML models for beam management.

[0222] Embodiment 16 is a method according to any one of Embodiments 12 to 15, wherein the state of the UE includes at least one of the following: the mobility state of the UE, the number of UEs in the area where the UE is located, the data processing capability of the UE, the uplink traffic volume sharing the bandwidth of the UE, or the uplink traffic load of the network including the UE.

[0223] Embodiment 17 is a method according to any one of embodiments 12 to 16, further comprising: receiving aggregate performance characteristics of an ML model for beam blocking prediction, the ML model for beam blocking prediction comprising an aggregation of ML-based feature data input from multiple UEs including the UE, wherein the determination to switch to the one of the ML-based feature data extraction models is further based on the aggregate performance characteristics.

[0224] Embodiment 18 is a method according to any one of embodiments 12 to 17, further comprising: sending a confidence level associated with the ML-based feature data.

[0225] Embodiment 19 is a method according to embodiment 18, wherein the determination to switch to said one of the ML-based feature data extraction models is based on the ability of said one of the ML-based feature data extraction models to derive the confidence level.

[0226] Embodiment 20 is a method according to any one of embodiments 12 to 19, wherein the message further includes an instruction for the UE to reconfigure a sensor of the UE, and the ML-based feature data is further based on the sensor.

[0227] Embodiment 21 is a method according to embodiment 20, wherein the message is received in response to criteria for sensor reconfiguration being met.

[0228] Embodiment 22 is a method according to embodiment 20 or 21, further comprising: determining criteria for satisfying sensor reconfiguration; and reconfiguring at least one of the field of view (FoV), range, measurement update rate, resolution, or frame rate of the sensor in response to satisfying the criteria.

[0229] Embodiment 23 is a network node, comprising: a processor; a memory coupled to the processor; and instructions, the instructions being stored in the memory and being operable when executed by the processor to cause the network node to: receive first machine learning (ML)-based feature data from a user equipment (UE) based on a first ML-based feature data extraction model of the UE; send a message instructing the UE to switch from the first ML-based feature data extraction model of the UE to a second ML-based feature data extraction model based on a state of the UE; receive second ML-based feature data from the UE based on the second ML-based feature data extraction model of the UE; and determine a beam blocking prediction in response to the second ML-based feature data.

[0230] Embodiment 24 is a network node according to Embodiment 23, wherein the first ML-based feature data extraction model and the second ML-based feature data extraction model include different calculation speeds and different detection accuracies.

[0231] Embodiment 25 is a network node according to embodiment 23 or 24, wherein the network node also includes multiple ML models for beam management, and the message is based on the performance of at least one ML model among the ML models for beam management.

[0232] Embodiment 26 is a network node according to any one of embodiments 23 to 25, wherein the message further includes instructions for the UE to send different ML-based feature data for different ML models for beam management of the network node.

[0233] Embodiment 27 is a method for performing wireless communications at a network node, comprising: receiving first machine learning (ML)-based feature data from a user equipment (UE) based on a first ML-based feature data extraction model of the UE; sending a message instructing the UE to switch from the first ML-based feature data extraction model of the UE to a second ML-based feature data extraction model based on a state of the UE; receiving second ML-based feature data from the UE based on the second ML-based feature data extraction model of the UE; and determining a beam blocking prediction in response to the second ML-based feature data.

[0234] Embodiment 28 is a method according to Embodiment 27, wherein the first ML-based feature data extraction model and the second ML-based feature data extraction model include different calculation speeds and different detection accuracies.

[0235] Embodiment 29 is a method according to embodiment 27 or 28, wherein the network node further includes a plurality of ML models for beam management, and the message is based on the performance of at least one of the ML models for beam management.

[0236] Embodiment 30 is a method according to embodiment 29, wherein the message further includes an instruction for the UE to send different ML-based feature data for different ML models for beam management of the network node.

Claims

1. A device, comprising: processor; a memory coupled to the processor; and instructions stored in the memory and operable when executed by the processor to cause the apparatus to: receiving, based on a state of the device, a message instructing the device to switch from a current machine learning (ML)-based feature data extraction model of the device to one of a plurality of ML-based feature data extraction models, the plurality of ML-based feature data extraction models each providing ML-based feature data for predicting beam blockage between the device and a network entity; determining to switch from the current ML-based feature data extraction model to the one of the ML-based feature data extraction models based at least in part on the message; as well as The ML-based feature data for predicting the beam blocking is transmitted based on the one of the ML-based feature data extraction models in response to the switching.

2. The apparatus according to claim 1, wherein the plurality of ML-based feature data extraction models include different calculation speeds and different detection accuracies. 3 . The apparatus of claim 1 , wherein the message indicates performance of at least one ML model among a plurality of ML models for beam management. 4 . The apparatus of claim 1 , wherein the message further comprises an instruction for the apparatus to send different ML-based feature data for different ML models for beam management.

5. The device of claim 1, wherein the state of the device comprises at least one of: the mobility state of the device, the number of user equipment (UE) in the area where the device is located, the data processing capabilities of the device, uplink traffic sharing the bandwidth of the device, or An uplink traffic load of a network including the apparatus.

6. The apparatus of claim 1, wherein the instructions, when executed by the processor, further cause the apparatus to: receiving an aggregate performance characteristic of an ML model for beam blocking prediction, the ML model for beam blocking prediction comprising an aggregation of input ML-based feature data from a plurality of UEs comprising the apparatus, wherein the determining to switch to the one of the ML-based feature data extraction models is further based on the aggregate performance characteristic.

7. The apparatus of claim 1, wherein the instructions, when executed by the processor, further cause the apparatus to: A confidence level associated with the ML-based feature data is transmitted.

8. The apparatus of claim 7, wherein the determining to switch to the one of the ML-based feature data extraction models is based on a capability of the one of the ML-based feature data extraction models to derive the confidence level.

9. The apparatus of claim 1, wherein the message further comprises an instruction for the apparatus to reconfigure a sensor of the apparatus, and the ML-based feature data is further based on the sensor.

10. The apparatus of claim 9, wherein the message is received in response to criteria for sensor reconfiguration being met.

11. The apparatus of claim 9, wherein the instructions, when executed by the processor, further cause the apparatus to: Determining criteria that meet sensor reconfiguration criteria; and At least one of a field of view (FoV), range, measurement update rate, resolution, or frame rate of the sensor is reconfigured in response to the criteria being met.

12. A method of wireless communication at a user equipment (UE), the method comprising: receiving, based on a state of the UE, a message instructing the UE to switch from a current machine learning (ML) based feature data extraction model of the UE to one of a plurality of ML based feature data extraction models, the plurality of ML based feature data extraction models each providing ML based feature data for predicting beam blocking between the UE and a network entity; determining, based at least in part on the message, to switch from the current ML-based feature data extraction model to the one of the ML-based feature data extraction models; and The ML-based feature data for predicting the beam blocking is transmitted based on the one of the ML-based feature data extraction models in response to the switching.

13. The method according to claim 12, wherein the plurality of ML-based feature data extraction models include different calculation speeds and different detection accuracies.

14. The method of claim 12, wherein the message indicates performance of at least one ML model among a plurality of ML models for beam management.

15. The method of claim 12, wherein the message further comprises an instruction for the UE to send different ML-based feature data for different ML models for beam management.

16. The method of claim 12, wherein the state of the UE comprises at least one of: the mobility state of the UE, The number of UEs in the area where the UE is located, The data processing capability of the UE, uplink traffic sharing the bandwidth of the UE, or An uplink traffic load of a network including the UE.

17. The method according to claim 12, further comprising: receiving an aggregate performance characteristic of an ML model for beam blocking prediction, the ML model for beam blocking prediction comprising an aggregation of input ML-based feature data from a plurality of UEs including the UE, wherein the determining to switch to the one of the ML-based feature data extraction models is further based on the aggregate performance characteristic.

18. The method according to claim 12, further comprising: A confidence level associated with the ML-based feature data is transmitted.

19. The method of claim 18, wherein the determining to switch to the one of the ML-based feature data extraction models is based on a capability of the one of the ML-based feature data extraction models to derive the confidence level.

20. The method of claim 12, wherein the message further comprises an instruction to the UE to reconfigure a sensor of the UE, and the ML-based feature data is further based on the sensor.

21. The method of claim 20, wherein the message is received in response to criteria for sensor reconfiguration being met.

22. The method according to claim 20, further comprising: determining criteria for satisfying sensor reconfiguration; as well as At least one of a field of view (FoV), range, measurement update rate, resolution, or frame rate of the sensor is reconfigured in response to the criteria being met.

23. A network node, the network node comprising: processor; a memory coupled to the processor; and instructions stored in the memory and operable when executed by the processor to cause the network node to: receiving first machine learning (ML) based feature data extraction model from a user equipment (UE) based on the UE; sending a message instructing the UE to switch from the first ML-based feature data extraction model of the UE to a second ML-based feature data extraction model based on a state of the UE; receiving second ML-based feature data from the UE based on the second ML-based feature data extraction model of the UE; as well as A beam blockage prediction is determined in response to the second ML-based feature data.

24. The network node of claim 23, wherein the first ML-based feature data extraction model and the second ML-based feature data extraction model comprise different calculation speeds and different detection accuracies.

25. The network node of claim 23, wherein the network node further comprises a plurality of ML models for beam management, and the message is based on a performance of at least one of the ML models for beam management.

26. The network node of claim 23, wherein the message further comprises instructions for the UE to send different ML-based feature data for different ML models for beam management of the network node.

27. A method for wireless communication at a network node, the method comprising: receiving first machine learning (ML) based feature data extraction model from a user equipment (UE) based on the UE; sending a message instructing the UE to switch from the first ML-based feature data extraction model of the UE to a second ML-based feature data extraction model based on a state of the UE; receiving second ML-based feature data from the UE based on the second ML-based feature data extraction model of the UE; as well as A beam blockage prediction is determined in response to the second ML-based feature data.

28. The method of claim 27, wherein the first ML-based feature data extraction model and the second ML-based feature data extraction model comprise different calculation speeds and different detection accuracies.

29. The method of claim 27, wherein the network node further comprises a plurality of ML models for beam management, and the message is based on a performance of at least one of the ML models for beam management.

30. The method of claim 29, wherein the message further comprises an instruction for the UE to send different ML-based feature data for different ML models for beam management of the network node.