Model monitoring using reference model

Abstract

Apparatus, methods, and computer-readable media for performing wireless communication are disclosed. An example method of wireless communication includes a method performed at a user equipment (UE). The method comprises: generating a first representation of control information associated with the communication channel using a machine learning model under test; generating a second representation of control information associated with the communication channel using a reference machine learning model; and transmitting information associated with a comparison based on the first representation of the control information and the second representation of the control information to a device. The comparison may occur on the UE or on the device, and may be based on the representation of the control information or a reconstructed representation of the control information.

Description

Technical Field

[0001] The present disclosure generally relates to machine learning (ML) systems for wireless communications. For example, aspects of the present disclosure relate to systems and techniques for monitoring the performance of machine learning models deployed at a device using one or more reference machine learning models. Background Art

[0002] Wireless communication systems are deployed to provide a variety of telecommunication and data services including telephone, video, data, messaging, and broadcasting. Broadband wireless communication systems have evolved over several generations, including first generation analog wireless telephone service (1G), second generation (2G) digital wireless telephone service (including transitional 2.5G networks), third generation (3G) high-speed data wireless devices with Internet capabilities, and fourth generation (4G) services (e.g., Long Term Evolution (LTE), WiMax). Examples of wireless communication systems 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, global system for mobile communications (GSM) systems, and the like. Other wireless communication technologies include 802.11 Wi-Fi, Bluetooth, and the like.

[0003] The fifth generation (5G) mobile standard requires higher data transfer speeds, a greater number of connections, and better coverage, among other improvements. According to the Next Generation Mobile Network Alliance, the 5G standard (also known as "new radio" or "NR") is designed to provide tens of megabits per second of data rates to each of tens of thousands of users, and 1 gigabit per second of data rates to dozens of employees in an office floor. To support large sensor deployments, hundreds of thousands of simultaneous connections should be supported. Algorithms based on artificial intelligence (AI) and ML can be incorporated into 5G and future standards to improve telecommunications and data services. Such ML-based algorithms can be trained for specific environments, such as indoor environments versus outdoor environments. Summary of the invention

[0004] The following presents a simplified summary of the invention related to one or more aspects disclosed herein. Therefore, the following summary of the invention should neither be considered as an exhaustive overview related to all conceived aspects, nor should it be considered to identify key or decisive elements related to all conceived aspects or to delineate the scope associated with any particular aspect. Accordingly, the following summary of the invention presents certain concepts related to one or more aspects of the mechanisms disclosed herein in a simplified form before the detailed embodiments presented below.

[0005] Systems and techniques are described herein for monitoring the performance of one or more test ML models (e.g., test neural network models) using one or more reference machine learning (ML) models (e.g., reference neural network models) to identify situations where reconstructed control information (e.g., channel state information (CSI)) is different from target control information (e.g., CSI) intended to be transmitted by the device. In some aspects, the test ML model deployed on a first device (e.g., a UE, a base station, or a portion of a base station (such as a central unit (CU), a distributed unit (DU), a radio unit (RU), etc.)) may be monitored at a second device (e.g., a base station, a user equipment (UE), etc.) based on data received at the second device from the first device.

[0006] In some cases, the data received at the second device may be based on the output generated using the test ML model at the first device and the output generated using the reference ML model at the first device. For example, the encoder of the test ML model on the first device may be trained to generate a compressed representation (e.g., a latent representation, such as a latent code) of control information (such as CSI or other control information) associated with the communication channel. The encoder of the reference ML model on the first device may also be trained to generate a compressed representation of the control information (e.g., CSI) associated with the communication channel. The first device may send the compressed representation of the control information to the second device. Upon receiving the two compressed representations of the control information, the second device may use the decoder of the ML model deployed on the second device to reconstruct the control information from each corresponding compressed representation. If the difference between the two versions of the reconstructed control information (e.g., the reconstructed CSI) is below a threshold difference, the second device may determine that the performance of the test model deployed on the first device is accurate for the communication channel. In some cases, the CSI corresponds to a pre-decoding vector, or includes a pre-decoding vector. In such cases, a metric such as a normalized mean square error (NMSE) or a cosine similarity metric may be used to determine the threshold difference. As an example, based on the NMSE calculation, the threshold difference may be -10dB. For example, if the value calculated based on the NMSE is -10 dB or less, then the test model may be determined to be accurate for the communication channel. Other values ​​are also contemplated. If the difference between two versions of the reconstructed control information (e.g., reconstructed CSI) is greater than (or not less than) a threshold difference, then the second device may determine that the performance of the test model deployed on the first device is inaccurate for the communication channel.

[0007] In other aspects, the monitoring or comparison described herein for determining the accuracy or adequacy of the test ML model can be performed on the first device or on both the first device and the second device. For example, in some cases, the first device can monitor the ML model deployed on the first device, such as by comparing the output generated using the test ML model with the output generated using the reference ML model.

[0008] According to at least one example, a method of wireless communication is performed at a user equipment (UE). The method includes: using a tested machine learning model to generate a first representation of control information associated with a communication channel; using a reference machine learning model to generate a second representation of the control information associated with the communication channel; and sending information associated with a comparison based on the first representation of the control information and the second representation of the control information to a device.

[0009] In another aspect, an apparatus for wireless communication may include: at least one memory; and at least one processor coupled to the at least one memory. The at least one processor may be configured to: generate a first representation of control information associated with a communication channel using a tested machine learning model; generate a second representation of the control information associated with the communication channel using a reference machine learning model; and send information associated with a comparison based on the first representation of the control information and the second representation of the control information to a device.

[0010] In another aspect, a non-transitory computer-readable medium is provided, which has instructions that, when executed by one or more processors, cause the one or more processors to: generate a first representation of control information associated with a communication channel using a tested machine learning model; generate a second representation of the control information associated with the communication channel using a reference machine learning model; and send to a device information associated with a comparison based on the first representation of the control information and the second representation of the control information.

[0011] In another example, an apparatus for wireless communication may include components for generating a first representation of control information associated with a communication channel using a tested machine learning model; generating a second representation of the control information associated with the communication channel using a reference machine learning model; and sending information associated with a comparison based on the first representation of the control information and the second representation of the control information to a device.

[0012] In some aspects, one or more of the devices described herein are, are part of, and / or include a user equipment (UE) such as a wireless communication device (e.g., a mobile device such as a mobile phone or other mobile device), an extended reality device (e.g., a virtual reality (VR) device, an augmented reality (AR) device, or a mixed reality (MR) device), a vehicle or a computing system, device, or component of a vehicle, a network-connected wearable device (e.g., a network-connected watch), a camera, a personal computer, a laptop computer, a server computer, or other UE. In some aspects, one or more of the devices described herein are, are part of, and / or include a base station (e.g., an eNodeB, a gNodeB, or other base station) or a part of a base station (e.g., a central unit (CU), a distributed unit (DU), a radio unit (RU), or other part of a base station with a decomposed architecture). In some cases, the device includes one or more cameras for capturing one or more images. In some examples, the device includes a display for displaying one or more images, notifications, and / or other displayable data. In some aspects, the device includes one or more sensors (e.g., one or more inertial measurement units (IMUs), such as one or more gyroscopes, one or more gyrometers, one or more accelerometers, any combination thereof, and / or other sensors). In some examples, the device includes a receiver, transmitter, or transceiver for receiving and / or sending information.

[0013] Aspects generally include methods, apparatus, systems, computer program products, non-transitory computer readable media, user equipment, base stations, wireless communication devices and / or processing systems as fully described herein with reference to and as illustrated by the accompanying drawings and description.

[0014] The features and technical advantages of examples according to the present disclosure have been outlined quite broadly above so that the following specific embodiments may be better understood. Additional features and advantages will be described below. The disclosed concepts and specific examples may be easily used as a basis for modifying or designing other structures for achieving the same purpose of the present disclosure. Such equivalent constructions do not depart from the scope of the appended claims. The characteristics of the concepts disclosed herein, both in terms of their organization and method of operation, and the associated advantages will be better understood by considering the following description in conjunction with the accompanying drawings. Each of the figures in the accompanying drawings is provided for the purpose of illustration and description and not as a definition of limitations of the claims.

[0015] Although various aspects are described in the present disclosure by illustrating some examples, it will be understood by those skilled in the art that such aspects can be implemented in many different arrangements and scenarios. The technology described herein can be implemented using different platform types, devices, systems, shapes, sizes and / or packaging arrangements. For example, some aspects can be implemented via integrated chip implementations or other devices based on non-module components (e.g., end-user devices, vehicles, communication equipment, computing equipment, industrial equipment, retail / shopping equipment, medical equipment and / or artificial intelligence devices). Various aspects can be implemented in chip-level components, modular components, non-modular components, non-chip-level components, device-level components and / or system-level components. The equipment incorporating the various aspects and features described may include additional components and features for implementing and practicing the various aspects claimed and described. For example, the transmission and reception of wireless signals may include one or more components for analog and digital purposes (e.g., hardware components, including antennas, radio frequency (RF) chains, power amplifiers, modulators, buffers, processors, interleavers, adders and / or summers). The various aspects described herein are intended to be practiced in various devices, components, systems, distributed arrangements and / or end-user devices of various sizes, shapes and configurations.

[0016] Other objects and advantages associated with the various aspects disclosed herein will be apparent to those skilled in the art based on the drawings and detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Examples of various specific implementations are described in detail below with reference to the following drawings:

[0018] Figure 1 is a block diagram illustrating an example of a wireless communication network according to some examples;

[0019] Figure 2 is a diagram illustrating a design of a base station and a user equipment (UE) device according to some examples, which enables sending and processing of signals exchanged between the UE and the base station;

[0020] Figure 3 is a diagram illustrating an example of a decomposed base station according to some examples;

[0021] Figure 4 is a block diagram illustrating components of user equipment according to some examples;

[0022] Figure 5 illustrates an example architecture of a neural network that may be used in accordance with some aspects of the present disclosure;

[0023] Figure 6 is a block diagram illustrating an ML engine according to aspects of the present disclosure;

[0024] Fig. 7A illustrates a block diagram associated with providing a scenario for testing a machine learning model on user equipment according to aspects of the present disclosure;

[0025] Figure 7B A UE-side model set for different scenarios and a NW-side model set for different scenarios according to various aspects of the present disclosure are illustrated;

[0026] Figure 7C Sending compressed data from different modes on a UE to a gNB to decompress the data via different modes thereof is illustrated in accordance with aspects of the present disclosure;

[0027] Fig.7D Sending compressed data from different models on a UE to a gNB to decompress the data via a common model in accordance with aspects of the present disclosure is illustrated;

[0028] FIG. 8A to FIG. 8B illustrating various flow charts associated with different aspects of testing the adequacy of a machine learning model in accordance with aspects of the present disclosure; and

[0029] Fig. 9 is a diagram illustrating an example of a system for implementing certain aspects of the present technology. DETAILED DESCRIPTION

[0030] Some aspects and embodiments of the present disclosure are provided below. Some of these aspects and embodiments can be applied independently, and some of them can be applied in combination, which is obvious to those skilled in the art. In the following description, specific details are set forth for explanation purposes in order to provide a thorough understanding of each embodiment of the application. However, it will be apparent that each embodiment can be put into practice without these specific details. Each drawing and description are not intended to be restrictive.

[0031] The following description only provides example embodiments and is not intended to limit the scope, applicability or configuration of the present disclosure. On the contrary, the following description of the exemplary embodiments will provide an enabling description for implementing the exemplary embodiments to those skilled in the art. It should be understood that various changes may be made to the function and arrangement of elements without departing from the essence and scope of the present application as set forth in the appended claims.

[0032] Wireless networks are deployed to provide various communication services, such as voice, video, packet data, messaging, broadcast, etc. Wireless networks may support access links for communication between wireless devices. An access link may refer to any communication link between a client device (e.g., user equipment (UE), station (STA), or other client device) and a base station (e.g., 3GPP gNodeB (gNB) for 5G / NR, 3GPP eNodeB (eNB) for LTE, Wi-Fi access point (AP), or other base station) or a decomposed base station (e.g., a central unit, a distributed unit, and / or a radio unit). In one example, the access link between the UE and the 3GPP gNB may be over the Uu interface. In some cases, the access link may support uplink signaling, downlink signaling, connection procedures, etc.

[0033] Various systems and techniques are provided to provide improvements to wireless communications with respect to wireless technologies (e.g., third generation partnership project (3GPP) 5G / new radio (NR) standards). A device (e.g., UE) may be configured to generate or determine control information related to a communication channel on which the device is communicating or is configured to communicate. For example, the UE may monitor the channel to determine information indicating the quality or state of the channel, which may be referred to as channel state information (CSI). In some cases, using an ML-based air interface, a first network device (e.g., UE) and a second network device (e.g., gNB) may use a trained ML model to implement functionality. For example, a UE that intends to deliver CSI to a gNB may use a neural network to derive a compressed representation of the CSI for transmission to the gNB. The gNB may use another neural network to reconstruct a target CSI from the compressed representation. In such cases, it is desirable to monitor the performance of the ML model at the UE and detect scenarios where the ML model performance is insufficient (e.g., when the reconstructed CSI is very different from the target CSI that the UE intends to deliver to the gNB).

[0034] In some cases, different ML models may be trained for different scenarios. For example, a first ML model may be trained using training data specific to an indoor environment to generate a compressed representation of control information (e.g., CSI), and a second ML model may be trained using training data specific to an outdoor environment to generate a compressed representation of control information (e.g., CSI). In another example, a first ML model may be trained using training data specific to a line-of-sight (LOS) scenario (e.g., without any obstructions, such as buildings) to generate a compressed representation of control information (e.g., CSI), and a first ML model may be trained using training data specific to a non-line-of-sight (NLOS) scenario to generate a compressed representation of control information (e.g., CSI). Other examples of different scenarios for which ML models may be trained include scenarios based on geographic location (e.g., region-specific), scenarios based on different serving cells, different categories of scenarios based on different statistics of a channel (e.g., delay spread, signal-to-noise ratio (SNR), ML-based features, etc.).

[0035] In such cases, if the device's ML model is trained using training data samples from one scenario, the ML model may perform poorly if used at inference or test time in a different scenario. For example, there may be a mismatch between the reconstructed control information generated from the compressed representation of the control information generated by the ML model and the target control information (e.g., CSI) that the device intended to convey. In an illustrative example, the ML model on the UE may be trained using training data from a first type of indoor environment, but when the test model is used at inference or test time, the UE may have moved to a different type of indoor environment or may have moved to an outdoor environment. In such examples, there may be a mismatch between the compressed representation of the control information output by the ML model compared to the target control information.

[0036] As described above, systems and techniques are described herein for monitoring the performance of one or more test ML models (e.g., test neural network models) using one or more reference machine learning (ML) models (e.g., reference neural network models) to identify situations where reconstructed control information is different from target control information intended to be transmitted by the device. The control information may include any type of control information or data that may need to be sent from a first network device to a second network device. One non-limiting example of control information is CSI. Another non-limiting example of control information is a reference signal, such as a demodulation reference signal (DMRS), a tracking reference signal (TRS), a positioning reference signal (PRS), a sounding reference signal (SRS), and / or other types of reference signals.

[0037] When a network device (e.g., a gNB) is determining whether reconstructed control information (e.g., CSI) received from another network device (e.g., a UE) is close to a target CSI, the network device needs to know the original target CSI originally determined by the other network device. The target control information may be considered to be "ground truth" or the actual condition of the channel. However, sending the target control information in its original form to the network device may require significant overhead and may reduce the benefits of using an ML model to compress the control information. According to various aspects described herein, a first network device may compress the ground truth control information using a test ML model to generate a first compressed representation of the control information, and may compress the ground truth control information using a reference ML model to generate a second compressed representation of the control information.

[0038] The performance of the test model can be monitored by determining the accuracy or adequacy of the test ML model based on the comparison between the first compressed representation and the second compressed representation, such as by comparing the compressed representation itself or by comparing corresponding reconstruction control information determined using the compressed representation. The monitoring or comparison described herein for determining the accuracy or adequacy of the test ML model can be performed by the first network device, by the second network device to which the compressed representation is sent by the first network device, or on both the first network device and the second network device.

[0039] According to some aspects, a test ML model deployed on a first network device can be monitored at a second network device based on data received at the second network device from the first network device (e.g., based on output generated at the first network device using the test ML model and output generated at the first network device using a reference ML model. For example, an encoder of the test ML model on the first network device can be trained to generate a compressed representation (e.g., a latent representation, such as a latent code) of control information associated with a communication channel (such as CSI or other control information indicating a quality or state of the communication channel). An encoder of the reference ML model on the first network device can also be trained to generate a compressed representation of control information (e.g., CSI) associated with the communication channel.

[0040] The first network device may send a compressed representation of the control information to the second network device. Upon receiving the two compressed representations of the control information, the second network device may use a decoder of the ML model deployed on the second network device to reconstruct the control information from each corresponding compressed representation. If the difference between the two versions of the reconstructed control information (e.g., the reconstructed CSI) is below a threshold difference, the second network device may determine that the performance of the test model deployed on the first network device is accurate for the communication channel. If the difference between the two versions of the reconstructed control information (e.g., the reconstructed CSI) is greater than (or not less than) the threshold difference, the second network device may determine that the performance of the test model deployed on the first network device is inaccurate for the communication channel.

[0041] The second network device may send information indicating a result of the comparison to the first network device (e.g., on a downlink channel such as a physical downlink control channel (PDCCH) or a physical downlink shared channel (PDSCH), on an uplink channel such as a physical uplink control channel (PUCCH) or a physical uplink shared channel (PUSCH), or on a sidelink channel such as a physical sidelink control channel (PSCCH) or a physical sidelink shared channel (PSSCH). Based on the received information, the first network device may update the test ML model (e.g., by retraining the test ML model), switch to a different ML model (e.g., an ML model trained for a different scenario), any combination thereof, and / or perform any other suitable operation.

[0042] In some cases, the ML model deployed on the first network device may be trained using training data including conditions for a specific scenario (e.g., an indoor environment, an outdoor environment, a specific cell, a specific geographic location, etc.), and the ML model deployed on the second network device may be trained using training data including conditions for multiple scenarios (e.g., for indoor and outdoor environments, for multiple cells, for multiple geographic locations, etc.). In such cases, the same ML model on the second network device may be compatible with multiple different scenario-specific ML models on the first network device. In some examples, the first network device may include ML models trained according to different scenarios. In some examples, the second network device may include an ML model trained for a specific scenario.

[0043] As described above, in some aspects, the first network device may monitor the performance of a test ML model deployed on the first network device. For example, the first network device may compare a first compressed representation with a second compressed representation to determine similarities or differences between the compressed representations. In another example, the first network device may reconstruct control information based on the first compressed representation (to generate a first reconstruction) and reconstruct control information based on the second compressed representation (to generate a second reconstruction). The first network device may compare the first reconstruction with the second reconstruction to determine similarities or differences between the first reconstruction and the second reconstruction. In such aspects, the first network device may send the results of the comparison to the second device, update the test ML model, switch to a different ML model (e.g., a model trained for a different scenario), any combination thereof, and / or perform any other suitable operation.

[0044] The first network device and the second network device may include any type of network device. For example, in some examples, the first network device may include a user equipment (UE) and the second network device may include a base station (e.g., an eNodeB, a gNodeB, or other base station) or a portion of a base station (e.g., a central unit (CU), a distributed unit (DU), a radio unit (RU), or other portion of a base station with a decomposed architecture). In some examples, the first network device may include a first UE, and the second network device may include a second UE. In some examples, the first network device may include a base station or a portion of a base station, and the second network device may include a UE.

[0045] One example of a benefit of the systems and techniques described herein is that the accuracy of an ML model deployed at a device can be monitored for a given channel. Another benefit is that the systems and techniques do not require raw control information (e.g., target or ground truth control CSI) to be sent from a first network device to a second network device for performance monitoring.

[0046] Further details regarding the systems and techniques described herein are provided with respect to the accompanying figures.

[0047] As used herein, the terms "user equipment" (UE) and "network entity" are not intended to be dedicated to or otherwise limited to any particular radio access technology (RAT), unless otherwise specified. In general, a UE can be any wireless communication device (e.g., a mobile phone, a router, a tablet computer, a laptop computer, and / or a tracking device, etc.), a wearable device (e.g., a smart watch, smart glasses, a wearable ring, and / or an extended reality (XR) device (such as a virtual reality (VR) headset, an augmented reality (AR) headset or glasses, or a mixed reality (MR) headset)), a vehicle (e.g., a car, a motorcycle, a bicycle, etc.), and / or an Internet of Things (IoT) device, etc., for a user to communicate on a wireless communication network. A UE can be mobile or can be stationary (e.g., at certain times) and can communicate with a radio access network (RAN). As used herein, the term "UE" may be interchangeably referred to as an "access terminal" or "AT," a "client device," a "wireless device," a "subscriber device," a "subscriber terminal," a "subscriber station," a "user terminal" or "UT," a "mobile device," a "mobile terminal," a "mobile station," or variations thereof. Generally speaking, a UE may communicate with a core network via a RAN, and through the core network, the UE may connect to external networks such as the Internet and to other UEs. Of course, other mechanisms for connecting to the core network and / or the Internet are also possible for the UE, such as through a wired access network, a wireless local area network (WLAN) network (e.g., based on IEEE 802.11 communication standards, etc.), and the like.

[0048] The network entity may be implemented in a converged or monolithic base station architecture, or alternatively, in a disaggregated base station architecture, and may include one or more of a central unit (CU), a distributed unit (DU), a radio unit (RU), a near real-time (near-RT) RAN intelligent controller (RIC), or a non-real-time (non-RT) RIC. A base station (e.g., having a converged / monolithic base station architecture or a disaggregated base station architecture) may operate in communication with a UE according to one of several RATs depending on the network in which the base station is deployed, and may alternatively be referred to as an access point (AP), a network node, a NodeB (NB), an evolved NodeB (eNB), a next generation eNB (ng-eNB), a new radio (NR) NodeB (also referred to as a gNB or gNodeB), etc. A base station may be primarily used to support wireless access for UEs, including supporting data, voice, and / or signaling connections for supported UEs. In some systems, a base station may provide edge node signaling functionality, while in other systems, a base station may provide additional control and / or network management functionality. The communication link through which the UE can transmit signals to the base station is called an uplink (UL) channel (e.g., a reverse traffic channel, a reverse control channel, an access channel, etc.). The communication link through which the base station can transmit signals to the UE is called a downlink (DL) or forward link channel (e.g., a paging channel, a control channel, a broadcast channel, or a forward traffic channel, etc.). As used herein, the term traffic channel (TCH) may refer to an uplink, a reverse or downlink, and / or a forward traffic channel.

[0049] The term "network entity" or "base station" (e.g., having an aggregated / monolithic base station architecture or a decomposed base station architecture) may refer to a single physical transmit receive point (TRP) or multiple physical TRPs that may or may not be co-located. For example, where the term "network entity" or "base station" refers to a single physical TRP, the physical TRP may be a base station antenna corresponding to a cell (or several cell sectors) of the base station. Where the term "network entity" or "base station" refers to multiple co-located physical TRPs, these physical TRPs may be antenna arrays of the base station (e.g., as in a multiple-input multiple-output (MIMO) system or where the base station employs beamforming). Where the term "base station" refers to multiple non-co-located physical TRPs, the physical TRP may be a distributed antenna system (DAS) (a network of spatially separated antennas connected to a common source via a transmission medium) or a remote radio head (RRH) (a remote base station connected to a serving base station). Alternatively, the non-co-located physical TRPs may be a serving base station that receives measurement reports from a UE and a neighbor base station whose reference radio frequency (RF) signal (or "reference signal" for short) the UE is measuring. Because, as used herein, a TRP is the point through which a base station transmits and receives wireless signals, references to transmitting from or receiving at a base station should be understood to refer to the specific TRP of a base station.

[0050] In some specific implementations of supporting UE positioning, a network entity or base station may not support wireless access of the UE (e.g., may not support data, voice, and / or signaling connections with respect to the UE), but may instead send a reference signal to be measured by the UE to the UE, and / or may receive and measure a signal sent by the UE. Such a base station may be referred to as a positioning beacon (e.g., in the case of sending a signal to the UE) and / or as a position measurement unit (e.g., in the case of receiving and measuring a signal from the UE).

[0051] An RF signal may include an electromagnetic wave of a given frequency that transmits information through the space between a transmitter and a receiver. As used herein, a transmitter may send a single "RF signal" or multiple "RF signals" to a receiver. However, due to the propagation characteristics of RF signals through multipath channels, a receiver may receive multiple "RF signals" corresponding to each transmitted RF signal. The same transmitted RF signal on different paths between a transmitter and a receiver may be referred to as a "multipath" RF signal. As used herein, an RF signal may also be referred to as a "wireless signal" or simply a "signal" where it is clear from the context that the term "signal" refers to a wireless signal or an RF signal.

[0052] Various aspects of the systems and techniques described herein are discussed below with respect to the accompanying drawings. According to various aspects, Figure 1An example of a wireless communication system 100 is illustrated. The wireless communication system 100 (which may also be referred to as a wireless wide area network (WWAN)) may include various base stations 102 and various UEs 104. In some aspects, the base station 102 may also be referred to as a "network entity" or a "network node". One or more of the base stations 102 may be implemented in an aggregated or monolithic base station architecture. Additionally or alternatively, one or more of the base stations 102 may be implemented in a decomposed base station architecture and may include one or more of a central unit (CU), a distributed unit (DU), a radio unit (RU), a near real-time (near RT) RAN intelligent controller (RIC), or a non-real-time (non-RT) RIC. The base station 102 may include a macro cell base station (a high-power cellular base station) and / or a small cell base station (a low-power cellular base station). In one aspect, the macro cell base station may include an eNB and / or an ng-eNB (where the wireless communication system 100 corresponds to a long-term evolution (LTE) network), or a gNB (where the wireless communication system 100 corresponds to an NR network), or a combination of both, and the small cell base station may include a femto cell, a pico cell, a micro cell, etc.

[0053] The base stations 102 may collectively form a RAN and interface with a core network 170 (e.g., an evolved packet core (EPC) or a 5G core (5GC)) through a backhaul link 122, and interface to one or more location servers 172 (which may be part of the core network 170 or may be external to the core network 170) through the core network 170. The base stations 102 may perform functions related to one or more of the following, among other functions: delivering 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, 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 (e.g., through the EPC or 5GC) through a backhaul link 134 (which may be wired and / 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. In one aspect, base station 102 in each coverage area 110 can support one or more cells. A "cell" is a logical communication entity used to communicate with a base station (e.g., on a certain frequency resource, referred to as a carrier frequency, component carrier, carrier, frequency band, etc.), and can be associated with an identifier (e.g., a physical cell identifier (PCI), a virtual cell identifier (VCI), a cell global identifier (CGI)) to distinguish cells operating via the same or different carrier frequencies. In some cases, different cells can be configured according to different protocol types (e.g., machine type communication (MTC), narrowband IoT (NB-IoT), enhanced mobile broadband (eMBB) or other protocol types) that can provide access to different types of UEs. Because a cell is supported by a specific base station, the term "cell" can refer to either or both of a logical communication entity and a base station supporting the logical communication entity depending on the context. In addition, since a TRP is typically a physical transmission point of a cell, the terms "cell" and "TRP" can be used interchangeably. In some cases, the term "cell" may also refer to a geographic coverage area (eg, a sector) of a base station, as long as a carrier frequency can be detected within some portion of the geographic coverage area 110 and is used for communications within that portion.

[0055] Although the geographic coverage areas 110 of neighboring macrocell base stations 102 may partially overlap (e.g., in a handover area), some areas of the geographic coverage areas 110 may substantially overlap with the larger geographic coverage areas 110. For example, a small cell base station 102' may have a coverage area 110' that substantially overlaps with the coverage areas 110 of one or more macrocell base stations 102. A network that includes both small cell base stations and macrocell base stations may be referred to as a heterogeneous network. A heterogeneous network may also include a home eNB (HeNB), which may provide services to a restricted group referred to as a closed subscriber group (CSG).

[0056] The communication link 120 between the base station 102 and the UE 104 may include uplink (also known as reverse link) transmissions from the UE 104 to the base station 102 and / or downlink (also known as forward link) transmissions from the base station 102 to the UE 104. The communication link 120 may use MIMO antenna technology, including spatial multiplexing, beamforming, and / or transmit diversity. The communication link 120 may be over one or more carrier frequencies. The allocation of carriers may be asymmetric for the downlink and uplink (e.g., more or fewer carriers may be allocated to the downlink than to the uplink).

[0057] The wireless communication system 100 may further include a WLAN AP 150 in communication with a WLAN station (STA) 152 via a communication link 154 in an unlicensed spectrum (e.g., 5 gigahertz (GHz)). When communicating in the unlicensed spectrum, the WLAN STA 152 and / or the WLAN AP 150 may perform a clear channel assessment (CCA) or a listen-before-talk (LBT) process before communicating to determine whether a channel is available. In some examples, the wireless communication system 100 may include a device (e.g., UE, etc.) that communicates with one or more UEs 104, base stations 102, APs 150, etc. using an ultra-wideband (UWB) spectrum. The UWB spectrum may range from 3.1 GHz to 10.5 GHz.

[0058] The small cell base station 102' can operate in a licensed and / or unlicensed spectrum. When operating in an unlicensed spectrum, the small cell base station 102' can adopt LTE or NR technology and use the same 5GHz unlicensed spectrum used by the WLAN AP 150. The small cell base station 102' adopting LTE and / or 5G in the unlicensed spectrum can boost the coverage of the access network and / or increase the capacity of the access network. NR in the unlicensed spectrum can be referred to as NR-U. LTE in the unlicensed spectrum can be referred to as LTE-U, License Assisted Access (LAA) or MulteFire.

[0059] The wireless communication system 100 may also include a millimeter wave (mmW) base station 180, which may operate at mmW frequencies and / or near mmW frequencies to communicate with UE 182. The mmW base station 180 may be implemented in an aggregated or monolithic base station architecture, or alternatively, in a decomposed base station architecture (e.g., including one or more of CU, DU, RU, near RT RIC, or non-RT RIC). Extremely high frequency (EHF) is part of RF in the electromagnetic spectrum. EHF has a range of 30 GHz to 300 GHz and a wavelength between 1 mm and 10 mm. The radio waves in this band may be referred to as millimeter waves. Near mmW may extend downward to a frequency of 3 GHz with a wavelength of 100 mm. The super high frequency (SHF) band extends between 3 GHz and 30 GHz, and is also referred to as centimeter waves. Communications using mmW and / or near mmW radio frequency bands have high path loss and relatively short distances. The mmW base station 180 and the UE 182 may utilize beamforming (transmit and / or receive) on the mmW communication link 184 to compensate for the extremely high path loss and short distance. In addition, it should be understood that in alternative configurations, one or more base stations 102 may also transmit using mmW or near-mmW and beamforming. Therefore, it should be understood that the foregoing illustrations are merely examples and should not be construed as limiting the various aspects disclosed herein.

[0060] In some aspects related to 5G, the spectrum in which a wireless network node or entity (e.g., base station 102 / 180, UE 104 / 182) operates is divided into multiple frequency ranges: FR1 (from 450 megahertz (MHz) to 6000 MHz), FR2 (from 24250 MHz to 52600 MHz), FR3 (above 52600 MHz), and FR4 (between FR1 and FR2). In a multi-carrier system such as 5G, one of the carrier frequencies is referred to as a "primary carrier" or "anchor carrier" or "primary serving cell" or "PCell", and the remaining carrier frequencies are referred to as "secondary carriers" or "secondary serving cells" or "SCells". In carrier aggregation, the anchor carrier is a carrier operating on the primary frequency (e.g., FR1) used by the UE 104 / 182 and the cell in which the UE 104 / 182 performs an initial radio resource control (RRC) connection establishment procedure or initiates an RRC connection reestablishment procedure. The primary carrier carries all common control channels as well as UE-specific control channels and can be a carrier in a licensed frequency (however, this is not always the case). A secondary carrier is a carrier operating on a second frequency (e.g., FR2) that can be configured and used to provide additional radio resources once an RRC connection is established between the UE 104 and the anchor carrier. In some cases, the secondary carrier may be a carrier in an unlicensed frequency. The secondary carrier may contain only necessary signaling information and signals, for example, since the primary uplink carrier and the primary downlink carrier are typically UE-specific, those signaling information and signals specific to the UE may not be present in the secondary carrier. This means that different UEs 104 / 182 in a cell may have different downlink primary carriers. This also holds true for the uplink primary carrier. The network is able to change the primary carrier of any UE 104 / 182 at any time. This is done, for example, to balance the load on different carriers. Because a "serving cell" (whether PCell or SCell) corresponds to a carrier frequency and / or component carrier through which some base stations are communicating, the terms "cell", "serving cell", "component carrier", "carrier frequency", etc. may be used interchangeably.

[0061] For example, still refer to Figure 1, one of the frequencies used by the macrocell base station 102 may be an anchor carrier (or "PCell"), and the other frequencies used by the macrocell base station 102 and / or the mmW base station 180 may be secondary carriers ("SCells"). In carrier aggregation, the base station 102 and / or the UE 104 may use a spectrum of up to Y MHz (e.g., 5 MHz, 10 MHz, 15 MHz, 20 MHz, 100 MHz) bandwidth per carrier, with up to a total of Yx MHz (x component carriers) in each direction for transmission. The component carriers may be adjacent to each other or may not be adjacent to each other in the spectrum. The allocation of carriers may be asymmetric with respect to the downlink and uplink (e.g., more or fewer carriers may be allocated to the downlink than to the uplink). The simultaneous transmission and / or reception of multiple carriers enables the UE 104 / 182 to significantly increase its data transmission and / or reception rate. For example, two 20 MHz aggregated carriers in a multi-carrier system would theoretically result in a doubling of the data rate (i.e., 40 MHz) compared to the data rate obtained with a single 20 MHz carrier.

[0062] In order to operate on multiple carrier frequencies, the base station 102 and / or the UE 104 may be equipped with multiple receivers and / or transmitters. For example, the UE 104 may have two receivers, namely "receiver 1" and "receiver 2", where "receiver 1" is a multi-band receiver that can be tuned to band (i.e., carrier frequency) "X" or band "Y", and "receiver 2" is a single-band receiver that can be tuned to only band "Z". In this example, if the UE 104 is being served in band "X", the band "X" will be referred to as the PCell or active carrier frequency, and "receiver 1" will need to tune from band "X" to band "Y" (SCell) to measure band "Y" (and vice versa). In contrast, regardless of whether the UE 104 is being served in band "X" or band "Y", due to the separate "receiver 2", the UE 104 can measure band "Z" without interrupting the service on band "X" or band "Y".

[0063] The wireless communication system 100 may further include a UE 164 that may communicate with the macrocell base station 102 over the communication link 120 and / or communicate with the mmW base station 180 over the mmW communication link 184. For example, the macrocell base station 102 may support a PCell and one or more SCells for the UE 164, and the mmW base station 180 may support one or more SCells for the UE 164.

[0064] The wireless communication system 100 may also include one or more UEs, such as UE 190, which are indirectly connected to one or more communication networks via one or more device-to-device (D2D) peer-to-peer (P2P) links (referred to as “side links”). Figure 1 In the example of FIG. 1 , UE 190 has a D2D P2P link 192 with one of UEs 104 connected to one of base stations 102 (e.g., UE 190 can indirectly obtain cellular connectivity through the D2D P2P link), and has a D2D P2P link 194 with WLAN STA 152 connected to WLAN AP 150 (UE 190 can indirectly obtain WLAN-based Internet connectivity through the D2D P2P link). In an example, D2D P2P links 192 and 194 can use any well-known D2D RAT (such as LTE Direct (LTE-D), WiFi Direct (WiFi-D), etc.) to support.

[0065] Figure 2 A block diagram of a design of a base station 102 and a UE 104 is shown that enables sending and processing of signals exchanged between the UE and the base station according to some aspects of the present disclosure. Design 200 includes components of the base station 102 and the UE 104, which may be Figure 1 1. A base station in base station 102 and a UE in UE 104. Base station 102 may be equipped with T antennas 234a through 234t, and UE 104 may be equipped with R antennas 252a through 252r, where in general T≥1 and R≥1.

[0066] At the base station 102, the transmit processor 220 may receive data for one or more UEs from the data source 212, select one or more modulation and coding schemes (MCS) for each UE based at least in part on a channel quality indicator (CQI) received from the UE, process (e.g., encode and modulate) the data for each UE based at least in part on the MCS selected for the UE, and provide data symbols for all UEs. The transmit processor 220 may also process system information (e.g., for semi-static resource partitioning information (SRPI), etc.) and control information (e.g., CQI requests, grants, upper layer signaling, channel state information, channel state feedback, etc.), and provide overhead symbols and control symbols. The transmit processor 220 may also generate reference symbols for reference signals (e.g., cell-specific reference signals (CRS)) and synchronization signals (e.g., primary synchronization signals (PSS) and secondary synchronization signals (SSS)). The transmit (TX) multiple-input multiple-output (MIMO) processor 230 may perform spatial processing (e.g., pre-decoding) on ​​data symbols, control symbols, overhead symbols, and / or reference symbols where applicable, and may provide T output symbol streams to T modulators (MOD) 232a to 232t. The modulators 232a to 232t are shown as combined modulator-demodulators (MOD-DEMOD). In some cases, the modulator and demodulator may be separate components. Each modulator in the modulators 232a to 232t may process a corresponding output symbol stream (e.g., for an orthogonal frequency division multiplexing (OFDM) scheme, etc.) to obtain an output sample stream. Each modulator in the modulators 232a to 232t may further process (e.g., convert to analog, amplify, filter, and up-convert) the output sample stream to obtain a downlink signal. The T downlink signals may be sent from the modulators 232a to 232t via T antennas 234a to 234t, respectively. According to certain aspects described in greater detail below, position encoding may be utilized to generate a synchronization signal to convey additional information.

[0067] At UE 104, antennas 252a to 252r may receive downlink signals from base station 102 and / or other base stations and may provide received signals to demodulators (DEMOD) 254a to 254r, respectively. Demodulators 254a to 254r are shown as combined modulator-demodulators (MOD-DEMOD). In some cases, modulators and demodulators may be separate components. Each demodulator in demodulators 254a to 254r may condition (e.g., filter, amplify, down-convert, and digitize) received signals to obtain input samples. Each demodulator in demodulators 254a to 254r may further process input samples (e.g., for OFDM, etc.) to obtain received symbols. MIMO detector 256 may obtain received symbols from all R demodulators 254a to 254r, perform MIMO detection on these received symbols where applicable, and provide detected symbols. The receive processor 258 may process (e.g., demodulate and decode) the detected symbols, provide decoded data for the UE 104 to a data sink 260, and provide decoded control information and system information to a controller / processor 280. The channel processor may determine a reference signal received power (RSRP), a received signal strength indicator (RSSI), a reference signal received quality (RSRQ), a channel quality indicator (CQI), etc.

[0068] On the uplink, at the UE 104, a transmit processor 264 may receive and process data from a data source 262 and control information from a controller / processor 280 (e.g., for reports including RSRP, RSSI, RSRQ, CQI, channel state information, channel feedback information, etc.). The transmit processor 264 may also generate reference symbols for one or more reference signals (e.g., based at least in part on a beta value or set of beta values ​​associated with the one or more reference signals). The symbols from the transmit processor 264 may be pre-decoded by the TX MIMO processor 266, if applicable, further processed by the modulators 254a through 254r (e.g., for DFT-s-OFDM, CP-OFDM, etc.), and transmitted to the base station 102. At the base station 102, uplink signals from the UE 104 and other UEs may be received by antennas 234a to 234t, processed by demodulators 232a to 232t, detected by a MIMO detector 236 where applicable, and further processed by a receive processor 238 to obtain decoded data and control information transmitted by the UE 104. The receive processor 238 may provide the decoded data to a data sink 239 and the decoded control information to a controller (processor) 240. The base station 102 may include a communication unit 244 and communicate with the network controller 231 via the communication unit 244. The network controller 231 may include a communication unit 294, a controller / processor 290, and a memory 292.

[0069] In some aspects, one or more components of the UE 104 may be included in a housing. The controller 240 of the base station 102, the controller / processor 280 of the UE 104, and / or Figure 2 Any other component of may perform one or more techniques associated with implicit UCI β value determination for NR.

[0070] Memories 242 and 282 may store data and program codes for base station 102 and UE 104, respectively. A scheduler 246 may schedule UEs for data transmission on the downlink, uplink, and / or sidelink.

[0071] In some aspects, the deployment of a communication system such as a 5G New Radio (NR) system can be arranged with various components or constituent parts 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 a network equipment (such as a base station (BS)) or one or more units (or one or more components) performing base station functions can be implemented in an aggregated or decomposed architecture. For example, a 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.

[0072] A converged base station may be configured to utilize a radio protocol stack physically or logically integrated within a single RAN node. A decomposed base station may be configured to utilize a protocol stack physically or logically distributed between two or more units, such as one or more central or centralized units (CUs), one or more distributed units (DUs), or one or more radio units (RUs). In some aspects, a CU may be implemented within a RAN node, and one or more DUs may be co-located with the CU, or alternatively, may be geographically or virtually distributed in one or more other RAN nodes. A DU may be implemented to communicate with one or more RUs. Each of a CU, a DU, and a RU may 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).

[0073] 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 used 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.

[0074] Figure 3A diagram illustrating an example decomposed base station 300 architecture is shown. The decomposed base station 300 architecture may include one or more central units (CUs) 310 that may communicate directly with a core network 320 via a backhaul link, or indirectly with the core network 320 through one or more decomposed base station units, such as a near real-time (near-RT) RAN intelligent controller (RIC) 325 via an E2 link, or a non-real-time (non-RT) RIC 315 associated with a service management and orchestration (SMO) framework 305, or both. The CU 310 may communicate with one or more distributed units (DUs) 330 via respective midhaul links, such as an F1 interface. The DU 330 may communicate with one or more radio units (RUs) 340 via respective fronthaul links. The RU 340 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 340.

[0075] Each of the units (e.g., CU 310, DU 330, RU 340, and near-RT RIC 325, non-RT RIC 315, and SMO framework 305) may include or be coupled to one or more interfaces configured to receive or send signals, data, or information (collectively referred to as signals) via a wired or wireless transmission medium. Each of these 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 a signal or send a signal to one or more of the other units via a wired transmission medium. Additionally, these units may include a wireless interface that may include a receiver, a transmitter, or a transceiver (such as a radio frequency (RF) transceiver) configured to receive a signal or send a signal to one or more of the other units via a wireless transmission medium, or both.

[0076] In some aspects, CU 310 may host one or more 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 with an interface configured to communicate signals with other control functions hosted by CU 310. CU 310 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 310 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 310 may be implemented to communicate with DU 330 for network control and signaling.

[0077] DU 330 may correspond to a logical unit that includes one or more base station functions for controlling the operation of one or more RU 340. In some aspects, DU 330 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.) depending at least in part on a functional split, such as that defined by the Third Generation Partnership Project (3GPP). In some aspects, DU 330 may also host one or more low PHY layers. Each layer (or module) may be implemented using an interface that is configured to communicate signals with other layers (and modules) hosted by DU 330 or with control functions hosted by CU 310.

[0078] The lower layer functionality may be implemented by one or more RUs 340. In some deployments, a RU 340 controlled by a DU 330 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 such an architecture, the RU 340 may be implemented to handle over-the-air (OTA) communications with one or more UEs 104. In some implementations, real-time and non-real-time aspects of control and user plane communications with the RU 340 may be controlled by the corresponding DU 330. In some scenarios, this configuration may enable the DU 330 and the CU 310 to be implemented in a cloud-based RAN architecture (such as a vRAN architecture).

[0079] The SMO framework 305 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 305 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 305 may be configured to interact with a cloud computing platform (such as an open cloud (O-Cloud) 390) to perform network element lifecycle management (such as to instantiate 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 310, DU 330, RU 340, and near-RT RIC 325. In some specific implementations, the SMO framework 305 may communicate with hardware aspects of the 4G RAN (such as an open eNB (O-eNB) 311) via the O1 interface. Additionally, in some specific implementations, the SMO framework 305 may communicate directly with one or more RUs 340 via the O1 interface. The SMO framework 305 may also include a non-RT RIC 315 configured to support the functionality of the SMO framework 305 .

[0080] The non-RT RIC 315 may be configured to include logic functions that enable 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 325. The non-RT RIC 315 may be coupled to or communicate with the near-RT RIC 325 (such as via an A1 interface). The near-RT RIC 325 may be configured to include logic functions that enable 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 310, one or more DUs 330, or both, and the O-eNB with the near-RT RIC 325.

[0081] In some implementations, in order to generate an AI / ML model to be deployed in the near-RT RIC 325, the non-RT RIC 315 may receive parameters or external enrichment information from an external server. Such information may be utilized by the near-RT RIC 325 and may be received from a non-network data source or from a network function at the SMO framework 305 or the non-RT RIC 315. In some examples, the non-RT RIC 315 or the near-RT RIC 325 may be configured to tune RAN behavior or performance. For example, the non-RT RIC 315 may monitor long-term trends and patterns of performance and employ AI / ML models to perform corrective actions through the SMO framework 305 (such as via reconfiguration of O1) or via the creation of RAN management policies (such as A1 policies).

[0082] Figure 4 An example of a computing system 470 of a wireless device 407 is illustrated. The wireless device 407 may include a client device such as a UE (e.g., UE 104, UE 152, UE 190) or other type of device that can be used by an end user (e.g., a station (STA) configured to communicate using a Wi-Fi interface). For example, the wireless device 407 may include a mobile phone, a router, a tablet computer, a laptop computer, a tracking device, a wearable device (e.g., a smart watch, glasses, an extended reality (XR) device (such as a virtual reality (VR), augmented reality (AR) or mixed reality (MR) device), etc.), an Internet of Things (IoT) device, an access point, and / or another device configured to communicate via a wireless communication network. The computing system 470 includes software and hardware components that can be electrically coupled or communicatively coupled via a bus 489 (or can communicate in other ways, as the case may be). For example, the computing system 470 includes one or more processors 484. One or more processors 484 may include one or more CPUs, ASICs, FPGAs, APs, GPUs, VPUs, NSPs, microcontrollers, dedicated hardware, any combination thereof, and / or other processing devices or systems. One or more processors 484 may use a bus 489 to communicate between cores and / or with one or more memory devices 486.

[0083] The computing system 470 may also include one or more memory devices 486, one or more digital signal processors (DSPs) 482, one or more subscriber identity modules (SIMs) 474, one or more modems 476, one or more wireless transceivers 478, one or more antennas 487, one or more input devices 472 (e.g., a camera, a mouse, a keyboard, a touch-sensitive screen, a touchpad, a keypad, a microphone, and / or the like), and one or more output devices 480 (e.g., a display, a speaker, a printer, and / or the like).

[0084] In some aspects, computing system 470 may include one or more RF interfaces configured to send and / or receive radio frequency (RF) signals. In some examples, the RF interface may include components such as modem 476, wireless transceiver 478, and / or antenna 487. One or more wireless transceivers 478 may send and receive wireless signals (e.g., signal 488) from one or more other devices via antenna 487, such as other wireless devices, network devices (e.g., base stations such as eNBs and / or gNBs, Wi-Fi access points (APs) such as routers, range extenders, etc.), cloud networks, etc. In some examples, computing system 470 may include multiple antennas or antenna arrays that can facilitate simultaneous transmission and reception functionality. Antenna 487 may be an omnidirectional antenna so that radio frequency (RF) signals can be received from all directions and transmitted in all directions. Wireless signal 488 may be transmitted via a wireless network. The wireless network may be any wireless network, such as a cellular or telecommunications network (e.g., 3G, 4G, 5G, etc.), a wireless local area network (e.g., a WiFi network), a Bluetooth® network, or a wireless communications network. TM network and / or other networks.

[0085] In some examples, wireless signals 488 may be transmitted directly to other wireless devices using sidelink communications (e.g., using a PC5 interface, using a DSRC interface, etc.). Wireless transceiver 478 may be configured to transmit RF signals via antenna 487 for performing sidelink communications according to one or more transmit power parameters that may be associated with one or more regulatory modes. Wireless transceiver 478 may also be configured to receive sidelink communication signals having different signal parameters from other wireless devices.

[0086] In some examples, one or more wireless transceivers 478 may include an RF front end that includes one or more components such as an amplifier, a mixer for down-converting a signal (also referred to as a signal multiplier), a frequency synthesizer (also referred to as an oscillator) that provides a signal to the mixer, a baseband filter, an analog-to-digital converter (ADC), one or more power amplifiers, and other components. The RF front end may generally handle the selection of wireless signals 488 and the conversion of wireless signals to baseband frequencies or intermediate frequencies, and may convert RF signals to the digital domain.

[0087] In some cases, computing system 470 may include a coding-decoding device (or CODEC) configured to encode and / or decode data transmitted and / or received using one or more wireless transceivers 478. In some cases, computing system 470 may include an encryption-decryption device or component configured to encrypt and / or decrypt data transmitted and / or received by one or more wireless transceivers 478 (e.g., according to the AES and / or DES standards).

[0088] One or more SIMs 474 may each securely store an International Mobile Subscriber Identity (IMSI) number and associated keys assigned to a user of the wireless device 407. The IMSI and keys may be used to identify and authenticate a subscriber when accessing a network provided by a network service provider or operator associated with the one or more SIMs 474. One or more modems 476 may modulate one or more signals to encode information for transmission using one or more wireless transceivers 478. One or more modems 476 may also demodulate signals received by one or more wireless transceivers 478 in order to decode the transmitted information. In some examples, one or more modems 476 may include a Wi-Fi modem, a 4G (or LTE) modem, a 5G (or NR) modem, and / or other types of modems. One or more modems 476 and one or more wireless transceivers 478 may be used to communicate data of one or more SIMs 474.

[0089] The computing system 470 may also include (and / or be in communication with) one or more non-transitory machine-readable storage media or storage devices (e.g., one or more memory devices 486), which may include, but are not limited to, local and / or network accessible storage, disk drives, drive arrays, optical storage devices, solid-state storage devices such as RAM and / or ROM, which may be programmable, flash-updatable, etc. Such storage devices may be configured to implement any suitable data storage, including, but not limited to, various file systems, database structures, etc.

[0090] In various embodiments, the functionality may be stored as one or more computer program products (e.g., instructions or code) in the memory device 486 and executed by the one or more processors 484 and / or the one or more DSPs 482. The computing system 470 may also include software elements (e.g., located within the one or more memory devices 486) including, for example, an operating system, device drivers, executable libraries, and / or other code, such as one or more applications, which may include computer programs that implement the functionality provided by various embodiments, and / or may be designed to implement methods and / or configure systems, as described herein.

[0091] Figure 5An example architecture of a neural network 500 that may be used according to some aspects of the present disclosure is illustrated. The example architecture of the neural network 500 may be defined by an example neural network description 502 in a neural controller 501. The neural network 500 is an example of a machine learning model that may be deployed and implemented at a base station 102, a central unit (CU) 310, a distributed unit (DU) 330, a radio unit (RU) 340, and / or a UE 104. The neural network 500 may be a feed-forward neural network or any other known or to be developed neural network or machine learning model.

[0092] Neural network description 502 may include a complete specification of neural network 500, including Figure 5 . For example, neural network description 502 may include: a description or specification of the architecture of neural network 500 (e.g., layers, layer interconnections, number of nodes in each layer, etc.); input and output descriptions indicating how inputs and outputs are formed or processed; indications of activation functions in the neural network, operations or filters in the neural network, etc.; neural network parameters such as weights, biases, etc.; and the like.

[0093] Neural network 500 may reflect the neural architecture defined in neural network description 502. Neural network 500 may include any suitable neural or deep learning type of network. In some cases, neural network 500 may include a feedforward neural network. In other cases, neural network 500 may include a recursive neural network, which may have loops that allow information to be carried across nodes when reading inputs. Neural network 500 may include any other suitable neural network or machine learning model. An example includes a convolutional neural network (CNN), which includes an input layer and an output layer, with multiple hidden layers between the input layer and the output layer. The hidden layers of the CNN include a series of hidden layers as described below, such as convolutional layers, nonlinear layers, pooling layers (for downsampling), and fully connected layers. In other examples, neural network 500 may represent any other neural network or deep learning network, such as an autoencoder, a deep belief network (DBN), a recursive neural network (RNN), a generative adversarial network (GAN), etc.

[0094] exist Figure 5In a non-limiting example, the neural network 500 includes an input layer 503 that can receive one or more input data sets. The input data can be any type of data (e.g., image data, video data, network parameter data, user data, etc.). The neural network 500 may include hidden layers 504A to 504N (hereinafter collectively referred to as "504"). The hidden layer 504 may include n number of hidden layers, where n is an integer greater than or equal to one. The n number of hidden layers may include as many layers as desired for processing results and / or presentation intent. In an illustrative example, any of the hidden layers 504 may include data representing one or more of the data provided at the input layer 503. The neural network 500 also includes an output layer 506 that provides an output resulting from the processing performed by the hidden layer 504. The output layer 506 may provide output data based on the input data.

[0095] exist Figure 5 In the example of , the neural network 500 is a multi-layer neural network of interconnected nodes. Each node can represent a piece of information. The information associated with these nodes is shared between different layers, and each layer retains the information when processing the information. Information can be exchanged between the nodes through node-to-node interconnections between the layers. The nodes of the input layer 503 can activate a set of nodes in the first hidden layer 504A. For example, as shown in the figure, each input node of the input layer 503 is connected to each node of the first hidden layer 504A. The nodes of the hidden layer 504A can transform the information by applying an activation function to the information of each input node. Then, the information derived from the transformation can be passed to the nodes of the next hidden layer (e.g., 504B) and these nodes can be activated, and these nodes can perform their own specified functions. Example functions include convolution, upsampling, data transformation, pooling and / or any other suitable function. The output of the hidden layer (e.g., 504B) can then activate the nodes of the next hidden layer (e.g., 504N), and so on. The output of the last hidden layer can activate one or more nodes of the output layer 506, providing an output at this point. In some cases, although nodes in neural network 500 (e.g., nodes 508A, 508B, 508C) are shown as having multiple output lines, the node can have a single output and all lines shown as output from the node can represent the same output value.

[0096] In some cases, each node or the interconnections between nodes may have a weight that is a set of parameters derived from training neural network 500. For example, an interconnection between nodes may represent a piece of information learned about the interconnected nodes. The interconnections may have numerical weights that may be tuned (e.g., based on a training data set), thereby allowing neural network 500 to adapt to inputs and learn as more data is processed.

[0097] The neural network 500 may be pre-trained to process features of the data from the input layer 503 using different hidden layers 504 in order to provide an output through the output layer 506. For example, in some cases, the neural network 500 may use a training process known as back propagation to adjust the weights of the nodes. Back propagation may include a forward pass, a loss function, a backward pass, and a weight update. The forward pass, loss function, backward pass, and parameter update may be performed for one training iteration. The process may be repeated for each training data set for a certain number of iterations until the weights of the layers are accurately tuned (e.g., to meet a configurable threshold determined based on experiments and / or empirical studies).

[0098] Increasingly, ML (eg, AI) algorithms (eg, models) are being incorporated into a variety of technologies including wireless telecommunications standards. For example, as described herein, systems and techniques for monitoring the performance of a test ML model using a reference ML model are described. Figure 6 600. As an example, one or more devices in a wireless system may include the ML engine 600. In some cases, the ML engine 600 may be similar to the neural network 500. In this example, the ML engine 600 receives and processes an input 602 to generate an output 604. The input 602 of the ML engine 600 may be data that the ML engine 600 may use to make predictions or otherwise operate on it. As an example, the ML engine 600 configured to select an RF beam may take as input 602 data about current RF conditions, location information, network load, etc. As another example, data related to packets transmitted to a UE along with historical packet data may be taken as input 602 to the ML engine 600 configured to predict a DRX schedule for the UE. In some cases, the output 604 may be a prediction or other information generated by the ML engine 600, and the output 604 may be used to configure the wireless device, adjust settings, parameters, operating modes, etc. Continuing with the previous example, the ML engine 600 configured to select an RF beam may output 604 an RF beam or a set of RF beams that may be used. Similarly, the ML engine 600 configured to predict a DRX schedule for a UE may output a DRX schedule for the UE. Also as described above, the ML engine 600 configured on the UE may be trained to compress control information (e.g., CSI) and send the compressed control information to another device (e.g., gNB, UE, etc.) over an air interface.

[0099] Fig. 7A is a diagram illustrating an example of a system 700 for implementing various aspects of monitoring the performance of a test ML model deployed on a network device. Fig. 7AAs shown, system 700 includes UE 701 and base station 703. Base station 703 may include a gNB, eNB, or other type of base station, or a portion of a base station (e.g., a CU, DU, RU, or other portion of a base station with a decomposed architecture). UE 701 may determine downlink channel estimate 702 (as an example of control information), such as based on one or more received CSI reference signals (CSI-RS) received from a network device (e.g., a gNB) (such as base station 703, another base station, or other network device). UE 701 may provide downlink channel estimate 702 to a channel state information (CSI) encoder 704.

[0100] The CSI encoder 704 may include at least one reference ML model and at least one test ML model. Although the examples described herein use one reference ML model and one test ML model for illustrative purposes, it will be appreciated by those of ordinary skill that the UE and / or other network devices may include multiple reference ML models and / or multiple test ML models. Each of the test ML model and the reference ML model of the CSI encoder 704 may encode the downlink channel estimate 702 (e.g., CSI) to generate a corresponding encoded or compressed representation of the downlink channel estimate 702. The encoded / compressed representation of the downlink channel estimate 702 may include a potential representation (e.g., a potential code) of the downlink channel estimate 702 (e.g., a potential code representing the CSI). For example, the potential representation may include a feature vector, a tensor, an array, or other representation including values ​​representing the downlink channel estimate 702.

[0101] The UE 701 may use antenna 708 to send the coded downlink channel estimate to the receiving antenna 712 of the base station 703 via a data or control channel 706 over a wireless or air interface 710. The coded or compressed downlink channel estimate 702 is provided to the CSI decoder 716 of the base station 703 via a data or control channel 714. The CSI decoder 716 may decode the coded downlink channel estimate to generate a reconstructed downlink channel estimate 718. The decoder 716 may also include a reference ML model and a test ML model. The encoder 704 may be on the UE, and the decoder may be on the base station (e.g., gNB) or a portion of the base station (e.g., CU, DU, RU, etc.). The encoder output from the UE is sent to the base station as an input to the decoder 716. In one example, the encoder at the UE outputs a compressed channel state feedback (CSF), which is input to the decoder at the base station. The decoder at the base station outputs a reconstructed CSF, such as a precoding vector.

[0102] Thus, aspects disclosed herein may be systems or methods associated with a gNB server, a gNB that receives and performs comparative analysis to monitor the performance of a test ML model, a UE server and / or a UE that may also receive data associated with the performance of the test ML model for monitoring purposes. When the test ML model does not perform correctly (as determined by monitoring), remedial steps may be taken, such as switching to a different model or transmitting full uncompressed data (e.g., such as raw CSI or CSF).

[0103] Figure 7B In more detail, a system 730 is illustrated that includes a UE 732 on which a UE-side model 736 trained for a first scenario referred to as scenario 1 (e.g., using training data specific to the first scenario) and another UE-side model 738 trained for one or more scenarios, such as scenarios labeled as scenario 2 and scenario 3, has been configured. As described above, different scenarios may include indoor use, outdoor use, line-of-sight (LOS) use, non-LOS use, UEs from a first UE vendor versus UEs from a second UE vendor, region-specific geographic locations, serving cell characteristics, abstract scenario categories based on different statistics of the channel, such as delay spread, signal-to-noise ratio, ML-based features, etc. If an ML model, such as model 738, is trained for a specific scenario (e.g., an indoor environment), the model may perform poorly in a different scenario (e.g., an outdoor environment).

[0104] On the network side, gNB 734 may have a network side model 740 trained for scenarios 1 and 2 and another network side model 742 trained for scenario 3. By training network side model 740 according to scenarios 1 and 2, model 740 may be compatible with many different scenario specific UE side models. For example, model 704 may allow gNB 734 to accurately determine whether model 736 and / or model 738 are performing accurately for a given channel for which channel estimate 702 was determined.

[0105] Figure 7C A system 746 including a UE 732 and a network device 734 (e.g., a base station or a portion thereof) is illustrated. The UE 732 includes a UE-side test model 748 that generates compressed control information (such as CSI feedback 750) and sends the CSI feedback 750 to the network device 734. The network device 734 may use the network-side test model 752 to decode the compressed CSI feedback 750. The output of the network-side test model 752 is a decoded image corresponding to the original (target) CSI feedback (e.g., from Fig. 7A The first version of the reconstructed CSI feedback of the channel estimate 702). Figure 7CAs further illustrated in FIG. 1 , UE-side reference model 754 generates compressed control information (such as CSI feedback 755) and sends CSI feedback 755 data to a network node, which decodes the compressed CSI feedback 855 using network-side reference model 756. The output from network-side reference model 756 is a CSI feedback corresponding to the original (target) CSI feedback (e.g., from Fig. 7A A second version of the reconstructed CSI feedback of the channel estimate 702).

[0106] The network device 734 may then compare (at operation 758) the two versions of the reconstructed CSI feedback to determine whether the CSI feedback is similar (e.g., within a threshold difference). The closer the two reconstructions are, the better the performance of the tested UE-side model 748. The result of the comparison operation may be a determination or comparison value that may indicate whether the UE-side test model 748 is accurate for the communication channel for which the original (target) CSI feedback was determined. For example, if the difference between the first version of the reconstructed CSI feedback and the second version of the reconstructed CSI feedback is within a threshold difference, the network device 734 (or UE 732) may determine that the performance of the UE-side test model 748 is accurate for the communication channel. If the difference between the two versions of the reconstructed control information (e.g., the reconstructed CSI) is greater than (or not less than) the threshold difference, the network device 734 (or UE 732) may determine that the performance of the UE-side test model 748 is not accurate for the communication channel.

[0107] The network device 734 may send information indicating the result of the comparison to the UE 732. For example, if the comparison value indicates that the test model 748 is performing accurately for the communication channel, the UE 732 may continue to use the test model 748 to compress the additional control information. However, if the model 748 is not performing accurately for the communication channel, the UE 732 may switch to another ML model (e.g., which may be trained for a different scenario) to compress the additional control data, may further train the test model 748 using the additional training data, may send control information (e.g., CSI feedback) without compression, and / or perform one or more other operations.

[0108] The method of monitoring the performance of the test model 748 may occur on one or both of the UE 732 and the gNB 734. For example, the gNB 734 may perform the comparison and / or also monitor the performance of the test ML model 748. In another aspect, the reconstructed data may be sent back to the UE 732 to perform the comparison step and take further action based on the results of the comparison.

[0109] In one aspect, the UE 732 generates two compressed representations 750, 755 and sends the two compressed representations to the gNB 734, and the comparison is performed after reconstruction on the gNB 734. However, other variations may also be applied. The reconstructed data of both the network-side test model 752 and the network-side reference model 756 may be sent back to the UE 732, and the comparison 758 may occur on the UE 732. In that case, the monitoring or determination of the performance of the test model 748 is actually performed on the UE 732.

[0110] In general, information may be sent from the gNB 734 to the UE 732 based on a comparison performed at the gNB 734 associated with receiving a first compressed representation of the control information and a second compressed representation of the control information. In a more specific example, the comparison may be a comparison of a first reconstruction of the control information based on the first compressed representation of the control information and a second reconstruction of the control information based on the second compressed representation of the control information.

[0111] In one aspect, the UE 732 may provide metadata, parameters, or indications along with the transmission of the compressed representations 750, 755 indicating that different data will be compared to monitor the performance of the test model 748. The indication may be based on initial data known on the UE 732, such as how comparable the CSI feedback 750, 755 are to each other. The indication may also be provided based on how different the first environment is from the second environment as the UE 732 moves from one environment to another. For example, a large change in the characteristics of the different environments may trigger an indicator to monitor the performance of the test UE side model 748.

[0112] A well-trained model 748 will ensure that typical realizations of the CSI (i.e., common realizations that are not outliers or uncommon cases) can be reconstructed well and will indicate adequate performance of the model 748. In contrast, realizations that are out of distribution of the training data will result in inaccurate reconstructions, which are shown in the comparison results 758. In this sense, the goal of model monitoring is to determine whether a realization is out of distribution (OOD).

[0113] As an example, the model M_indoor 748 may be trained based on data collected in indoor situations, such as at home, in an office, at a stadium, or other indoor scenes. The M_reference model 766 may be a model trained on a mixture of indoor and outdoor datasets, and may also allow for a larger size of compressed representation. In this case, if the UE 732 is currently using the M_indoor 748, but the UE 732 moves outdoors, the M_indoor model 748 will experience an OOD situation because the data is from an outdoor scene, and this may result in poor performance.

[0114] The M_reference model 754 trained on a mixture of indoor and outdoor data may produce a non-OOD implementation, and the reconstructed CSI may be similar to the target CSI. The gNB 734 may then detect whether the implementation is OOD by comparing 758 the CSI reconstructed from M_ref 754 and M_indoor 748.

[0115] Fig.7D A system 760 is illustrated in which a UE 732 includes a UE-side test model 762 that generates a first CSI feedback 764 for transmission to a gNB 734. The UE 734 also includes a UE-side reference model 766 that generates a second CSI feedback 768 that is transmitted to the gNB 734. The gNB reconstructs the first CSI feedback 764 using a generic network-side model 770, then reconstructs the second CSI feedback 768 using the same generic network-side model 770, and compares 772 the two reconstructions to determine whether the test UE-side model 762 performs adequately.

[0116] The network may configure the test models 748, 762 for inference purposes and the reference models 754, 766 for model monitoring purposes. In this scenario, the network may trigger the UE 732 to use the test model 762 for inference. Additionally, the network may trigger the UE 732 to use the reference models 754, 766 for inference. In this case, inference using the reference model 766 may be triggered less frequently than using the test models 748, 762 to save computational requirements at the UE 732. The reference model may also be associated with more relaxed processing time requirements relative to the test model. The triggering of which model to use may be based on one or more factors, such as the location of the UE 732, the characteristics of the environment around the UE 732 (i.e., indoors or outdoors, etc.), the moving speed of the UE 732, whether the UE is in a vehicle and / or the type of vehicle (such as a train, airplane, or car), etc. There may be any number of triggering events that may cause the UE 732 to switch to one model or another, and monitor whether the selected model performs adequately.

[0117] In one aspect, instructions for monitoring the performance of the test models 748, 762 may occur in a periodic manner, such as every hour, every day, or any periodic time frame. The timing may also be based on a predicted schedule, such as the time that the UE 732 arrives at an office or a certain location every day. For example, the UE 732 or other device may cause performance monitoring when the user with the UE 732 arrives at work or goes home every day and according to the predetermined time or location of the UE 732. The user may manually request performance monitoring, or a graphical user interface may be presented to the user to confirm the location or environmental conditions in order to test the performance of the ML model. The machine learning model may also be implemented to predict or infer whether changes to the AI / ML model 748 are needed or expected based on the historical activities of the UE 732 or the trained activities or movements. On the other hand, monitoring may occur based on some other event, such as throughput degradation, high block error rate.

[0118] Fig.7D The example of only illustrates the monitoring of the UE-side test model 762 based on the network-side model 770 being trained using training data specific to a wide range of scenarios. In such examples, a reference model is only needed on the UE side, and the network-side model 770 can be used on the network side (without being monitored by the reference model) because it is known to be universal for a wide range of scenarios. In some aspects, as described herein, a system (e.g., system 746, system 760, or other system) may monitor the end-to-end performance of both the UE-side model and the network-side model (where both the UE-side model and the NW-side model are under test). In such cases, the reference model may include a UE-side reference model and a NW-side reference model. In other cases, the UE may have a UE-side model trained using training data specific to multiple scenarios (e.g., indoor environments and outdoor environments), and the base station may include a network-side test model tested using a reference model on the base station.

[0119] The information associated with the performance of the tested machine learning model may indicate that the tested machine learning model is inaccurate for the communication channel. The UE may then take remedial steps based on the information, such as switching to an alternative machine learning model for further communication with the device or an additional device. The UE may send uncompressed control information, which may occupy more bandwidth but may be more accurate. The information associated with the performance of the tested machine learning model may indicate that the tested machine learning model is accurate for the communication channel. The UE may then continue to use the same model for further communication.

[0120] Fig. 8A800 for performing wireless communications at a UE. At block 802, process 800 includes using a machine learning model under test (referred to as a test machine learning model) to generate a first representation of control information associated with a communication channel. In some examples, the control information may include channel state information (CSI) or channel state feedback (CSF) associated with the communication channel. In other examples, the control information may include other types of data in addition to CSI or CSF. In some aspects, the machine learning model may generate a representation of the control information based on a rate-distortion tradeoff that provides a tradeoff between the size of the representation and the accuracy of a reconstructed version of the control information (e.g., the distortion between the reconstructed control information and the original control information). For example, in some cases, the goal may be to compress (reduce the size of) the control information (e.g., a CSI or CSF message), in which case the first representation may include a first compressed representation of the control information. In other cases, the goal may be to improve the accuracy of the reconstructed control information (e.g., the reconstructed CSI or CSF).

[0121] At block 804, process 800 includes using a reference machine learning model to generate a second representation of control information associated with a communication channel. In some cases, the second representation may include a second compressed representation of the control information. In some aspects, the test machine learning model may include a first encoder neural network model (e.g., UE-side test model 748, UE-side test model 762, or other model) that is trained to compress control information of a first environment into a compressed representation. In some aspects, the reference machine learning model may include a second encoder neural network model (e.g., UE-side reference model 754, UE-side reference model 766, or other model) that is trained to compress control information of a first environment and a second environment into a compressed representation. In one illustrative example, the first environment may include an indoor environment, and the second environment may include an outdoor environment. Many other environments are also contemplated.

[0122] At block 806, process 800 includes sending information associated with a comparison based on a first representation of control information and a second representation of control information to a device. The device may perform a comparison of the first representation of control information and the second representation of control information to monitor the performance of the test machine learning model 748, 762. In some cases, the device may perform the comparison on the original representation. In other cases, when the representation is compressed (referred to herein as a compressed representation), the device may use a first decoder to process the first representation of control information to generate a first reconstructed representation of the control information, and may use a second decoder to process the second representation of control information to generate a second reconstructed representation of the control information. In such cases, the device may then compare the first reconstructed representation of the control information with the second reconstructed representation of the control information to produce a comparison.

[0123] In some cases, the device may also perform a comparison and information associated with the comparison may be sent to the device as a result of the comparison.

[0124] In some aspects, such as at box 808, process 800 may include receiving information associated with the performance of the tested machine learning model from the device. In some examples, the information associated with the comparison of the first reconstruction of the control information and the second reconstruction of the control information may indicate that the tested machine learning model is inaccurate for the communication channel. In such examples, process 800 may also include switching to an alternative machine learning model to further communicate with the device or an additional device based on the information indicating that the tested machine learning model is inaccurate for the communication channel. In some examples, the information associated with the comparison of the first representation of the control information and the second representation of the control information indicates that the tested machine learning model is accurate for the communication channel. In such examples, process 800 may also include continuing to use the tested machine learning model based on the information indicating that the tested machine learning model is accurate for the communication channel.

[0125] On the other hand, the method may include: updating the machine learning model under test based at least in part on the information to generate an updated machine learning model; or receiving information from the device including a first trigger for using the machine learning model under test and generating a first representation of control information using the machine learning model under test based on the first trigger.

[0126] In some aspects, process 800 may include: receiving information from the device including a second trigger to use the reference machine learning model; and generating a second representation of control information using the reference machine learning model based on the second trigger. In some cases, use of the machine learning model under test may be triggered more frequently than use of the reference machine learning model.

[0127] In some aspects, at least one of the use of the tested machine learning model or the use of the reference machine learning model may be triggered based on an event. The event may include at least one of: the UE moving to a new environment for which the test machine learning model was not trained, a degradation of throughput via a communication channel, a high block error rate (BLER) condition, or a periodic time to monitor the performance of the tested machine learning model. The tested machine learning model may be configured on a UE or a device or a base station.

[0128] Figure 8BA process 820 for wireless communication at a first device is illustrated. In some cases, the first network device may be a gNB 734 or other network device. At block 822, the process 820 may include receiving from a second device a first representation of control information associated with a communication channel generated using a first tested machine learning model. The second network device may be a UE 732 or other device. At block 824, the process 820 may include receiving from the second device a second compressed representation of control information associated with a communication channel generated using a first reference machine learning model.

[0129] At block 826, process 820 may include reconstructing the control information from the first representation of the control information using the second test machine learning model to generate a first reconstruction of the control information. At block 828, process 820 may include reconstructing the control information from the second representation of the control information using the second reference machine learning model to generate a second reconstruction of the control information. At block 830, process 820 may include determining an accuracy of the first tested machine learning model for the communication channel based on a comparison of the first reconstruction of the control information and the second reconstruction of the control information (830). The representation of the control information may be compressed at the first device. The first device and the second device may be a UE or a base station or a gNB.

[0130] Fig. 9 is a diagram illustrating an example of a system for implementing certain aspects of the present technology. Specifically, Fig. 9 An example of a computing system 900 is illustrated, which may be any computing device, for example, constituting an internal computing system, a remote computing system, a camera, or any components thereof, wherein the components of the system communicate with each other using a connection 905. The connection 905 may be a physical connection using a bus, or a direct connection into the processor 910, such as in a chipset architecture. The connection 905 may also be a virtual connection, a networked connection, or a logical connection.

[0131] In some embodiments, computing system 900 is a distributed system, where the functionality described in the present disclosure may be distributed within a data center, multiple data centers, a peer-to-peer network, etc. In some embodiments, one or more of the described system components represent a number of such components that each perform some or all of the functionality for which the component is described. In some embodiments, a component may be a physical device or a virtual device.

[0132] The example system 900 includes at least one processing unit (CPU or processor) 910 and connections 905 that communicatively couple various system components including system memory 915, such as read only memory (ROM) 920 and random access memory (RAM) 925, to the processor 910. The computing system 900 may include a cache 912 of high-speed memory directly connected to the processor 910, in close proximity to the processor, or integrated as part of the processor.

[0133] Processor 910 may include any general purpose processor and hardware or software services such as services 932, 934, and 936 stored in storage device 930 configured to control processor 910, as well as special purpose processors where software instructions are incorporated into the actual processor design. Processor 910 may essentially be a completely independent computing system containing multiple cores or processors, buses, memory controllers, caches, etc. Multi-core processors may be symmetric or asymmetric.

[0134] To enable user interaction, the computing system 900 includes an input device 945 that can represent any number of input mechanisms, such as a microphone for voice, a touch-sensitive screen for gesture or graphical input, a keyboard, a mouse, motion input, voice, etc. The computing system 900 may also include an output device 935, which may be one or more of a plurality of output mechanisms. In some cases, a multimodal system may enable a user to provide multiple types of input / output to communicate with the computing system 900.

[0135] The computing system 900 may include a communication interface 940, which may generally govern and manage user input and system output. The communication interface may perform or facilitate receiving and / or sending wired or wireless communications using wired and / or wireless transceivers, including using an audio jack / plug, a microphone jack / plug, a Universal Serial Bus (USB) port / plug, an Apple TM Lightning TM Ports / plugs, Ethernet ports / plugs, Fiber optic ports / plugs, Dedicated wired ports / plugs, 3G, 4G, 5G and / or other cellular data network wireless signal transmission, Bluetooth TM Wireless signal transmission, Bluetooth TM Low energy (BLE) wireless signal transmission, IBEACON TMThe communication interface 940 may also include one or more global navigation satellite system (GNSS) receivers or transceivers for determining the location of the computing system 900 based on receiving one or more signals from one or more satellites associated with one or more GNSS systems. GNSS systems include, but are not limited to, the United States' Global Positioning System (GPS), Russia's Global Navigation Satellite System (GLONASS), China's BeiDou Navigation Satellite System (BDS), and Europe's Galileo GNSS. There is no restriction to operating on any particular hardware arrangement, and thus the base features herein may be easily substituted for improved hardware or firmware arrangements as they are developed.

[0136] The storage device 930 may be a non-volatile and / or non-transitory and / or computer-readable memory device and may be a hard disk or other type of computer-readable medium that can store data accessible by a computer, such as a magnetic cassette, a flash memory card, a solid-state memory device, a digital versatile disk, a cassette, a floppy disk, a floppy disk, a hard disk, a magnetic tape, a magnetic stripe / magnetic stripe, any other magnetic storage medium, flash memory, a memristor memory, any other solid-state memory, a compact disk-read only memory (CD-ROM) optical disk, a rewritable compact disk (CD) optical disk, a digital video disk (DVD) optical disk, a Blu-ray disc (BDD) optical disk, a holographic optical disk, another optical medium, a secure digital (SD) card, a micro secure digital (microSD) card, a Memory Card, or a memory card. card, a smart card chip, an EMV chip, a subscriber identity module (SIM) card, a mini / micro / nano / pico SIM card, another integrated circuit (IC) chip / card, a random access memory (RAM), a static RAM (SRAM), a dynamic RAM (DRAM), a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a flash EPROM (FLASH EPROM), a cache memory (e.g., a layer 1 (L1) cache, a layer 2 (L2) cache, a layer 3 (L3) cache, a layer 4 (L4) cache, a layer 5 (L5) cache, other (L#) cache), a resistive random access memory (RRAM / ReRAM), a phase change memory (PCM), a spin transfer torque RAM (STT-RAM), another memory chip or box, and / or a combination thereof.

[0137] Storage device 930 may include software services, servers, services, etc., which, when the code defining such software is executed by processor 910, causes the system to perform functions. In some embodiments, hardware services that perform specific functions may include software components for performing functions stored in a computer-readable medium connected to necessary hardware components such as processor 910, connection 905, output device 935, etc. The term "computer-readable medium" includes, but is not limited to, portable or non-portable storage devices, optical storage devices, and various other media capable of storing, containing or carrying instructions and / or data. Computer-readable media may include non-transient media in which data may be stored and does not include carrier waves and / or transient electronic signals propagated wirelessly or through wired connections. Examples of non-transient media may include, but are not limited to, disks or tapes, optical storage media (such as compact discs (CDs) or digital versatile discs (DVDs)), flash memory, memory, or memory devices. Computer readable media may store thereon code and / or machine executable instructions, which may represent a process, function, subprogram, program, routine, subroutine, module, software package, category, or any combination of instructions, data structures, or program statements. A code segment may be coupled to another code segment or a hardware circuit by transmitting and / or receiving information, data, independent variables, parameters, or memory contents. Information, independent variables, parameters, data, etc. may be transmitted, forwarded, or sent via any suitable means, including memory sharing, message passing, token passing, network sending, etc.

[0138] Specific details are provided in the above description to provide a thorough understanding of each embodiment and each example provided herein, but those skilled in the art will recognize that the application is not limited thereto. Thus, although the exemplary embodiments of the present application have been described in detail herein, it is to be understood that the inventive concept can be embodied and adopted in various other ways, and the appended claims are intended to be interpreted as including such variations, unless limited by the prior art. Various features and aspects of the above-mentioned applications can be used individually or in combination. In addition, without departing from the broader scope of this specification, the embodiments can be used in any number of environments and applications beyond the environment and application described herein. Therefore, the description and the accompanying drawings should be considered as illustrative rather than restrictive. For the purpose of illustration, each method is described in a specific order. It should be understood that in an alternative embodiment, each method can be performed in a different order than described.

[0139] For clarity of explanation, in some cases, the present technology may be presented as including separate functional blocks, which include devices, device components, steps or routines in the method embodied in software or a combination of hardware and software. Additional components other than those components shown in the drawings and / or described herein may be used. For example, circuits, systems, networks, processes and other components may be shown as components in block diagram form to avoid burying these embodiments in unnecessary details. In other cases, known circuits, processes, algorithms, structures and techniques may be shown without necessary details to avoid confusing each embodiment.

[0140] In addition, it will be appreciated by those skilled in the art that the various exemplary logic blocks, modules, circuits and algorithmic steps described in conjunction with the various aspects disclosed herein can be implemented as electronic hardware, computer software or a combination of the two. In order to clearly illustrate this interchangeability of hardware and software, various exemplary components, blocks, modules, circuits and steps have been generally described above in terms of their functionality. Whether such functionality is implemented as hardware or software depends on the specific application and the design constraints proposed to the entire system. The technician can implement the described functionality in different ways for each specific application, but such specific implementation decisions should not be interpreted as causing a departure from the scope of the present disclosure.

[0141] Individual embodiments may be described above as processes or methods depicted as flow charts, flow diagrams, data flow diagrams, structure diagrams, or block diagrams. Although a flow chart may describe an operation as a sequential process, many of the operations may be performed in parallel or concurrently. In addition, the order of the operations may be rearranged. The process is terminated when the operations of the process are completed, but the process may have additional steps not included in the accompanying drawings. A process may correspond to a method, function, procedure, subroutine, subprogram, etc. When a process corresponds to a function, the termination of the process may correspond to the function returning to the calling function or main function.

[0142] The processes and methods according to the above examples can be implemented using stored computer executable instructions or otherwise available computer executable instructions from a computer readable medium. Such instructions may include, for example, instructions and data that configure a general-purpose computer, a special-purpose computer, or a processing device to perform a certain function or group of functions. Portions of the computer resources used may be accessed via a network. Computer executable instructions may be, for example, binary, intermediate format instructions such as assembly language, firmware, source code. Examples of computer readable media that can be used to store instructions, information used, and / or information created during the methods according to the described examples include disks or optical disks, flash memory, USB devices with non-volatile memory, networked storage devices, etc.

[0143] In some embodiments, computer-readable storage devices, media, and memories may include wired or wireless signals containing bit streams, etc. However, when referred to, non-transitory computer-readable storage media explicitly exclude media such as energy, carrier signals, electromagnetic waves, and signals themselves.

[0144] Those skilled in the art will appreciate that information and signals may be represented using any of a variety of different techniques and methods. For example, data, instructions, commands, information, signals, bits, symbols, and chips that may be mentioned in the above description may be represented by voltage, current, electromagnetic waves, magnetic fields or magnetic particles, optical fields or optical particles, or any combination thereof, depending in some cases on the specific application, the desired design, the corresponding technology, etc.

[0145] The various illustrative logic blocks, modules, and circuits described in conjunction with the various aspects disclosed herein may be implemented or executed using hardware, software, firmware, middleware, microcode, hardware description language, or any combination thereof, and may be implemented in any of a variety of form factors. When implemented in software, firmware, middleware, or microcode, program code or code segments (e.g., computer program products) for performing the necessary tasks may be stored in a computer-readable or machine-readable medium. The processor may perform the necessary tasks. Examples of form factors include: laptops, smartphones, mobile phones, tablet devices, or other small form factor personal computers, personal digital assistants, rack-mounted devices, stand-alone devices, etc. The functions described herein may also be embodied in peripheral devices or add-in cards. By way of further example, such functionality may also be implemented on circuit boards in different chips or different processes executed on a single device.

[0146] The instructions, the media for conveying such instructions, the computing resources for executing them, and other structures for supporting such computing resources are example means for providing the functionality described in this disclosure.

[0147] The techniques described herein may also be implemented in electronic hardware, computer software, firmware, or any combination thereof. Such techniques may be implemented in any of a variety of devices, such as general-purpose computers, wireless communication device handsets, or integrated circuit devices with multiple uses, including applications in wireless communication device handsets and other devices. Any features described as modules or components may be implemented together in an integrated logic device or separately as discrete but interoperable logic devices. If implemented in software, the techniques may be implemented at least in part by a computer-readable data storage medium including a program code, which includes instructions for executing one or more of the methods, algorithms, and / or operations described above when executed. The computer-readable data storage medium may form part of a computer program product, which may include packaging materials. The computer-readable medium may include a memory or data storage medium, such as a random access memory (RAM) (such as a synchronous dynamic random access memory (SDRAM)), a read-only memory (ROM), a non-volatile random access memory (NVRAM), an electrically erasable programmable read-only memory (EEPROM), a flash memory, a magnetic or optical data storage medium, and the like. Additionally or alternatively, the techniques may be implemented at least in part by a computer-readable communication medium that carries or communicates program code in the form of instructions or data structures and that can be accessed, read, and / or executed by a computer, such as a propagated signal or wave.

[0148] The program code may be executed by a processor, which may include one or more processors, such as one or more digital signal processors (DSPs), general-purpose microprocessors, application-specific integrated circuits (ASICs), field programmable logic arrays (FPGAs), or other equivalent integrated or discrete logic circuits. Such processors may be configured to perform any of the techniques described in this disclosure. A general-purpose processor may be a microprocessor; however, in an alternative, the processor may be any conventional processor, controller, microcontroller, or state machine. The processor may also be implemented as a combination of computing devices, such as a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors combined with a DSP core, or any other such configuration. Therefore, the term "processor" as used herein may refer to any of the aforementioned structures, any combination of the aforementioned structures, or any other structure or device suitable for implementing the techniques described herein.

[0149] It should be understood by those of ordinary skill in the art that the less than ("<") and greater than (">") symbols or terms used herein may be replaced by less than or equal to ("≤") and greater than or equal to ("≥") symbols, respectively, without departing from the scope of the present specification.

[0150] Where a component is described as being “configured to” perform certain operations, such configuration may be achieved, for example, by designing electronic circuits or other hardware to perform the operations, by programming programmable electronic circuits (e.g., a microprocessor or other suitable electronic circuits) to perform the operations, or any combination thereof.

[0151] The phrases “coupled to” or “communicatively coupled to” refer to any component being physically connected directly or indirectly to another component, and / or any component being in communication directly or indirectly with another component (e.g., connected to the other component via a wired or wireless connection and / or other suitable communication interface).

[0152] Claim language or other language stating "at least one of" a set and / or "one or more of" a set indicates that one member of the set or multiple members of the set (in any combination) satisfies the claim. For example, claim language stating "at least one of A and B" or "at least one of A or B" means A, B, or A and B. In another example, claim language stating "at least one of A, B, and C" or "at least one of A, B, or C" means A, B, C, or A and B, or A and C, or B and C, or any repetition is information or data (e.g., A and A, B and B, C and C, A and A and B, etc.), or any other ordering, repetition, or combination of A, B, and C. The language "at least one of" a set and / or "one or more of" a set does not limit the set to the items listed in the set. For example, claim language stating "at least one of A and B" or "at least one of A or B" may mean A, B, or A and B, and may additionally include items not listed in the set of A and B.

[0153] Illustrative aspects of the present disclosure include:

[0154] Aspect 1. A method of wireless communication performed at a user equipment (UE), the method comprising: using a tested machine learning model to generate a first representation of control information associated with a communication channel; using a reference machine learning model to generate a second representation of the control information associated with the communication channel; and sending to a device information associated with a comparison based on the first representation of the control information and the second representation of the control information.

[0155] Aspect 2. A method according to Aspect 1, wherein the first representation of the control information and the second representation of the control information are sent to the device for performing the comparison of the first representation of the control information and the second representation of the control information to monitor the performance of the machine learning model under test.

[0156] Aspect 3. A method according to any one of Aspects 1 or 2, wherein the first representation of the control information and the second representation of the control information are sent to the device for: using a first decoder to process the first representation of the control information to generate a first reconstructed representation of the control information; using a second decoder to process the second representation of the control information to generate a second reconstructed representation of the control information; and performing the comparison at least in part by comparing the first reconstructed representation of the control information with the second reconstructed representation of the control information.

[0157] Aspect 4. The method according to aspect 1, further comprising performing the comparison and sending the information associated with the comparison to the device as a result of the comparison.

[0158] Aspect 5. A method according to any one of aspects 1 to 4, wherein the control information includes channel state information associated with the communication channel.

[0159] Aspect 6. A method according to any one of Aspects 1 to 5, wherein the machine learning model under test includes a first encoder neural network model, which is trained to compress control information into a first compressed representation when operating in a first environment, and wherein the reference machine learning model includes a second encoder neural network model, which is trained to compress control information into a second compressed representation when operating in the first environment and the second environment.

[0160] Aspect 7. The method according to any one of aspects 1 to 6, wherein the first environment comprises an indoor environment, and wherein the second environment comprises an outdoor environment.

[0161] Aspect 8. According to any one of Aspects 1 to 7, the method also includes: receiving information associated with the performance of the tested machine learning model from the device.

[0162] Aspect 9. A method according to any one of Aspects 1 to 8, wherein the information associated with the performance of the tested machine learning model indicates that the tested machine learning model is inaccurate for the communication channel.

[0163] Aspect 10. According to any one of Aspects 1 to 9, the method further includes: based on the information indicating that the tested machine learning model is inaccurate for the communication channel, switching to an alternative machine learning model to further communicate with the device or additional device.

[0164] Aspect 11. A method according to any one of Aspects 1 to 10, wherein the information associated with the performance of the tested machine learning model indicates that the tested machine learning model is accurate for the communication channel.

[0165] Aspect 12. According to the method described in any one of Aspects 1 to 11, the method further includes: continuing to use the machine learning model under test based on the information indicating that the machine learning model under test is accurate for the communication channel.

[0166] Aspect 13. A method according to any one of Aspects 1 to 12, the method further comprising: updating the tested machine learning model based at least in part on the information to generate an updated machine learning model.

[0167] Aspect 14. According to any one of Aspects 1 to 13, the method further includes: receiving information including a first trigger for using the machine learning model under test from the device; and generating the first representation of the control information using the machine learning model under test based on the first trigger.

[0168] Aspect 15. According to any one of Aspects 1 to 14, the method further includes: receiving information including a second trigger for using the reference machine learning model from the device; and generating the second representation of the control information using the reference machine learning model based on the second trigger.

[0169] Aspect 16. A method according to any one of Aspects 1 to 15, wherein the use of the tested machine learning model is triggered more frequently than the use of the reference machine learning model.

[0170] Aspect 17. A method according to any one of Aspects 1 to 16, wherein the use of at least one of the tested machine learning model or the reference machine learning model is triggered based on an event.

[0171] Aspect 18. A method according to any one of Aspects 1 to 17, wherein the event includes at least one of the following: the UE moves to a new environment for which the machine learning model under test is not trained, degradation of throughput via the communication channel, high block error rate (BLER) conditions, or periodic time for monitoring the performance of the machine learning model under test.

[0172] Aspect 19. A method according to any one of aspects 1 to 18, wherein the device comprises a base station.

[0173] Aspect 20. An apparatus for wireless communication, the apparatus comprising: at least one memory; and at least one processor, the at least one processor being coupled to the at least one memory, the at least one processor being configured to: generate a first representation of control information associated with a communication channel using a tested machine learning model; generate a second representation of the control information associated with the communication channel using a reference machine learning model; and send information associated with a comparison based on the first representation of the control information and the second representation of the control information to a device.

[0174] Aspect 21. An apparatus according to Aspect 20, wherein the device performs a comparison of the first representation of the control information with the second representation of the control information to monitor the performance of the machine learning model under test.

[0175] Aspect 22. An apparatus according to any one of Aspects 20 or 21, wherein the first representation of the control information and the second representation of the control information are sent to the device for: processing the first representation of the control information using a first decoder to generate a first reconstructed representation of the control information; processing the second representation of the control information using a second decoder to generate a second reconstructed representation of the control information; and performing the comparison at least in part by comparing the first reconstructed representation of the control information with the second reconstructed representation of the control information.

[0176] Aspect 23. The apparatus according to aspect 20, wherein the apparatus performs the comparison and sends the information associated with the comparison to the device as a result of the comparison.

[0177] Aspect 24. An apparatus according to any one of aspects 20 to 23, wherein the control information comprises channel state information associated with the communication channel.

[0178] Aspect 25. An apparatus according to any one of Aspects 20 to 24, wherein the tested machine learning model comprises a first encoder neural network model, which is trained to compress control information into a first compressed representation when operating in a first environment, and wherein the reference machine learning model comprises a second encoder neural network model, which is trained to compress control information into a second compressed representation when operating in the first environment and the second environment.

[0179] Aspect 26. The apparatus according to any one of Aspects 20 to 25, wherein the first environment comprises an indoor environment, and wherein the second environment comprises an outdoor environment.

[0180] Aspect 27. An apparatus according to any one of Aspects 20 to 26, wherein the at least one processor is further configured to: receive information associated with the performance of the tested machine learning model from the device.

[0181] Aspect 28. An apparatus according to any one of Aspects 20 to 27, wherein the information associated with the performance of the tested machine learning model indicates that the tested machine learning model is inaccurate for the communication channel.

[0182] Aspect 29. An apparatus according to any one of Aspects 20 to 28, wherein the at least one processor is further configured to: based on the information indicating that the tested machine learning model is inaccurate for the communication channel, switch to an alternative machine learning model to further communicate with the device or an additional device.

[0183] Aspect 30. An apparatus according to any one of Aspects 20 to 29, wherein the information associated with the performance of the tested machine learning model indicates that the tested machine learning model is accurate for the communication channel.

[0184] Aspect 31. An apparatus according to any one of Aspects 20 to 30, wherein the at least one processor is further configured to: continue to use the tested machine learning model based on the information indicating that the tested machine learning model is accurate for the communication channel.

[0185] Aspect 32. An apparatus according to any one of Aspects 20 to 31, wherein the at least one processor is further configured to: update the tested machine learning model based at least in part on the information to generate an updated machine learning model.

[0186] Aspect 33. An apparatus according to any one of Aspects 20 to 32, wherein the at least one processor is further configured to: receive information from the device including a first trigger for using the machine learning model under test; and generate the first representation of the control information using the machine learning model under test based on the first trigger.

[0187] Aspect 34. An apparatus according to any one of Aspects 20 to 33, wherein the at least one processor is further configured to: receive information from the device including a second trigger for using the reference machine learning model; and generate the second representation of the control information using the reference machine learning model based on the second trigger.

[0188] Aspect 35. An apparatus according to any one of Aspects 20 to 34, wherein use of the tested machine learning model is triggered more frequently than use of the reference machine learning model.

[0189] Aspect 36. An apparatus according to any one of Aspects 20 to 35, wherein the use of at least one of the tested machine learning model or the reference machine learning model is triggered based on an event.

[0190] Aspect 37. An apparatus according to any one of Aspects 20 to 36, wherein the event includes at least one of the following: the UE moves to a new environment for which the machine learning model under test is not trained, a degradation of the throughput via the communication channel, a high block error rate (BLER) condition, or a periodic time for monitoring the performance of the machine learning model under test.

[0191] Aspect 38. An apparatus according to any one of aspects 20 to 37, wherein the device comprises a base station.

[0192] Aspect 39. A method for performing wireless communications at a first device, the method comprising: receiving a first representation of control information associated with a communication channel generated using a first tested machine learning model from a second device; receiving a second representation of the control information associated with the communication channel generated using a first reference machine learning model from the second device; reconstructing the control information from the first representation of the control information using a second tested machine learning model at the first device to generate a first reconstruction of the control information; reconstructing the control information from the second representation of the control information using a second reference machine learning model at the first device to generate a second reconstruction of the control information; and determining the accuracy of the first tested machine learning model for the communication channel at the first device based on a comparison of the first reconstruction of the control information and the second reconstruction of the control information.

[0193] Aspect 40. A method according to claim 39, wherein the second tested machine learning model includes a first decoder, which is configured to generate the first reconstruction of the control information, and wherein the second reference machine learning model includes a second decoder, which is configured to generate the second reconstruction of the control information.

[0194] Aspect 41. An apparatus for wireless communication, the apparatus comprising: at least one memory; and at least one processor, the at least one processor being coupled to the at least one memory, the at least one processor being configured to: receive from a device a first representation of control information associated with a communication channel generated using a first tested machine learning model; receive from the device a second representation of the control information associated with the communication channel generated using a first reference machine learning model; reconstruct the control information from the first representation of the control information using a second tested machine learning model at the apparatus to generate a first reconstruction of the control information; reconstruct the control information from the second representation of the control information using a second reference machine learning model at the apparatus to generate a second reconstruction of the control information; and determine at the apparatus the accuracy of the first tested machine learning model for the communication channel based on a comparison of the first reconstruction of the control information and the second reconstruction of the control information.

[0195] Aspect 42. An apparatus according to claim 41, wherein the second tested machine learning model includes a first decoder, which is configured to generate the first reconstruction of the control information, and wherein the second reference machine learning model includes a second decoder, which is configured to generate the second reconstruction of the control information.

[0196] Aspect 43. A non-transitory computer-readable storage medium, comprising instructions stored thereon, which, when executed by at least one processor, cause the at least one processor to perform operations according to any one of claims 1 to 19 and / or claims 39 to 40.

Claims

1. A method of wireless communication performed at a user equipment (UE), the method comprising: generating a first representation of control information associated with a communication channel using the machine learning model under test; generating a second representation of the control information associated with the communication channel using a reference machine learning model; as well as Information associated with a comparison based on the first representation of the control information and the second representation of the control information is sent to a device.

2. The method of claim 1, wherein the first representation of the control information and the second representation of the control information are sent to the device for performing the comparison of the first representation of the control information and the second representation of the control information to monitor the performance of the machine learning model under test.

3. A method according to any one of claims 1 or 2, wherein the first representation of the control information and the second representation of the control information are sent to the device for: using a first decoder to process the first representation of the control information to generate a first reconstructed representation of the control information; using a second decoder to process the second representation of the control information to generate a second reconstructed representation of the control information; and performing the comparison at least in part by comparing the first reconstructed representation of the control information with the second reconstructed representation of the control information. 4 . The method of claim 1 , further comprising performing the comparison and sending the information associated with the comparison to the device as a result of the comparison.

5. The method of claim 1, wherein the control information comprises channel state information associated with the communication channel.

6. A method according to any one of claims 2 to 5, wherein the tested machine learning model includes a first encoder neural network model, which is trained to compress control information into a first compressed representation when operating in a first environment, and wherein the reference machine learning model includes a second encoder neural network model, which is trained to compress control information into a second compressed representation when operating in the first environment and the second environment. The method of claim 6 , wherein the first environment comprises an indoor environment, and wherein the second environment comprises an outdoor environment.

8. The method according to any one of claims 1 to 7, further comprising: Information associated with the performance of the tested machine learning model is received from the device.

9. The method of claim 8, wherein the information associated with the performance of the tested machine learning model indicates that the tested machine learning model is inaccurate for the communication channel.

10. The method according to claim 9, further comprising: Based on the information indicating that the tested machine learning model is inaccurate for the communication channel, switch to an alternative machine learning model to further communicate with the device or additional device.

11. The method of claim 8, wherein the information associated with the performance of the tested machine learning model indicates that the tested machine learning model is accurate for the communication channel.

12. The method according to claim 11, further comprising: Based on the information indicating that the tested machine learning model is accurate for the communication channel, continue to use the tested machine learning model.

13. The method according to any one of claims 8 to 12, further comprising: The tested machine learning model is updated based at least in part on the information to generate an updated machine learning model.

14. The method according to any one of claims 1 to 13, further comprising: receiving, from the device, information including a first trigger for using the machine learning model under test; as well as Generate the first representation of the control information using the machine learning model under test based on the first trigger.

15. The method according to any one of claims 1 to 14, further comprising: receiving, from the device, information including a second trigger to use the reference machine learning model; as well as The second representation of the control information is generated using the reference machine learning model based on the second trigger.

16. A method according to claim 15, wherein the use of the tested machine learning model is triggered more frequently than the use of the reference machine learning model.

17. A method according to any one of claims 1 to 16, wherein the use of at least one of the tested machine learning model or the reference machine learning model is triggered based on an event.

18. The method of claim 17, wherein the event comprises at least one of: the UE moving to a new environment for which the machine learning model under test was not trained, a degradation in throughput via the communication channel, a high block error rate (BLER) condition, or a periodic time of monitoring the performance of the machine learning model under test.

19. The method of any one of claims 1 to 18, wherein the device comprises a base station.

20. An apparatus for wireless communication, the apparatus comprising: at least one memory; and at least one processor, the at least one processor coupled to the at least one memory, the at least one processor configured to: generating a first representation of control information associated with a communication channel using the machine learning model under test; generating a second representation of the control information associated with the communication channel using a reference machine learning model; as well as Information associated with a comparison based on the first representation of the control information and the second representation of the control information is sent to a device.

21. An apparatus according to claim 20, wherein the at least one processor is configured to send the first representation of the control information and the second representation of the control information to the device for performing the comparison of the first representation of the control information and the second representation of the control information to monitor the performance of the machine learning model under test.

22. An apparatus according to any one of claims 20 or 21, wherein the first representation of the control information and the second representation of the control information are sent to the device for: using a first decoder to process the first representation of the control information to generate a first reconstructed representation of the control information; using a second decoder to process the second representation of the control information to generate a second reconstructed representation of the control information; and performing the comparison at least in part by comparing the first reconstructed representation of the control information with the second reconstructed representation of the control information.

23. The apparatus of claim 20, wherein the at least one processor is configured to: performing the comparison; and The information associated with the comparison is sent to the device as a result of the comparison.

24. The apparatus of claim 20, wherein the control information comprises channel state information associated with the communication channel.

25. An apparatus according to any one of claims 20 to 24, wherein the tested machine learning model comprises a first encoder neural network model, which is trained to compress control information into a first compressed representation when operating in a first environment, and wherein the reference machine learning model comprises a second encoder neural network model, which is trained to compress control information into a second compressed representation when operating in the first environment and the second environment.

26. The apparatus of claim 25, wherein the first environment comprises an indoor environment, and wherein the second environment comprises an outdoor environment.

27. The apparatus of any one of claims 20 to 26, wherein the at least one processor is further configured to: Information associated with the performance of the tested machine learning model is received from the device.

28. An apparatus according to claim 27, wherein the information associated with the performance of the tested machine learning model indicates that the tested machine learning model is inaccurate for the communication channel.

29. The apparatus of claim 28, wherein the at least one processor is further configured to: Based on the information indicating that the tested machine learning model is inaccurate for the communication channel, switch to an alternative machine learning model to further communicate with the device or additional device.

30. An apparatus according to claim 29, wherein the information associated with the performance of the tested machine learning model indicates that the tested machine learning model is accurate for the communication channel.

31. The apparatus of claim 30, wherein the at least one processor is further configured to: Based on the information indicating that the tested machine learning model is accurate for the communication channel, continue to use the tested machine learning model.

32. The apparatus of any one of claims 27 to 31, wherein the at least one processor is further configured to: The machine learning model under test is updated based at least in part on the information associated with the performance of the machine learning model under test to generate an updated machine learning model.

33. The apparatus of any one of claims 20 to 32, wherein the at least one processor is further configured to: receiving, from the device, information including a first trigger for using the machine learning model under test; and Generate the first representation of the control information using the machine learning model under test based on the first trigger.

34. The apparatus of any one of claims 20 to 33, wherein the at least one processor is further configured to: receiving, from the device, information including a second trigger to use the reference machine learning model; and The second representation of the control information is generated using the reference machine learning model based on the second trigger.

35. An apparatus according to claim 34, wherein use of the tested machine learning model is triggered more frequently than use of the reference machine learning model.

36. An apparatus according to any one of claims 20 to 35, wherein the use of at least one of the tested machine learning model or the reference machine learning model is triggered based on an event.

37. A device according to claim 36, wherein the event includes at least one of the following: the device moves to a new environment for which the machine learning model under test was not trained, a degradation of the throughput via the communication channel, a high block error rate (BLER) condition, or a periodic time to monitor the performance of the machine learning model under test.

38. An apparatus according to any one of claims 20 to 37, wherein the device comprises a base station.

39. A method of wireless communication at a first device, the method comprising: receiving, from a second device, a first representation of control information associated with a communication channel generated using a first tested machine learning model; receiving, from the second device, a second representation of the control information associated with the communication channel generated using a first reference machine learning model; reconstructing the control information from the first representation of the control information using a second tested machine learning model at the first device to generate a first reconstruction of the control information; reconstructing the control information from the second representation of the control information using a second reference machine learning model at the first device to generate a second reconstruction of the control information; as well as Determining, at the first device, an accuracy of the first tested machine learning model for the communication channel based on a comparison of the first reconstruction of the control information and the second reconstruction of the control information.

40. A method according to claim 39, wherein the second tested machine learning model includes a first decoder, which is configured to generate the first reconstruction of the control information, and wherein the second reference machine learning model includes a second decoder, which is configured to generate the second reconstruction of the control information.

41. An apparatus for wireless communication, the apparatus comprising: at least one memory; and at least one processor, the at least one processor coupled to the at least one memory, the at least one processor configured to: receiving, from a device, a first representation of control information associated with a communication channel generated using a first tested machine learning model; receiving, from the device, a second representation of the control information associated with the communication channel generated using a first reference machine learning model; reconstructing the control information from the first representation of the control information using a second tested machine learning model at the device to generate a first reconstruction of the control information; reconstructing the control information from the second representation of the control information using a second reference machine learning model at the device to generate a second reconstruction of the control information; as well as Determining, at the device, an accuracy of the first tested machine learning model for the communication channel based on a comparison of the first reconstruction of the control information and the second reconstruction of the control information.

42. An apparatus according to claim 41, wherein the second tested machine learning model includes a first decoder, which is configured to generate the first reconstruction of the control information, and wherein the second reference machine learning model includes a second decoder, which is configured to generate the second reconstruction of the control information.

43. A non-transitory computer-readable storage medium comprising instructions stored thereon, which, when executed by at least one processor, cause the at least one processor to perform operations according to any one of claims 1 to 19 and / or claims 39 to 40.

44. An apparatus for wireless communication, the apparatus comprising one or more means for performing the operations of any one of claims 1 to 19 and / or claims 39 to 40.

Images