Machine learning at a wireless device

Abstract

Various aspects of the present disclosure relate to machine learning at a wireless device. An apparatus, such as a node (e.g., a user equipment (UE) and / or a network entity (NE)), receives, from at least one second node, a set of data samples associated with a wireless channel. The first node obtains a subset of data samples from the set of data samples, where the subset of data samples includes at least a first subset of data samples and a second subset of data samples. The first node determines, from the subset of data samples, one or more parameters of a machine learning model of a wireless device associated with the first node. In some instances, the first node transmits the parameters of the machine learning model to a second node for implementation at a machine learning model of a wireless device associated with the second node.

Description

[0001] Related applications

[0002] This application claims priority to U.S. Provisional Application Serial No. 63 / 627,296, filed January 31, 2024, entitled "Machine Learning at a Wireless Device," the disclosure of which is incorporated herein by reference in its entirety. Technical Field

[0003] This disclosure relates to wireless communication, and more specifically to machine learning techniques for wireless communication. Background Technology

[0004] A wireless communication system may include one or more network communication devices, such as base stations, which can support wireless communication with one or more user communication devices, which may also be referred to as user equipment (UE) or other suitable terms. The wireless communication system can support wireless communication with one or more user communication devices by utilizing the resources of the wireless communication system (e.g., time resources (e.g., symbols, time slots, subframes, frames, etc.) or frequency resources (e.g., subcarriers, carriers, etc.)). Furthermore, the wireless communication system can support wireless communication across various radio access technologies, including third-generation (3G), fourth-generation (4G), fifth-generation (5G), and other suitable radio access technologies above 5G (e.g., sixth-generation (6G)). Summary of the Invention

[0005] The article “a(a)” preceding an element is not limited and should be understood to mean “at least one” or “one or more” of those elements. The terms “a,” “at least one,” “one or more,” and “at least one of one or more” are interchangeable. As used herein (included in the claims), “or” as used in a list of items (for example, a list of items followed by phrases such as “at least one of…” or “one or more of…” or “one or both of…”) indicates a list of inclusion, such that (for example) a list of at least one of A, B, or C means A or B or C or AB or AC or BC or ABC (i.e., A and B and C). Furthermore, as used herein, the phrase “based on” should not be construed as a reference to a set of closed conditions. For example, an exemplary step described as “based on condition A” could be based on both condition A and condition B without departing from the scope of this disclosure. In other words, as used herein, the phrase “based on” should be interpreted in the same manner as the phrase “at least partially based on.” Furthermore, as used herein (included in the claims), a “set” may contain one or more elements.

[0006] Some embodiments of the methods and apparatus described herein may further include a first node for wireless communication, configured to: receive a set of data samples associated with a wireless channel from at least one second node; obtain a subset of data samples from the set of data samples, wherein the subset of data samples includes at least a first subset of data samples and a second subset of data samples; and determine one or more parameters of a machine learning model of a wireless device associated with the first node from the subset of data samples.

[0007] In some embodiments of the methods and apparatus described herein, the first node transmits signaling indicating one or more parameters of the machine learning model to the at least one second node. Alternatively, a corresponding data sample in the data sample set includes at least one of a first value representing an output corresponding to the wireless channel and a second value representing a message associated with the output corresponding to the wireless channel. Alternatively, the message corresponds to a transmission from one or more of the at least one second node or at least one third node. Alternatively, to determine the one or more parameters of the machine learning model, the first node: in response to training a first instance of the machine learning model using supervised learning techniques based on the first data sample subset, obtains a first parameter set associated with the first data sample subset; in response to training a second instance of the machine learning model using the supervised learning techniques based on a second data sample subset, obtains a second parameter set associated with the second data sample subset; and obtains the one or more parameters by combining the first parameter set associated with the first data sample subset and the second parameter set associated with the second data sample subset. Alternatively, the one or more parameters define the machine learning model.

[0008] Alternatively or alternatively, the data sample set comprises one or more different data sample types, and a corresponding subset of data samples comprises at least one data sample of the one or more different data sample types. Alternatively or alternatively, the number of corresponding values ​​of data samples in the subset of data samples satisfies a threshold, and wherein the threshold is based on the total number of values ​​of data samples in the data sample set and the number of values ​​in the subset of data samples. Alternatively or alternatively, the first node: obtains an output from the machine learning model associated with at least a portion of the received transmission based on inputting an additional data sample set associated with the received transmission into the machine learning model, wherein the at least a portion of the received transmission corresponds to corrupted signaling; and decodes the received transmission based on obtaining the at least a portion of the received transmission. Alternatively or alternatively, the first node: obtains an output from the machine learning model associated with at least one of a beam selection procedure or an estimation procedure, the beam selection procedure or the estimation procedure being associated with the wireless channel, based on inputting an additional data sample set into the machine learning model; and selects a beam for the beam selection procedure or updates one or more communication parameters based on the output. Alternatively or concurrently, the wireless device includes at least one of a receiver or a decoder.

[0009] Some embodiments of the methods and apparatus described herein may further include a processor for wireless communication, configured to: receive a set of data samples associated with a wireless channel from at least one second node; obtain a subset of data samples from the set of data samples, wherein the subset of data samples includes at least a first subset of data samples and a second subset of data samples; and determine one or more parameters from the subset of data samples of a machine learning model of a wireless device associated with the processor.

[0010] In some embodiments of the methods and apparatus described herein, the processor transmits signaling to the at least one second node indicating one or more parameters of the machine learning model. Alternatively, a corresponding data sample in the data sample set includes at least one of a first value representing an output corresponding to the wireless channel and a second value representing a message associated with the output corresponding to the wireless channel. Alternatively, the message corresponds to a transmission from one or more of the at least one second node or at least one third node. Alternatively, to determine the one or more parameters of the machine learning model, the processor: in response to a first instance of training the machine learning model using supervised learning techniques based on a first subset of data samples, obtains a first parameter set associated with the first subset of data samples; in response to a second instance of training the machine learning model using the supervised learning techniques based on a second subset of data samples, obtains a second parameter set associated with the second subset of data samples; and obtains the one or more parameters by combining the first parameter set associated with the first subset of data samples and the second parameter set associated with the second subset of data samples. Alternatively, the one or more parameters define the machine learning model.

[0011] Alternatively or alternatively, the data sample set comprises one or more different data sample types, and a corresponding subset of data samples comprises at least one data sample of the one or more different data sample types. Alternatively or alternatively, the number of corresponding values ​​of data samples in the subset of data samples satisfies a threshold, and wherein the threshold is based on the total number of values ​​of data samples in the data sample set and the number of values ​​in the subset of data samples. Alternatively or alternatively, the processor: obtains an output from the machine learning model associated with at least a portion of the received transmission based on inputting an additional data sample set associated with the received transmission into the machine learning model, wherein the at least a portion of the received transmission corresponds to corrupted signaling; and decodes the received transmission based on obtaining the at least a portion of the received transmission. Alternatively or alternatively, the processor: obtains an output from the machine learning model associated with at least one of a beam selection procedure or an estimation procedure, the beam selection procedure or the estimation procedure being associated with the wireless channel, based on inputting an additional data sample set into the machine learning model; and selects a beam for the beam selection procedure or updates one or more communication parameters based on the output. Alternatively or concurrently, the wireless device includes at least one of a receiver or a decoder.

[0012] Some embodiments of the methods and apparatus described herein may further include a method performed by a UE, the method comprising: receiving a set of data samples associated with a wireless channel from at least one second node; obtaining a subset of data samples from the set of data samples, wherein the subset of data samples includes at least a first subset of data samples and a second subset of data samples; and determining one or more parameters of a machine learning model of a wireless device associated with the first node from the subset of data samples.

[0013] In some embodiments of the methods and apparatus described herein, the method includes transmitting signaling to the at least one second node indicating one or more parameters of the machine learning model. Alternatively, a corresponding data sample in the data sample set includes at least one of a first value representing an output corresponding to the wireless channel and a second value representing a message associated with the output corresponding to the wireless channel. Alternatively, the message corresponds to a transmission from one or more of the at least one second node or at least one third node. Alternatively, to determine the one or more parameters of the machine learning model, the method includes: obtaining a first parameter set associated with the first data sample subset in response to training a first instance of the machine learning model using supervised learning techniques based on a first subset of data samples; obtaining a second parameter set associated with the second data sample subset in response to training a second instance of the machine learning model using the supervised learning techniques based on a second subset of data samples; and obtaining the one or more parameters by combining the first parameter set associated with the first data sample subset and the second parameter set associated with the second data sample subset. Alternatively, the one or more parameters define the machine learning model.

[0014] Alternatively or alternatively, the data sample set comprises one or more different data sample types, and a corresponding subset of data samples comprises at least one data sample of the one or more different data sample types. Alternatively or alternatively, the number of corresponding values ​​of data samples in the subset of data samples satisfies a threshold, and wherein the threshold is based on the total number of values ​​of data samples in the data sample set and the number of values ​​in the subset of data samples. Alternatively or alternatively, the method comprises: obtaining an output from the machine learning model associated with at least a portion of the received transmission based on inputting an additional data sample set associated with the received transmission into the machine learning model, wherein the at least a portion of the received transmission corresponds to corrupted signaling; and decoding the received transmission based on obtaining the at least a portion of the received transmission. Alternatively or alternatively, the method comprises: obtaining an output from the machine learning model associated with at least one of a beam selection procedure or an estimation procedure, the beam selection procedure or the estimation procedure being associated with the wireless channel, based on inputting an additional data sample set into the machine learning model; and selecting a beam for the beam selection procedure or updating one or more communication parameters based on the output. Alternatively or concurrently, the wireless device includes at least one of a receiver or a decoder.

[0015] Some embodiments of the methods and apparatus described herein may further include a first node for wireless communication, configured to: transmit a first signaling to a second node, the first signaling indicating a data sample set associated with a wireless channel; and receive a second signaling from the second node, the second signaling indicating one or more parameters of a machine learning model of a wireless device associated with the first node, wherein the one or more parameters are obtained from a subset of data samples from the data sample set.

[0016] In some embodiments of the methods and apparatus described herein, a corresponding data sample in the data sample set includes at least one of a first value representing an output corresponding to the wireless channel and a second value representing a message associated with the output corresponding to the wireless channel. Alternatively, the message corresponds to a transmission from one or more of the first node or at least one third node. Alternatively, the one or more parameters of the machine learning model are obtained based on combining a corresponding parameter set associated with the subset of data samples, wherein the corresponding parameter set is obtained in response to training a corresponding instance of the machine learning model using supervised learning techniques based on the subset of data samples. Alternatively, the one or more parameters define the machine learning model.

[0017] Alternatively or alternatively, the data sample set comprises one or more different data sample types, and a corresponding subset of data samples comprises at least one data sample of the one or more different data sample types. Alternatively or alternatively, the number of corresponding values ​​of data samples in the subset of data samples satisfies a threshold, and wherein the threshold is based on the total number of values ​​of data samples in the data sample set and the number of values ​​in the subset of data samples. Alternatively or alternatively, the first node: obtains an output from the machine learning model associated with at least a portion of the received transmission based on inputting an additional data sample set associated with the received transmission into the machine learning model, wherein the at least a portion of the received transmission corresponds to corrupted signaling; and decodes the received transmission based on obtaining the at least a portion of the received transmission. Alternatively or alternatively, the first node: obtains an output from the machine learning model associated with at least one of a beam selection procedure or an estimation procedure, the beam selection procedure or the estimation procedure being associated with the wireless channel, based on inputting an additional data sample set into the machine learning model; and selects a beam for the beam selection procedure or updates one or more communication parameters based on the output. Alternatively or concurrently, the wireless device includes at least one of a receiver or a decoder.

[0018] Some embodiments of the methods and apparatus described herein may further include a method performed by a first node, the method comprising: transmitting a first signaling to a second node, the first signaling indicating a data sample set associated with a wireless channel; and receiving a second signaling from the second node, the second signaling indicating one or more parameters of a machine learning model of a wireless device associated with the first node, wherein the one or more parameters are obtained from a subset of data samples of the data sample set.

[0019] In some embodiments of the methods and apparatus described herein, a corresponding data sample in the data sample set includes at least one of a first value representing an output corresponding to the wireless channel and a second value representing a message associated with the output corresponding to the wireless channel. Alternatively, the message corresponds to a transmission from one or more of the first node or at least one third node. Alternatively, the one or more parameters of the machine learning model are obtained based on combining a corresponding parameter set associated with the subset of data samples, wherein the corresponding parameter set is obtained in response to training a corresponding instance of the machine learning model using supervised learning techniques based on the subset of data samples. Alternatively, the one or more parameters define the machine learning model.

[0020] Alternatively or alternatively, the data sample set comprises one or more different data sample types, and a corresponding subset of data samples comprises at least one data sample of the one or more different data sample types. Alternatively or alternatively, the number of corresponding values ​​of data samples in the subset of data samples satisfies a threshold, and wherein the threshold is based on the total number of values ​​of data samples in the data sample set and the number of values ​​in the subset of data samples. Alternatively or alternatively, the method comprises: obtaining an output from the machine learning model associated with at least a portion of the received transmission based on inputting an additional data sample set associated with the received transmission into the machine learning model, wherein the at least a portion of the received transmission corresponds to corrupted signaling; and decoding the received transmission based on obtaining the at least a portion of the received transmission. Alternatively or alternatively, the method comprises: obtaining an output from the machine learning model associated with at least one of a beam selection procedure or an estimation procedure, the beam selection procedure or the estimation procedure being associated with the wireless channel, based on inputting an additional data sample set into the machine learning model; and selecting a beam for the beam selection procedure or updating one or more communication parameters based on the output. Alternatively or concurrently, the wireless device includes at least one of a receiver or a decoder. Attached Figure Description

[0021] Figure 1Illustrated examples of wireless communication systems according to various aspects of this disclosure.

[0022] Figure 2 Illustrated examples of machine learning schemas based on various aspects of this disclosure.

[0023] Figure 3 Illustrated examples of wireless communication systems according to various aspects of this disclosure.

[0024] Figure 4 Illustrated examples of signaling diagrams based on various aspects of this disclosure.

[0025] Figure 5 Illustrated examples of UEs based on various aspects of this disclosure.

[0026] Figure 6 Illustrated examples of processors according to various aspects of this disclosure.

[0027] Figure 7 Illustrated examples of network equipment (NE) according to various aspects of this disclosure.

[0028] Figure 8 The diagram illustrates a flowchart of a method performed by a UE and / or NE according to various aspects of this disclosure.

[0029] Figure 9 The diagram illustrates a flowchart of a method performed by a UE and / or NE according to various aspects of this disclosure. Detailed Implementation

[0030] A wireless communication system may include one or more means for transmitting and receiving signaling. The means in a wireless communication system may include a UE and / or NE, as well as other means, and may be referred to as a node. A node may include receiving means, such as a receiver and / or decoder, for receiving signaling from other nodes. In variations, the receiving means may implement one or more machine learning models for decoding the transmission. For example, the receiving means may obtain a decoded message as the output of one or more machine learning models based on inputting one or more values ​​representing a message output from a wireless channel into one or more machine learning models. In some cases, the machine learning model may have a generalization error, which refers to the performance of the machine learning model on new data not presented during its training. That is, the generalization error measures the difference between the performance of the machine learning model on a training dataset and the performance of the machine learning model on a test dataset containing data not used in the training dataset. In some instances, to reduce the generalization error of the machine learning model, the user and / or node may manually adjust one or more parameters of the machine learning model and / or increase the number of numerical values ​​of the training data samples in the training dataset. However, obtaining a sufficient number of training data samples to meet the generalization error threshold of a machine learning model may increase signaling overhead and processing due to the collection of training data samples. Alternatively, increasing the number of training data samples may lead to communication latency due to the increased training duration associated with processing training data samples during the machine learning model's training process. Alternatively, manually adjusting the parameters of the machine learning model by the user may result in errors in the parameters, and generalization error may still occur due to the user incorrectly adjusting the parameters (e.g., by selecting parameters that increase generalization error instead of decreasing it).

[0031] As described herein, to reduce the generalization error of a machine learning model implemented at a node's receiver, the training dataset is divided into subsets of data samples for training multiple instances of the machine learning model. In some instances, a node (e.g., a UE and / or NE) may use a corresponding subset of data samples to train instances of the machine learning model to obtain one or more parameters of those instances. The parameters of the machine learning model instances may be selected to minimize empirical risk, which defines the average loss or error caused by the machine learning model on the training dataset. Nodes may combine or aggregate the parameters of the machine learning model instances to obtain parameters of a global machine learning model to be implemented at a receiving device (e.g., a receiver and / or decoder). By aggregating the parameters of the machine learning model instances, variations in one or more of the parameters can be reduced or eliminated, thereby reducing the generalization error of the global machine learning model. In a variant, a node may implement the global machine learning model at the node's receiving device. In other variants, a node may transfer the parameters of the global machine learning model to another node for implementation at the receiving device of that other node.

[0032] Various aspects of this disclosure are described in the context of wireless communication systems.

[0033] Figure 1 The illustration depicts examples of a wireless communication system 100 according to various aspects of this disclosure. The wireless communication system 100 may include one or more NEs 102, one or more UEs 104, and a core network (CN) 106. The wireless communication system 100 may support various radio access technologies. In some embodiments, the wireless communication system 100 may be a 4G network, such as an LTE network or an LTE-A advanced network. In some other embodiments, the wireless communication system 100 may be an NR network, such as a 5G network, an 5G-A advanced network, or a 5G ultra-wideband (5G-UWB) network. In other embodiments, the wireless communication system 100 may be a combination of 4G and 5G networks, or other suitable radio access technologies, including IEEE 802.11 (Wi-Fi), IEEE 802.16 (WiMAX), and IEEE 802.20. The wireless communication system 100 may support radio access technologies beyond 5G, such as 6G. In addition, the wireless communication system 100 can support technologies such as Time Division Multiple Access (TDMA), Frequency Division Multiple Access (FDMA), or Code Division Multiple Access (CDMA).

[0034] One or more NEs 102 may be distributed throughout a geographic area to form a wireless communication system 100. One or more of the NEs 102 described herein may be, include, or be referred to as a network node, base station, network element, network function, network entity, radio access network (RAN), NodeB, eNodeB (eNB), next-generation NodeB (gNB), or other suitable terms. NEs 102 and UEs 104 may communicate via a communication link, which may be a wireless or wired connection. For example, NEs 102 and UEs 104 may perform wireless communication (e.g., receiving signaling, transmitting signaling) via a Uu interface.

[0035] NE 102 can provide a geographic coverage area, and NE 102 can support service to one or more UEs 104 within the geographic coverage area. For example, NE 102 and UE 104 can support wireless communication of signals related to services (e.g., voice, video, packet data, message sending and receiving, broadcasting, etc.) according to one or more radio access technologies. In some embodiments, NE 102 can be mobile, such as a satellite associated with a non-terrestrial network (NTN). In some embodiments, different geographic coverage areas associated with the same or different radio access technologies can overlap, but different geographic coverage areas can be associated with different NEs 102.

[0036] One or more UEs 104 may be distributed throughout the geographic area of ​​the wireless communication system 100. UE 104 may include or be referred to as a remote unit, mobile device, wireless device, remote device, subscriber device, transmitter device, receiver device, or some other suitable term. In some embodiments, UE 104 may be referred to as a unit, station, terminal, or client, and other instances thereof. Additionally or alternatively, UE 104 may be referred to as an Internet of Things (IoT) device, an Internet of Everything (IoE) device, or a Machine Type Communication (MTC) device, and other instances thereof.

[0037] UE 104 may be able to support direct wireless communication with other UE 104 via a communication link. For example, UE 104 may support direct wireless communication with another UE 104 via a device-to-device (D2D) communication link. In some implementations, such as vehicle-to-vehicle (V2V) deployments, vehicle-to-everything (V2X) deployments, or cellular V2X deployments, the communication link may be referred to as a sidelink. For example, UE 104 may support direct wireless communication with another UE 104 via a PC5 interface.

[0038] NE 102 may support communication with CN 106, another NE 102, or both. For example, NE 102 may interface with other NE 102 or CN 106 via one or more backhaul links (e.g., S1, N2, N6, or other network interfaces). In some embodiments, NE 102 may communicate directly with each other. In other embodiments, NE 102 may communicate indirectly with each other (e.g., via CN 106). In some embodiments, one or more NE 102 may include sub-components such as access network entities, which may be instances of access node controllers (ANCs). The ANC may communicate with one or more UE 104s via one or more other access network transport entities, which may be referred to as radio headends, smart radio headends, or transmit-receive points (TRPs).

[0039] CN 106 can support user authentication, access authorization, tracking, connectivity, and other access, routing, or mobility functions. CN106 can be an evolved packet core (EPC) or a 5G core (5GC), which may include control plane entities that manage access and mobility (e.g., a mobility management entity (MME), access and mobility management functions (AMF)) and user plane entities that route packets or interconnects to external networks (e.g., a serving gateway (S-GW), a packet data network (PDN) gateway (P-GW), or a user plane function (UPF)). In some implementations, the control plane entities may manage non-access plane (NAS) functions, such as mobility, authentication, and bearer management (data bearers, signaling bearers, etc.) for one or more UEs 104 served by one or more NEs 102 associated with CN 106.

[0040] CN 106 can communicate with a packet data network via one or more backhaul links (e.g., via S1, N2, N6, or other network interfaces). The packet data network may contain an application server. In some implementations, one or more UEs 104 can communicate with the application server. UE 104 can establish a session (e.g., a Protocol Data Unit (PDU) session, etc.) with CN 106 via NE 102. CN 106 can use the established session (e.g., an established PDU session) to route services (e.g., control information, data, etc.) between UE 104 and the application server. A PDU session can be an instance of a logical connection between UE 104 and CN 106 (e.g., one or more network functions of CN 106).

[0041] In the wireless communication system 100, NE 102 and UE 104 can use the resources of the wireless communication system 100 (e.g., time resources (e.g., symbols, time slots, subframes, frames, etc.) or frequency resources (e.g., subcarriers, carriers)) to perform various operations (e.g., wireless communication). In some embodiments, NE 102 and UE 104 can support different resource structures. For example, NE 102 and UE 104 can support different frame structures. In some embodiments, such as in 4G, NE 102 and UE 104 can support a single frame structure. In some other embodiments, such as in 5G and other suitable radio access technologies, NE 102 and UE 104 can support various frame structures (i.e., multiple frame structures). NE 102 and UE 104 can support various frame structures based on one or more parameter sets (numerologies).

[0042] The wireless communication system 100 may support one or more parameter sets, and the parameter sets may include subcarrier spacing and cyclic prefixes. A first parameter set (e.g., μ=0) may be associated with a first subcarrier spacing (e.g., 15 kHz) and a normal cyclic prefix. In some embodiments, the first parameter set (e.g., μ=0) associated with the first subcarrier spacing (e.g., 15 kHz) may utilize one time slot per subframe. A second parameter set (e.g., μ=1) may be associated with a second subcarrier spacing (e.g., 30 kHz) and a normal cyclic prefix. A third parameter set (e.g., μ=2) may be associated with a third subcarrier spacing (e.g., 60 kHz) and a normal cyclic prefix or an extended cyclic prefix. A fourth parameter set (e.g., μ=3) may be associated with a fourth subcarrier spacing (e.g., 120 kHz) and a normal cyclic prefix. A fifth parameter set (e.g., μ=4) may be associated with a fifth subcarrier spacing (e.g., 240 kHz) and a normal cyclic prefix.

[0043] Time intervals for resources (e.g., communication resources) can be organized according to frames (also referred to as radio frames). Each frame may have a duration, such as 10 milliseconds (ms). In some embodiments, each frame may contain multiple subframes. For example, each frame may contain 10 subframes, and each subframe may have a duration, such as 1 ms. In some embodiments, each frame may have the same duration. In some embodiments, each subframe of a frame may have the same duration.

[0044] Alternatively or concurrently, the time intervals of resources (e.g., communication resources) can be organized according to time slots. For example, a subframe may contain a certain number (e.g., quantity) of time slots. The number of time slots in each subframe may also depend on one or more parameter sets supported in the wireless communication system 100. For example, a first parameter set, a second parameter set, a third parameter set, a fourth parameter set, and a fifth parameter set (i.e., μ=0, μ=1, μ=2, μ=3, μ=4) associated with corresponding subcarrier intervals of 15 kHz, 30 kHz, 60 kHz, 120 kHz, and 240 kHz may utilize one time slot per subframe, two time slots per subframe, four time slots per subframe, eight time slots per subframe, and 16 time slots per subframe, respectively. Each time slot may contain a certain number (e.g., quantity) of symbols (e.g., OFDM symbols). In some embodiments, the number (e.g., quantity) of time slots in a subframe may depend on the parameter set. For a normal cyclic prefix, a time slot may contain 14 symbols. For an extended cyclic prefix (e.g., applicable to a 60 kHz subcarrier spacing), a time slot may contain 12 symbols. The relationship between the number of symbols per time slot, the number of time slots per subframe, and the number of time slots per frame for both normal and extended cyclic prefixes may depend on the parameter set. It should be understood that a reference to the first parameter set (e.g., μ=0) associated with the first subcarrier spacing (e.g., 15 kHz) may be used interchangeably between subframes and time slots.

[0045] In the wireless communication system 100, the electromagnetic (EM) spectrum can be divided into various categories, frequency bands, channels, etc., based on frequency or wavelength. For example, the wireless communication system 100 may support one or more operating frequency bands, such as frequency ranges specified as FR1 (410 MHz to 7.125 GHz), FR2 (24.25 GHz to 52.6 GHz), FR3 (7.125 GHz to 24.25 GHz), FR4 (52.6 GHz to 114.25 GHz), FR4a or FR4-1 (52.6 GHz to 71 GHz), and FR5 (114.25 GHz to 300 GHz). In some embodiments, NE 102 and UE 104 may perform wireless communication on one or more of the said operating frequency bands. In some embodiments, FR1 may be used by NE 102 and UE 104, as well as other equipment or devices for cellular communication services (e.g., control information, data). In some implementations, FR2 may be used by NE 102 and UE 104, as well as other equipment or devices for short-range, high-data-rate capabilities.

[0046] FR1 may be associated with one or more parameter sets (e.g., at least three parameter sets). For example, FR1 may be associated with a first parameter set containing a 15 kHz subcarrier spacing (e.g., μ=0), a second parameter set containing a 30 kHz subcarrier spacing (e.g., μ=1), and a third parameter set containing a 60 kHz subcarrier spacing (e.g., μ=2). FR2 may be associated with one or more parameter sets (e.g., at least two parameter sets). For example, FR2 may be associated with a third parameter set containing a 60 kHz subcarrier spacing (e.g., μ=2) and a fourth parameter set containing a 120 kHz subcarrier spacing (e.g., μ=3).

[0047] One or more devices (e.g., nodes) of the wireless communication system 100 (e.g., UE 104 and / or NE 102) can transmit and receive signaling via a wireless channel. In variations, the node may implement one or more machine learning models for decoding signaling, performing channel estimation procedures, performing beam selection procedures, and other implementations. In some instances, the machine learning model may have a generalization error value, where the generalization error represents the measured difference between the performance of the machine learning model on training data and the performance of the machine learning model on test or validation data. The node may implement one or more techniques to reduce the generalization error of the machine learning model implemented at the receiver device of the node. For example, the node may increase the number of numerical values ​​of the training data samples used to train the machine learning model. However, in some instances, collecting a sufficiently large training dataset (e.g., greater than a threshold) may result in additional processing and resource usage at the node. For example, the node may obtain additional data samples from other nodes and / or generate additional data samples, resulting in increased signaling overhead and power consumption at the node. Furthermore, the training time for machine learning models can increase with the size of the training dataset, leading to delays associated with implementing and / or deploying the machine learning model. Alternatively, users and / or nodes may manually update one or more parameters of the machine learning model to reduce generalization error. However, manually updating one or more parameters of the machine learning model may depend on the implementation scheme (e.g., by the user and / or node), and if the selected parameters do not reduce generalization error, then errors will occur.

[0048] In some instances, to reduce generalization error and prevent or reduce errors caused by manually selecting one or more parameters of the machine learning model and / or to prevent or reduce inefficiencies caused by collecting additional training data for training the machine learning model, a node can use a subset of data samples to train multiple instances of the machine learning model. In addition to or as an alternative to the techniques described above for reducing generalization error, a node can use a subset of data samples to train multiple instances of the machine learning model. In some cases, a node can divide the data sample set into one or more subsets. A node can select a machine learning model and initialize one or more weights and / or parameters of the machine learning model. In variations, a node can initialize the weights and / or parameters of multiple instances of the machine learning model (e.g., create identical instances of the machine learning model). A node can train machine learning model instances using appropriate subsets of the training dataset. A node can then aggregate the parameters of the machine learning model instances to obtain a global machine learning model. In some instances, once a node obtains a global machine learning model, it can then implement the global machine learning model at its receiving device (e.g., receiver and / or decoder) to decode signaling corrupted by the wireless channel, perform channel estimation, and / or perform beam selection.

[0049] Figure 2 The diagram illustrates examples of a machine learning schema 200 according to various aspects of this disclosure. In some instances, the machine learning schema 200 implements various aspects of the wireless communication system 100. For example, the machine learning schema 200 may be implemented by nodes such as UEs and / or NEs, which may be as described in the reference... Figure 1 Instances of UE 104 and / or NE 102 are described. Machine learning schema 200 can graphically illustrate the generation of instances of the global machine learning model 206 from the aggregated parameters 208 of the global machine learning model 206 using several subsets of data samples from data sample set 204 (e.g., subsets 202-a to 202-T, where T is any integer value). Although Figure 2 The illustration shows an instance of a single global machine learning model 206, but the global machine learning model 206 may additionally or alternatively include or implement any numerical number of machine learning models.

[0050] One or more wireless devices (e.g., UE and / or NE) in a wireless communication system can transmit and receive signaling via a wireless channel. A wireless device may be referred to as a network node, a wireless node, or a node. A node can receive signaling, including data and / or control information, from another node. In some instances, a node transmitting signaling may be referred to as a transmitting node or a transmitting wireless device, and a node receiving signaling may be referred to as a receiving node or a receiving wireless device. In some instances, a transmitting node and / or a receiving node may include one or more receiving devices, such as receivers and / or decoders.

[0051] In some instances, machine learning models, such as deep neural networks (DNNs), are developed in response to training the model using one or more datasets. The datasets may be labeled or unlabeled, resulting in supervised and / or learning or unsupervised training and / or learning, respectively. For machine learning models used in wireless communications, the model is developed based on training and testing using a dataset (e.g., data sample set 204), which is constructed from simulated data or real-world data collected from an operating wireless network, or a combination of simulated and real-world data. Once the machine learning model is developed (e.g., trained and tested using available data during the development phase), it is deployed or implemented (e.g., deployed in the wireless network) at the receiver of the node. In variations, the node may implement one or more machine learning models for decoding signaling, performing channel estimation procedures, performing beam selection procedures, and other implementations.

[0052] In some instances, a machine learning model can be implemented by the receiver device to perform various aspects and / or functions of the receiver device in a wireless communication system. For example, a receiver device (e.g., a wireless receiver and / or decoder) can be represented as a classifier that categorizes input (e.g., one or more corrupted symbols from channel output) into one or more transmitted symbols and / or transmitted messages from a defined list of symbols and / or messages. In some cases, discrete message sets... The data is transmitted (e.g., transmitted or received by a node) from one node to another via a wireless channel h. For example, a base station (e.g., NE) may transmit a message to a UE, which may be referred to as a downlink message. In some other instances, the UE may transmit a message to the NE, which may be referred to as an uplink message. The wireless channel is time-varying (e.g., due to the movement of one or more nodes in the wireless communication system and other factors), and the value of the wireless channel at time t is represented as h. tA wireless channel can be characterized by the probability density function (PDF) of the channel gain. For example, a wireless channel can be modeled using Rayleigh fading, a statistical model representing the impact of the propagation environment on signaling. A Rayleigh fading channel corresponds to... A transmitter device (e.g., a transmitter) encodes a message into one or more resources in the time domain, called symbols. For example, a transmitter device encodes a message... Encoded into symbols suitable for use in the channel. (For example, where d = 2), where This is referred to as a signal constellation. In some cases, the mapping from message set to symbol is a one-to-one mapping. Discrete symbols are mapped to analog waveforms and transmitted via a wireless channel. Wireless channel corruption and / or symbol transformation occur, where y i Indicates input x to the wireless channel i The wireless channel output. The receiver device (e.g., a receiver and / or encoder at the node receiving signaling) outputs from y i Recovery message m i .

[0053] In some instances, the receiver device implements one or more mathematical models to recover transmitted messages from the output of a damaged channel. In other instances, in addition to or as an alternative to mathematical models, the receiver device may implement one or more machine learning models to recover transmitted messages from the output of a damaged channel. Machine learning models can be trained on the receiver device, for example, using supervised learning from y i get That is, the estimated value of the transmitted symbols, to minimize the error probability (e.g., using a number of labeled data samples greater than a threshold or a sufficiently large number of values). The error probability can be calculated according to Equation 1:

[0054]

[0055] in This is the prior probability of the input message. The labeled training data samples are... Or, or equivalently, is In some instances, the machine learning model trained as a receiver device is represented as D. W , where w is a set of trainable and / or learnable parameters. Exemplary trainable and / or learnable parameters may include, but are not limited to, weights or coefficients of connections between network nodes (e.g., neural network nodes) applied to a machine learning model, biases of connections between network nodes applied to a machine learning model, and other parameters. Receiver device The receiver device is trained to decode the wireless channel output to reproduce the corresponding input at the encoder. In other words, the decoder is trained to classify the channel output into one of a set of messages (e.g., one of M messages) in a way that minimizes the average error probability in the classification.

[0056] In some instances, machine learning models are trained based on a training data sample set. Although the number of training data samples is relatively large (e.g., greater than a threshold number of data samples), the training data does indeed include all the data samples encountered by the machine learning model when it is deployed after the training phase. The period for deploying the machine learning model can be referred to as the inference phase. During the inference phase, the machine learning model outputs the channel output y. i Classified into one of M messages and the machine learning model can encounter channel outputs (y) that were not included in the training data during the training phase. i While classifying such channel outputs, the machine learning model may output one or more misclassified messages (e.g., one or more errors) from the channel outputs, resulting in an error probability P. e P higher than during the training phase e .

[0057] In supervised learning and / or training, machine learning models are trained by nodes using labeled data samples. For example, making Denotes the training dataset, where It is the kth training sample, and y k Indicates the corrupted symbol at the channel output and x k This represents the label of the transmitted symbol. The label can also be equal to the transmitted message, meaning the training data sample Z... k It can also be Given. Note that for all values ​​of k, and Training data samples ,in This represents the space of training samples. Nodes can use supervised learning algorithms to determine one or more parameters of a machine learning model. For example, a node can use a supervised learning algorithm to determine the parameters of a machine learning model. Represents one or more machine learning model weights, where The space representing the weights of a machine learning model (e.g., This makes the overall risk... Minimize. The overall risk is given by Equation 2:

[0058]

[0059] in, It is a non-negative loss function. Describe the expectation operator and The indicator sample Z is distributed according to the PDF distribution μ. In some instances, Let represent the set of positive real numbers. Therefore, supervised learning methods can determine the overall risk as given by Equation 2. Minimize one or more machine learning model parameters. However, PDF and μ are unknown, which hinders the assessment of the overall risk. The direct calculation. Therefore, substitution increases the overall risk. Minimize the empirical risk on the training dataset S. minimize.

[0060] Empirical risk is defined according to Equation 3:

[0061]

[0062] Minimizing empirical risk rather than population risk leads to generalization error in supervised learning. Generalization error is defined as the expected difference between population risk and empirical risk and is given by Equation 4:

[0063]

[0064] In some instances, the generalization error depends on the loss function. The training dataset S (e.g., the diversity and size of the samples it contains) ), distribution μ and the learning algorithm used by the nodes.

[0065] In some instances, once a node determines the parameters of a machine learning model, it can deploy the model by configuring it with those parameters. The node can provide inputs to the machine learning model to obtain one or more output values. In some instances, because the node provides one or more inferences and / or predictions during the testing phase of the machine learning model (e.g., before deployment), the deployed model is expected to perform at a defined level of accuracy and / or precision. The training dataset may contain a finite number of data samples (e.g., it may not represent all possible types of data samples). Therefore, in real-world wireless networks, the machine learning model can make predictions and / or inferences based on input data that was not used to train the model during the training phase. Thus, the node can minimize the generalization error of the machine learning model to reduce errors associated with machine learning models provided with input data different from the training data when deployed.

[0066] In some contexts, generalization error measures the difference between the performance of a machine learning model on training data and its performance on test or validation data. Performance can be represented by one or more metrics and / or measurements, such as the accuracy, precision, recall, and / or F1 score of the machine learning model. In some instances, accuracy is the ratio of the number of correctly predicted instances to the total number of instances. In some instances, precision is the ratio of the number of correctly predicted positive instances to the total number of predicted positive instances. In some instances, recall is the ratio of the number of correctly predicted positive instances to the total number of instances. In some instances, the F1 score is the harmonic mean of precision and recall. In some cases, low generalization error indicates that the machine learning model has successfully generalized its learning process from the training dataset to new instances in data not included in the training dataset, while high generalization error indicates that the machine learning model may be overfitting the training dataset and performing poorly on new instances in data not included in the training dataset.

[0067] In some instances, a node may implement one or more techniques to reduce the generalization error of a machine learning model implemented at the node's receiver device. For example, a node may increase the number of numerical training data samples used to train the machine learning model. Increasing the number of numerical training data samples allows the node to train the machine learning model on a majority of typical data samples, enabling the machine learning model to be trained on a majority of data samples and receive these samples as input when deployed after training. However, in some instances, collecting a sufficiently large training dataset (e.g., greater than a threshold) may result in additional processing and resource usage at the node. For example, the node may obtain additional data samples from other nodes and / or generate additional data samples, leading to increased signaling overhead and power consumption at the node. Furthermore, the training time of the machine learning model may increase with the size of the training dataset, resulting in latency associated with implementing and / or deploying the machine learning model.

[0068] Alternatively, users and / or nodes may manually update one or more parameters of the machine learning model to reduce generalization error. For example, users and / or nodes may introduce regularization (e.g., adding constraints) during the training of the machine learning model to ensure that the machine learning model does not overfit the training data, thereby increasing the generalization ability of the machine learning model. However, regularization may be implementation-dependent (e.g., by the user and / or node), and if the machine learning model is not properly regularized, it will lead to error. Alternatively, users and / or nodes may choose the model size and complexity to reduce generalization error. For example, a small machine learning model (e.g., features and / or parameters with a threshold number of values) may not be able to generalize to data samples not provided in the training data, resulting in higher generalization error. In some other cases, a relatively large machine learning model (e.g., features and / or parameters with a number of values ​​greater than a threshold) may learn the training data with very low training error, but may not be able to generalize to data samples not provided in the training data. By choosing a machine learning model size that meets a threshold (e.g., not too small and not too large), generalization error can be reduced. However, the choice of machine learning model size may depend on the implementation (e.g., by the user and / or nodes), and choosing an incorrectly sized machine learning model (e.g., a model size that increases generalization error) will result in errors. Alternatively, the user and / or nodes may add a relatively small amount of noise (e.g., less than a threshold) to the input samples during training to improve the performance of the machine learning model on data samples not provided in the training data. However, the choice of the noise amount may depend on the implementation (e.g., by the user and / or nodes), and choosing an incorrect amount of noise (e.g., by selecting an amount that increases generalization error) will result in errors.

[0069] In some instances, to reduce generalization error and prevent or reduce errors caused by manually selecting one or more parameters of the machine learning model and / or to prevent or reduce inefficiencies caused by collecting additional training data for training the machine learning model, a node may use a subset of data samples to train multiple instances of the machine learning model. In addition to or as an alternative to the techniques described above for reducing generalization error, a node may use a subset of data samples to train multiple instances of the machine learning model. In some cases, a node may divide the data sample set 204 into one or more subsets, including data sample subset 202-a, data sample subset 202-b, and one or more additional subsets (e.g., up to data sample subset 202-T), such that there are T subsets of data samples. The value of T can be any integer value.

[0070] In some cases, the data sample set 204 can be the training data sample set S, where And K is greater than the threshold. Supervised learning methods for machine learning models may involve selecting a machine learning model by fixing the number of layers, the number of neurons in each layer, and one or more activation functions, such as a support vector machine (SVM) model with a core and / or neural network. Once a machine learning model is selected, a learning or optimization algorithm is chosen. (e.g., stochastic gradient descent (SGD) or adaptive moment estimation (ADAM)), and a learning rate α is also chosen. In some instances, the loss function... Selected for use in a machine learning model. One or more weights and / or parameters of the machine learning model are initialized, where the weights are represented by W and Alternatively, other hyperparameters of the machine learning model are fixed, where hyperparameters are configuration settings of the machine learning model that are not learned from the training dataset (learning rate, number of epochs, batch size, architecture-related parameters, regularization parameters, etc.). In some instances, empirical risk is introduced during supervised training. Minimize to find (e.g., optimize) machine learning model parameters .

[0071] In some instances, to reduce generalization error, a node may split the training dataset S (e.g., data sample set 204) into two or more smaller training datasets, such that the node trains a machine learning model instance (e.g., the same but separate machine learning model) using the corresponding smaller training dataset by minimizing the empirical risk on one of the smaller training datasets. The node may then aggregate the parameters of the machine learning model instances to obtain a global machine learning model 206. In some instances, although the diagram illustrates more than two data sample subsets and machine learning model instances, the data sample set 204 may be split into any numerical number of data sample subsets (e.g., two or more), and the node may train any numerical number of machine learning model instances (e.g., two or more).

[0072] In some cases, nodes split the training dataset S into two or more subsets of training data samples S1, S2, ..., S3. T subset S i , In the variant, the data sample subset contains a portion of the training data samples from the original training dataset, making... ,in Representation set S i cardinality (e.g., Representation set S iThe total number of elements in the dataset. A node can select a number T of machine learning model instances to train using a subset of the data samples. For example, a node can use machine learning model instance 210-a, machine learning model instance 210-b, and / or one or more additional machine learning model instances (e.g., up to machine learning model instance 210-T). In some instances, the node can choose the value of T when partitioning and / or splitting the data sample set 204 into subsets 202-a to 202-T. That is, the number of machine learning model instances can be based on the number of values ​​in the data sample subset. In some other instances, the node can choose the number of machine learning model instances before partitioning and / or splitting the data sample set 204 into subsets 202-a to 202-T. In some instances, the machine learning model instances can represent a neural network with a fixed number of layers, a fixed number of neurons in each layer, and a fixed number of activation functions. In some other instances, the machine learning model instances can represent an SVM model with a defined core. In other instances, the machine learning model can be any type of artificial intelligence and / or machine learning model. In some cases, the corresponding instances of machine learning model instances (e.g., machine learning model instances 210-a to 210-T) can be identical after initialization (e.g., before training). For example, the machine learning model instance is identical to the other T-1 remaining machine learning models. In some cases, W... i Representing model i ( The parameter vector or weight vector of ), where .

[0073] In some cases, nodes can choose T number of learning or optimization algorithms. For example, SGD or ADAM and their corresponding learning rate α i , i = 1, ..., T. In some instances, and for It can be the same value and / or it can be different values. Alternatively, the learning rate α... i and α j for The values ​​can be the same and / or different. Nodes can choose a loss function. And it can initialize the weights and / or parameters of the corresponding machine learning model instances (e.g., the weights from machine learning model instance 210-a to machine learning model instance 210-T), so that Let i = 1, ..., T. In some cases, S is used. i As a training dataset and used in a learning algorithm Nodes can mitigate experiential risks. Minimize one or more weights and / or parameters of a machine learning model instance such that W i For i = 1, ..., T, the empirical risk is modeled according to Equation 5:

[0074]

[0075] in The training set S is represented by i The k-th training data sample in the dataset.

[0076] There are currently T machine learning models, where the i-th machine learning model has weights and / or parameters. For example, machine learning model instance 210-a may have a parameter set 212-a (e.g., weights and / or other parameters), machine learning model instance 210-b may have a parameter set 212-b, and one or more machine learning model instances may have corresponding parameter sets (e.g., parameter set 212-T up to machine learning model instance 210-T). In some instances, a node may obtain a global and / or holistic machine learning model, for example, with weights and / or parameters. The global machine learning model 206. For example, nodes can be aggregated, for instance, according to the following equation 6. (i = 1, ..., T) and obtain aggregated parameters 208 from the parameter set of the machine learning model instance (e.g., parameter set 212-a to parameter set 212-T):

[0077]

[0078] One or more weighting parameters β are selected. i , making .

[0079] In some instances, once a node obtains the global machine learning model 206 (e.g., by obtaining aggregated parameters 208 and applying them to the global machine learning model 206), the node can implement the global machine learning model 206 at its receiving device (e.g., receiver and / or decoder) to decode signaling corrupted by the wireless channel, perform channel estimation, and / or perform beam selection. Although the global machine learning model 206 is described as being implemented at the node's receiving device, the global machine learning model 206 can be implemented by a transmitting device and / or used in any other use case in the wireless communication system (e.g., for other classification use cases in the wireless communication system).

[0080] Figure 3The illustrations illustrate examples of a wireless communication system 300 according to various aspects of this disclosure. In some examples, the wireless communication system 300 implements aspects of the wireless communication system 100 and / or the machine learning schema 200. For example, the wireless communication system 300 includes UE 104-a, UE 104-b, and NE 102, which may be as described in the references. Figure 1 Examples of UE 104 and NE 102 described herein. The wireless communication system 300 may also include a global machine learning model 206 having one or more aggregated parameters 208, which may be as referenced... Figure 2 Examples of the corresponding features described. In some instances, UE104-a, UE104-b and / or NE102 may be instances of nodes (e.g., wireless nodes and / or wireless devices in wireless communication system 300).

[0081] In some instances, NE 102 may wirelessly communicate with one or more other nodes in the wireless communication system 300. For example, NE 102 may transmit and / or receive signaling from one or more UEs (including UE 104-a and UE 104-b). NE 102 may transmit signaling to UE 104-a via uplink communication link 302. Alternatively, UE 104-a may transmit signaling to NE 102 via downlink communication link 304-a and / or UE 104-b may transmit signaling to NE 102 via downlink communication link 304-b. Signaling between nodes in the wireless communication system 300 (e.g., NE 102, UE 104-a, and / or UE 104-b) may include control signaling and / or data transmission. For example, the signaling may carry one or more messages of defined message types (e.g., in a list of defined message types). NE 102 may implement a global machine learning model 206 at its receiver device (e.g., receiver and / or decoder) to decode signaling, perform channel estimation, and / or perform beam selection procedures. The channel estimation procedure may include obtaining channel characteristics of the wireless channel from the global machine learning model 206 by providing one or more data values ​​(e.g., including one or more channel measurements) as input to the global machine learning model 206. The beam selection procedure may include selecting a transmit and / or receive beam for transmission by receiving an indication of a beam as an output from the global machine learning model 206. The input to the global machine learning model 206 for the beam selection procedure may include one or more measurements performed by the node, and other parameters defining the signaling between nodes in the wireless communication system 300.

[0082] In some instances, one or more other nodes (e.g., in addition to NE 102 or as an alternative to NE 102) may implement one or more machine learning models 306. These nodes may train their own machine learning models or receive aggregated parameter indications 308 (e.g., aggregated parameter 208) from another node. For example, UE 104-a may implement one or more machine learning models 306 and may receive aggregated parameter indications 308 from NE 102 via uplink communication link 302. NE 102 may include aggregated parameter indications 308 in control signaling sent to UE 104-a (e.g., in DCI messages, in RRC signaling, and / or in MAC-CE). In some cases, UE 104-a may apply the parameters to the machine learning model 306 at UE 104-a.

[0083] In some instances, before obtaining the aggregation parameter 208, NE 102 may receive and / or generate data samples from one or more nodes of the wireless communication system 300, where NE 102 uses the data samples to train a machine learning model, as seen in references. Figure 2 As described. For example, NE 102 may receive data sample 310-a from UE 104-a and / or may receive data sample 310-b from UE 104-b. UE 104-a may transmit data sample 310-a to NE 102 via downlink communication link 304-a. Alternatively, UE 104-b may transmit data sample 310-b to NE 102 via downlink communication link 304-b. In some cases, UE 104-a and / or UE 104-b may explicitly transmit data sample 310-a and / or data sample 310-b in control signaling. In other cases, NE 102 may receive signaling from UE 104-a and / or UE 104-b and may use said signaling to generate data samples.

[0084] In some cases, data sample 310-a and / or data sample 310-b may contain one or more values ​​representing the time resources (e.g., symbols) of the output of the wireless channel and / or values ​​representing the content of the output of the wireless channel. In some instances, although the global machine learning model 206 is illustrated as being implemented and / or deployed by NE 102, the global machine learning model 206 may be implemented or deployed by any node in the wireless communication system 300 (e.g., UE 104-a, UE 104-b, and / or NE 102). In some instances, implementing or deploying the global machine learning model 206 may involve obtaining the output of the wireless channel from the global machine learning model 206 (e.g., by providing one or more measured and / or received signaling symbols as inputs to the global machine learning model 206).

[0085] Figure 4 The diagram illustrates examples of signaling schema 400 according to various aspects of this disclosure. In some instances, signaling schema 400 may implement aspects of wireless communication system 100, machine learning schema 200, and / or wireless communication system 300. Signaling schema 400 may illustrate examples of aggregating parameters of a global machine learning model by training a machine learning model using a subset of a data sample set. Nodes 402-a and 402-b may be as referenced. Figures 1 to 3 Examples of the described wireless devices (e.g., UE 104 and / or NE 102). Alternative examples may be implemented, some of which may be performed in a different order than described or not performed at all. In some cases, processes may include additional features not mentioned below, or other processes may be added.

[0086] In some instances, at 402, node 402-b transmits a data sample set to node 402-a. In some instances, node 402-b transmits the data sample set in control signaling, such as downlink control signaling and / or uplink control signaling (e.g., uplink control information (UCI) messages, DCI messages, RRC signaling, MAC-CE, or any other type of control signaling). Node 402-b may be a UE or an NE, and node 402-a may be a UE or an NE. In some instances, data samples constitute a data sample set. The data samples contain a first value representing the output from the radio channel (e.g., symbol x). i ) and a second value representing the message transmitted via the output from the wireless channel (e.g., symbol y). i At least one of the following. For example, node 402-b can transmit message m. i Encoding to x i In, and transmitted via wireless channel x i The channel output is y. i It may contain x i The corrupted version. Therefore, the input to the wireless channel is x. i And the output is the corresponding damaged signal y. i Therefore, the first value can represent y. i (e.g., output symbols from a wireless channel), and the second value may be contained in y i The message in the symbol (e.g., contained in y) i (The transmitted message). In some instances, node 402-a may generate one or more data samples from a data sample set (e.g., from signaling received from one or more other nodes). For example, node 402-a may generate data samples in addition to receiving data samples in explicit signaling from another node, or as an alternative.

[0087] At 404, node 402-a can obtain a subset of data samples. For example, node 402-a can segment or divide the data sample set into multiple subsets. Node 402-a can divide the data sample set into any number of subsets, such as two or more data samples. That is, a subset of data samples may contain at least a first subset and a second subset. In a variant, a subset of data samples contains at least one type of data sample from different types of data samples in the data sample set. For example, if the data sample set contains data sample types A, B, and C, then any subset of data samples may contain at least one data sample of type A, at least one data sample of type B, and at least one data sample of type C. Exemplary data sample types include, but are not limited to, different types of channel outputs (e.g., different types of messages, including data transmission, control signaling, and other types of messages) or one or more measurements performed on signaling received from another node. In some instances, different types of messages may be messages from a defined or configured list of message types transmitted by nodes in a wireless communication system. In some instances, the message is sent by another node, such as from node 402-b and / or from any other node (e.g., a third node). In some instances, different subsets of the data sample contain the same or similar number of data samples. That is, the corresponding number of values ​​of the data samples in the subsets of the data sample satisfies a threshold, wherein the threshold is based on the total number of values ​​of the data samples in the data sample set and the number of values ​​of the subsets of the data sample.

[0088] At 406, node 402-a can determine one or more parameters of the machine learning model. For example, node 402-a can aggregate or combine one or more parameters from different instances of the machine learning model to produce a global machine learning model with aggregated or combined parameters.

[0089] At point 408, node 402-a can use a subset of data samples to train an instance of a machine learning model to obtain a parameter set. Node 402-a can obtain multiple parameter sets for a given subset of data samples. For example, node 402-a can obtain a first parameter set in response to training a first instance of a machine learning model using supervised learning techniques and a first subset of data samples. Alternatively, node 402-a can obtain a second parameter set in response to training a second instance of a machine learning model using supervised learning techniques and a second subset of data samples. That is, node 402-a can determine T parameter sets W1, W2, ..., W... T One of the sets contains N parameters. This is achieved by using the i-th data sample subset S. iThe i-th set W in the T parameter sets is determined using a supervised learning method as the training dataset. i The values ​​of the N parameters in Node 402-a. Any number of instances of the machine learning model that can be trained. In some cases, instances of the machine learning model can be initialized with the same settings and / or parameters (e.g., so the instances of the machine learning model are identical before training).

[0090] At position 410, node 402-a can combine parameter sets. For example, node 402-a can combine or aggregate the T parameter sets W1, W2, ..., Wn for T instances used in a machine learning model. T The value of the i-th parameter in the parameter set W used for the global machine learning model is calculated by taking the value of the i-th parameter in each of the parameters.

[0091] At 412, node 402-a can transmit an indication of combined parameters to node 402-b. For example, node 402-a can transmit signaling (e.g., control signaling) indicating the parameters to node 402-b. The parameters can define a global machine learning model, such as one or more weights and / or biases of the global machine learning model. The global machine learning model can be a neural network and / or an SVM model.

[0092] In some instances, at 414, nodes 402-a and / or 402-b may transmit and / or receive signaling via a wireless channel. The receiving devices (e.g., receivers and / or decoders) of nodes 402-a and / or 402-b may use combined parameters to implement a global machine learning model.

[0093] In some instances, at 416 and / or 418, nodes 402-a and / or 402-b may obtain outputs indicating decoded signaling from a machine learning model (e.g., a global machine learning model). For example, the output may be included in the message sent in the signaling at 414. In other instances, nodes 402-a and / or 402-b may obtain outputs for performing channel estimation procedures and / or beam selection procedures. Nodes 402-a and / or 402-b may use the outputs from the machine learning model to perform channel estimation procedures and / or beam selection procedures. For example, nodes 402-a and / or 402-b may select a beam or update one or more communication parameters in the beam selection procedure based on measurements from the channel estimation procedure.

[0094] In some instances, the beam selection procedure may include measuring the radio channel to evaluate the quality or performance of different beams. The quality or performance of different beams may be measured based on the received signal strength and / or signal-to-noise ratio (SNR) and other characteristics of the radio channel. The NE or other nodes may transmit signals to another node (e.g., the UE), and the other node may perform one or more channel estimations to estimate the characteristics of the radio channel for different beam directions. In some instances, the other node transmits feedback to the NE, indicating information about beams that satisfy one or more threshold channel characteristic values ​​(e.g., the beam with the highest performance and / or transmission quality). This information may include the beam index of the beam or one or more channel quality indicators (CQIs) and other parameters. The NE selects beams for communication for one or more nodes based on the feedback from the respective nodes. The NE may provide information collected from different nodes into a machine learning model, and the machine learning model may output beam selections for the respective nodes.

[0095] In some instances, a channel estimation procedure (e.g., an estimation procedure for a wireless channel) may involve one or more nodes (e.g., node 402-a and / or node 402-b) measuring the characteristics of signaling transmitted via the wireless channel. The signaling may include one or more reference signal transmissions. The nodes may evaluate the measurements to obtain the channel characteristics of the wireless channel. Channel characteristics may include path loss (defined as the reduction in signal strength as a signal travels through the wireless channel), delay spread, Doppler shift, SNR or interference, and other characteristics. The nodes may update one or more communication parameters based on the channel characteristics. Communication parameters may include modulation and coding schemes (MCS), beamforming parameters (beamforming direction, beamwidth, antenna element spacing, etc.), or transmission power, and other transmission parameters. The nodes (e.g., node 402-a and / or node 402-b) may provide the measurements obtained from the wireless channel to a machine learning model, and the machine learning model may output updated communication parameters based on the measurements.

[0096] Figure 5 The diagram illustrates an example of a UE 500 according to various aspects of this disclosure. The UE 500 may include a processor 502, a memory 504, a controller 506, and a transceiver 508. The processor 502, memory 504, controller 506, or transceiver 508, or various combinations thereof, or various components thereof, may be examples of components for performing the various aspects of this disclosure as described herein. These components may be coupled via one or more interfaces (e.g., operational ground, communication ground, functional ground, electronic ground, electrical ground). In some instances, the UE 500 may be an example of a node, as referenced... Figures 1 to 4 As described.

[0097] Processor 502, memory 504, controller 506, or transceiver 508, or various combinations or components thereof, may be implemented in hardware (e.g., a circuit system). The hardware may include a processor, digital signal processor (DSP), application-specific integrated circuit (ASIC), or other programmable logic device, or any combination thereof, configured or otherwise supporting components for performing the functions described in this disclosure.

[0098] Processor 502 may include intelligent hardware devices (e.g., a general-purpose processor, DSP, CPU, ASIC, FPGA, or any combination thereof). In some embodiments, processor 502 may be configured to operate memory 504. In some other embodiments, memory 504 may be integrated into processor 502. Processor 502 may be configured to execute computer-readable instructions stored in memory 504, thereby enabling UE 500 to perform various functions of this disclosure.

[0099] Memory 504 may include volatile or non-volatile memory. Memory 504 may store computer-readable, computer-executable code containing instructions that, when executed by processor 502, cause UE 500 to perform the various functions described herein. The code may be stored in a non-transitory computer-readable medium, such as memory 504 or another type of memory. Computer-readable medium includes both non-transitory computer storage media and communication media, wherein the communication media includes any medium that facilitates the transfer of a computer program from one location to another. Non-transitory storage media may be any available medium accessible by a general-purpose or special-purpose computer.

[0100] In some implementations, processor 502 and memory 502 coupled to processor 504 may be configured to cause UE 500 to perform one or more of the functions described herein (e.g., processor 502 executing instructions stored in memory 504). For example, according to the examples disclosed herein, processor 502 may support wireless communication at UE 500. UE 500 may be configured or operable to support components for performing: receiving a set of data samples associated with a wireless channel from at least one second node; obtaining a subset of data samples from the set of data samples, wherein the subset of data samples includes at least a first subset of data samples and a second subset of data samples; and determining one or more parameters from the subset of data samples of a machine learning model of a wireless device associated with the first node.

[0101] Additionally, UE 500 may be configured to support any or a combination of the following: transmitting signaling to the at least one second node indicating one or more parameters of the machine learning model. Alternatively, a corresponding data sample in the data sample set includes at least one of a first value representing an output corresponding to the wireless channel and a second value representing a message associated with the output corresponding to the wireless channel. Alternatively, the message corresponds to a transmission from one or more of the at least one second node or at least one third node. Alternatively, to determine the one or more parameters of the machine learning model, UE 500 may be configured to: obtain a first parameter set associated with the first data sample subset in response to training a first instance of the machine learning model using supervised learning techniques based on the first data sample subset; obtain a second parameter set associated with the second data sample subset in response to training a second instance of the machine learning model using the supervised learning techniques based on the second data sample subset; and obtain the one or more parameters by combining the first parameter set associated with the first data sample subset and the second parameter set associated with the second data sample subset. Alternatively or concurrently, the one or more parameters define the machine learning model.

[0102] Alternatively or alternatively, the data sample set comprises one or more different data sample types, and a corresponding subset of data samples comprises at least one data sample of the one or more different data sample types. Alternatively or alternatively, the number of corresponding values ​​of data samples in the subset of data samples satisfies a threshold, and wherein the threshold is at least partially based on the total number of values ​​of data samples in the data sample set and the number of values ​​in the subset of data samples. Alternatively or alternatively, the UE 500 may be configured to support: obtaining an output associated with at least a portion of the received transmission from the machine learning model based on inputting an additional data sample set associated with the received transmission into the machine learning model, wherein the at least portion of the received transmission corresponds to corrupted signaling; and decoding the received transmission at least partially based on obtaining the at least portion of the received transmission. Alternatively, the UE 500 may be configured to support: obtaining an output from the machine learning model associated with at least one of a beam selection procedure or an estimation procedure, which is associated with the radio channel, based on inputting an additional set of data samples into the machine learning model; and selecting a beam for the beam selection procedure or updating one or more communication parameters, at least in part, based on the output. Alternatively, the radio device may include at least one of a receiver or a decoder.

[0103] Alternatively, UE 500 may support at least one memory (e.g., memory 504) and at least one processor (e.g., processor 502), the at least one processor being coupled to the at least one memory and configured to cause the UE to: receive a set of data samples associated with a wireless channel from at least one second node; obtain a subset of data samples from the set of data samples, wherein the subset of data samples includes at least a first subset of data samples and a second subset of data samples; and determine one or more parameters of a machine learning model of a wireless device associated with the first node from the subset of data samples.

[0104] Additionally, the UE 500 may be configured to support any or a combination of the following: the at least one processor is configured to transmit signaling indicating one or more parameters of the machine learning model to the at least one second node. Alternatively, a corresponding data sample in the data sample set includes at least one of a first value representing an output corresponding to the wireless channel and a second value representing a message associated with the output corresponding to the wireless channel. Alternatively, the message corresponds to a transmission from one or more of the at least one second node or at least one third node. Alternatively, to determine the one or more parameters of the machine learning model, the at least one processor is configured to: obtain a first parameter set associated with the first data sample subset in response to training a first instance of the machine learning model using supervised learning techniques based on the first data sample subset; obtain a second parameter set associated with the second data sample subset in response to training a second instance of the machine learning model using the supervised learning techniques based on the second data sample subset; and obtain the one or more parameters by combining the first parameter set associated with the first data sample subset and the second parameter set associated with the second data sample subset. Alternatively or concurrently, the one or more parameters define the machine learning model.

[0105] Alternatively or alternatively, the data sample set comprises one or more different data sample types, and a corresponding subset of data samples comprises at least one data sample of the one or more different data sample types. Alternatively or alternatively, the number of corresponding values ​​of the data samples in the subset of data samples satisfies a threshold, and wherein the threshold is based on the total number of values ​​of the data samples in the data sample set and the number of values ​​of the subset of data samples. Alternatively or alternatively, the at least one processor is configured to: obtain an output from the machine learning model associated with at least a portion of the received transmission based on inputting an additional data sample set associated with the received transmission into the machine learning model, wherein the at least portion of the received transmission corresponds to corrupted signaling; and decode the received transmission at least in part based on obtaining the at least portion of the received transmission. Alternatively or alternatively, the at least one processor is configured to: obtain an output from the machine learning model associated with at least one of a beam selection procedure or an estimation procedure, which is associated with the wireless channel, based on inputting an additional set of data samples into the machine learning model; and select a beam for the beam selection procedure or update one or more communication parameters based on the output. Alternatively or alternatively, the wireless device includes at least one of a receiver or a decoder.

[0106] Controller 506 manages the input and output signals of UE 500. Controller 506 can also manage peripheral devices not integrated into UE 500. In some embodiments, controller 506 may utilize an operating system such as iOS®, ANDROID®, WINDOWS®, or other operating systems. In some embodiments, controller 506 may be implemented as part of processor 502.

[0107] In some embodiments, UE 500 may include at least one transceiver 508. In other embodiments, UE 500 may have more than one transceiver 508. Transceiver 508 may represent a wireless transceiver. Transceiver 508 may include one or more receiver chains 510, one or more transmitter chains 512, or a combination thereof.

[0108] Receiver chain 510 may be configured to receive signals (e.g., control information, data, packets) via wireless media. For example, receiver chain 510 may include one or more antennas for receiving signals via air or wireless media. Receiver chain 510 may include at least one amplifier (e.g., a low-noise amplifier (LNA)) configured to amplify the received signal. Receiver chain 510 may include at least one demodulator configured to demodulate the received signal and obtain transmitted data by reversing the modulation technique applied during signal transmission. Receiver chain 510 may include at least one decoder for decoding the demodulated signal to receive transmitted data.

[0109] Transmitter chain 512 can be configured to generate and transmit signals (e.g., control information, data, packets). Transmitter chain 512 may include at least one modulator for modulating data onto a carrier signal, thereby preparing the signal for transmission over a wireless medium. The at least one modulator may be configured to support one or more techniques, such as amplitude modulation (AM), frequency modulation (FM), or digital modulation schemes like phase shift keying (PSK) or quadrature amplitude modulation (QAM). Transmitter chain 512 may also include at least one power amplifier configured to amplify the modulated signal to an appropriate power level suitable for transmission over a wireless medium. Transmitter chain 512 may also include one or more antennas for transmitting the amplified signal over the air or in the wireless medium.

[0110] Figure 6 The illustration depicts an example of a processor 600 according to various aspects of this disclosure. The processor 600 may be an example of a processor configured to perform various operations according to the examples described herein. The processor 600 may include a controller 602 configured to perform various operations according to the examples described herein. The processor 600 may optionally include at least one memory 604, which may be, for example, an L1 / L2 / L3 cache memory. Alternatively or additionally, the processor 600 may optionally include one or more arithmetic logic units (ALUs) 606. One or more of these components may be electronically communicated or otherwise coupled (e.g., operative ground, communicative ground, functional ground, electronic ground, electrical ground) via one or more interfaces (e.g., buses). In some instances, the processor 600 may be implemented as a node or may be implemented by a node, as referenced... Figures 1 to 4 As described.

[0111] Processor 600 may be a processor chipset and includes a protocol stack (e.g., a software stack) executed by the processor chipset to perform various operations (e.g., receiving, obtaining, retrieving, transmitting, outputting, forwarding, storing, determining, identifying, accessing, writing, and reading) according to the examples described herein. The processor chipset may include one or more cores, one or more cache memories (e.g., memory native to or contained within the processor chipset (e.g., processor 600), or other memories (e.g., random access memory (RAM), read-only memory (ROM), dynamic RAM (DRAM), synchronous dynamic RAM (SDRAM), static RAM (SRAM), ferroelectric RAM (FeRAM), magnetic RAM (MRAM), resistive RAM (RRAM), flash memory, phase-change memory (PCM), and others)).

[0112] The controller 602 can be configured to manage and coordinate various operations of the processor 600 (e.g., signaling, receiving, acquiring, retrieving, transmitting, outputting, forwarding, storing, determining, identifying, accessing, writing, and reading) to enable the processor 600 to support various operations according to the examples described herein. For example, the controller 602 can operate as a control unit of the processor 600, thereby generating control signals that manage the operation of various components of the processor 600. These control signals include enabling or disabling functional units, selecting data paths, initiating memory accesses, and coordinating operation timing.

[0113] Controller 602 may be configured to fetch (e.g., obtain, retrieve, receive) instructions from memory 604 and determine subsequent instructions to be executed to enable processor 600 to support various operations according to the examples described herein. Controller 602 may be configured to track the memory addresses of instructions associated with memory 604. Controller 602 may be configured to decode instructions to determine the operation to be performed and the operands involved. For example, controller 602 may be configured to interpret instructions and determine control signals to be output to other components of processor 600, thereby enabling processor 600 to support various operations according to the examples described herein. Alternatively or additionally, controller 602 may be configured to manage data flow within processor 600. Controller 602 may be configured to control data transfers between registers, ALU 606, and other functional units of processor 600.

[0114] Memory 604 may include one or more cache memories (e.g., memory native to or contained therein of processor 600, or other memory), such as RAM, ROM, DRAM, SDRAM, SRAM, MRAM, flash memory, etc. In some embodiments, memory 604 may reside within or on the processor chipset (e.g., native to processor 600). In some other embodiments, memory 604 may reside external to the processor chipset (e.g., remote from processor 600).

[0115] Memory 604 may store computer-readable, computer-executable code containing instructions that, when executed by processor 600, cause processor 600 to perform the various functions described herein. The code may be stored in a non-transitory computer-readable medium, such as system memory or another type of memory. Controller 602 and / or processor 600 may be configured to execute the computer-readable instructions stored in memory 604, thereby causing processor 600 to perform various functions. For example, processor 600 and / or controller 602 may be coupled together with or to memory 604, and processor 600 and controller 602 may be configured to perform the various functions described herein. In some instances, processor 600 may include multiple processors and memory 604 may include multiple memories. One or more of the multiple processors may be coupled to one or more of the multiple memories, which may be individually or jointly configured to perform the various functions described herein.

[0116] One or more ALUs 606 may be configured to support various operations according to the examples described herein. In some embodiments, one or more ALUs 606 may reside within or on a processor chipset (e.g., processor 600). In some other embodiments, one or more ALUs 606 may reside external to the processor chipset (e.g., processor 600). One or more ALUs 606 may perform one or more operations on data, such as addition, subtraction, multiplication, and division. For example, one or more ALUs 606 may receive input operands and an opcode, the opcode determining the operation to be performed. One or more ALUs 606 are configured with various logic and arithmetic circuits, including adders, subtractors, shifters, and logic gates, to process and manipulate data according to the operation. Alternatively, one or more ALU 606s may support logical operations such as AND, OR, XOR, NOR, and NAND, thereby enabling one or more ALU 606s to handle conditional operations, comparisons, and bitwise operations.

[0117] According to the examples disclosed herein, processor 600 may support wireless communication. Processor 600 may be configured or operable to support at least one controller (e.g., controller 602), said at least one controller being coupled to at least one memory (e.g., memory 604) and configured to cause the processor to: receive a set of data samples associated with a wireless channel from at least one second node; obtain a subset of data samples from said data sample set, said subset of data samples comprising at least a first subset of data samples and a second subset of data samples; and determine one or more parameters from said subset of data samples of a machine learning model of a wireless device associated with the processor.

[0118] Additionally, the processor 600 may be configured or operable to support any or a combination of the following: the at least one controller is configured to cause the processor 600 to transmit signaling to the at least one second node indicative of the one or more parameters of the machine learning model. Alternatively, a corresponding data sample in the data sample set includes at least one of a first value representing an output corresponding to the wireless channel and a second value representing a message associated with the output corresponding to the wireless channel. Alternatively, the message corresponds to a transmission from one or more of the at least one second node or at least one third node. Alternatively or alternatively, to determine the one or more parameters of the machine learning model, the at least one controller is configured to cause the processor 600 to: obtain a first set of parameters associated with the first data sample subset in response to training a first instance of the machine learning model using supervised learning techniques based on the first data sample subset; obtain a second set of parameters associated with the second data sample subset in response to training a second instance of the machine learning model using the supervised learning techniques based on the second data sample subset; and obtain the one or more parameters by combining the first set of parameters associated with the first data sample subset and the second set of parameters associated with the second data sample subset. Alternatively or alternatively, the one or more parameters define the machine learning model.

[0119] Alternatively or alternatively, the data sample set comprises one or more different data sample types, and a corresponding subset of data samples comprises at least one data sample of the one or more different data sample types. Alternatively or alternatively, the number of corresponding values ​​of the data samples in the subset of data samples satisfies a threshold, and wherein the threshold is based on the total number of values ​​of the data samples in the data sample set and the number of values ​​of the subset of data samples. Alternatively or alternatively, the at least one controller is configured to cause the processor 600 to: obtain an output associated with at least a portion of the received transmission from the machine learning model based on inputting an additional data sample set associated with the received transmission into the machine learning model, wherein the at least portion of the received transmission corresponds to corrupted signaling; and decode the received transmission at least in part based on obtaining the at least portion of the received transmission. Alternatively or alternatively, the at least one controller is configured to cause the processor 600 to: obtain an output from the machine learning model associated with at least one of a beam selection procedure or an estimation procedure, which is associated with the wireless channel, based on inputting an additional set of data samples into the machine learning model; and select a beam for the beam selection procedure or update one or more communication parameters based on the output. Alternatively or alternatively, the wireless device includes at least one of a receiver or a decoder.

[0120] Figure 7 The diagram illustrates an example of an NE 700 according to various aspects of this disclosure. The NE 700 may include a processor 702, a memory 704, a controller 706, and a transceiver 708. The processor 702, memory 704, controller 706, or transceiver 708, or various combinations thereof, or various components thereof, may be examples of components for performing various aspects of this disclosure as described herein. These components may be coupled via one or more interfaces (e.g., operational ground, communication ground, functional ground, electronic ground, electrical ground). In some instances, the NE 700 may be an example of a node, as referenced... Figures 1 to 4 As described.

[0121] Processor 702, memory 704, controller 706, or transceiver 708, or various combinations or components thereof, may be implemented in hardware (e.g., a circuit system). The hardware may include a processor, digital signal processor (DSP), application-specific integrated circuit (ASIC), or other programmable logic device, or any combination thereof, configured or otherwise supporting components for performing the functions described in this disclosure.

[0122] Processor 702 may include intelligent hardware devices (e.g., a general-purpose processor, DSP, CPU, ASIC, FPGA, or any combination thereof). In some embodiments, processor 702 may be configured to operate memory 704. In some other embodiments, memory 704 may be integrated into processor 702. Processor 702 may be configured to execute computer-readable instructions stored in memory 704, thereby enabling NE 700 to perform various functions of this disclosure.

[0123] Memory 704 may include volatile or non-volatile memory. Memory 704 may store computer-readable, computer-executable code containing instructions that, when executed by processor 702, cause NE 700 to perform the various functions described herein. The code may be stored in a non-transitory computer-readable medium, such as memory 704 or another type of memory. Computer-readable media includes both non-transitory computer storage media and communication media, wherein the communication media includes any medium that facilitates the transfer of a computer program from one location to another. Non-transitory storage media may be any available medium accessible by a general-purpose or special-purpose computer.

[0124] In some implementations, processor 702 and memory 704 coupled to processor 702 may be configured to cause NE 700 to perform one or more of the functions described herein (e.g., processor 702 executing instructions stored in memory 704). For example, according to the examples disclosed herein, processor 702 may support wireless communication at NE 700. NE 700 may be configured or operable to support components for performing: transmitting first signaling to a second node, the first signaling indicating a set of data samples associated with a wireless channel; and receiving second signaling from the second node, the second signaling indicating one or more parameters of a machine learning model of a wireless device associated with the first node, wherein said one or more parameters are obtained from a subset of data samples of the data sample set.

[0125] Additionally, the NE 700 can be configured or operated to support any or a combination of the following: a corresponding data sample in the data sample set includes at least one of a first value representing an output corresponding to the wireless channel and a second value representing a message associated with the output corresponding to the wireless channel. Alternatively, the message corresponds to a transmission from one or more of the first node or at least one third node. Alternatively, the one or more parameters of the machine learning model are obtained based on a combination of corresponding parameter sets associated with the subset of data samples, wherein the corresponding parameter sets are obtained in response to training corresponding instances of the machine learning model using supervised learning techniques based on the subset of data samples. Alternatively, the one or more parameters define the machine learning model.

[0126] Alternatively or alternatively, the data sample set comprises one or more different data sample types, and a corresponding subset of data samples comprises at least one data sample of the one or more different data sample types. Alternatively or alternatively, the number of corresponding values ​​of the data samples in the subset of data samples satisfies a threshold, and wherein the threshold is based on the total number of values ​​of the data samples in the data sample set and the number of values ​​of the subset of data samples. Alternatively or alternatively, the NE 700 may be configured or operable to support: obtaining an output associated with at least a portion of the received transmission from the machine learning model based on inputting an additional data sample set associated with the received transmission into the machine learning model, wherein the at least portion of the received transmission corresponds to corrupted signaling; and decoding the received transmission based on obtaining the at least portion of the received transmission. Alternatively or alternatively, the NE700 may be configured or operable to support: obtaining an output from the machine learning model associated with at least one of a beam selection procedure or an estimation procedure, which is associated with the wireless channel, based on inputting an additional set of data samples into the machine learning model; and selecting a beam for the beam selection procedure or updating one or more communication parameters based on the output. Alternatively or alternatively, the wireless device includes at least one of a receiver or a decoder.

[0127] Alternatively, the NE 700 may support at least one memory (e.g., memory 704) and at least one processor (e.g., processor 702), the at least one processor being coupled to the at least one memory and configured to cause the NE to: transmit a first signaling to a second node, the first signaling indicating a set of data samples associated with a wireless channel; and receive a second signaling from the second node, the second signaling indicating one or more parameters of a machine learning model of a wireless device associated with the first node, wherein the one or more parameters are obtained from a subset of data samples of the set of data samples.

[0128] Additionally, the NE 700 can be configured to support any or a combination of the following: a corresponding data sample in the data sample set includes at least one of a first value representing an output corresponding to the wireless channel and a second value representing a message associated with the output corresponding to the wireless channel. Alternatively, the message corresponds to a transmission from one or more of the first node or at least one third node. Alternatively, the one or more parameters of the machine learning model are obtained based on a combination of corresponding parameter sets associated with the subset of data samples, wherein the corresponding parameter sets are obtained in response to training corresponding instances of the machine learning model using supervised learning techniques based on the subset of data samples. Alternatively, the one or more parameters define the machine learning model.

[0129] Alternatively or alternatively, the data sample set comprises one or more different data sample types, and a corresponding subset of data samples comprises at least one data sample of the one or more different data sample types. Alternatively or alternatively, the number of corresponding values ​​of the data samples in the subset of data samples satisfies a threshold, and wherein the threshold is based on the total number of values ​​of the data samples in the data sample set and the number of values ​​of the subset of data samples. Alternatively or alternatively, the at least one processor is configured such that the NE: obtains an output from the machine learning model associated with at least a portion of the received transmission based on inputting an additional data sample set associated with the received transmission into the machine learning model, wherein the at least portion of the received transmission corresponds to corrupted signaling; and decodes the received transmission based on obtaining the at least portion of the received transmission. Alternatively or alternatively, the at least one processor is configured such that the NE: obtains an output from the machine learning model associated with at least one of a beam selection procedure or an estimation procedure, which is associated with the wireless channel, based on inputting an additional set of data samples into the machine learning model; and selects a beam for the beam selection procedure or updates one or more communication parameters based on the output. Alternatively or alternatively, the wireless device includes at least one of a receiver or a decoder.

[0130] Controller 706 manages the input and output signals of NE 700. Controller 706 can also manage peripheral devices not integrated into NE 700. In some embodiments, controller 706 may utilize an operating system such as iOS®, ANDROID®, WINDOWS®, or other operating systems. In some embodiments, controller 706 may be implemented as part of processor 702.

[0131] In some embodiments, the NE 700 may include at least one transceiver 708. In other embodiments, the NE 700 may have more than one transceiver 708. The transceiver 708 may represent a wireless transceiver. The transceiver 708 may include one or more receiver chains 710, one or more transmitter chains 712, or a combination thereof.

[0132] Receiver chain 710 may be configured to receive signals (e.g., control information, data, packets) via wireless media. For example, receiver chain 710 may include one or more antennas for receiving signals via air or wireless media. Receiver chain 710 may include at least one amplifier (e.g., a low-noise amplifier (LNA)) configured to amplify the received signal. Receiver chain 710 may include at least one demodulator configured to demodulate the received signal and obtain transmitted data by reversing the modulation technique applied during signal transmission. Receiver chain 710 may include at least one decoder for decoding the demodulated signal to receive transmitted data.

[0133] Transmitter chain 712 can be configured to generate and transmit signals (e.g., control information, data, packets). Transmitter chain 712 may include at least one modulator for modulating data onto a carrier signal, thereby preparing the signal for transmission over a wireless medium. The at least one modulator may be configured to support one or more techniques, such as amplitude modulation (AM), frequency modulation (FM), or digital modulation schemes like phase shift keying (PSK) or quadrature amplitude modulation (QAM). Transmitter chain 712 may also include at least one power amplifier configured to amplify the modulated signal to an appropriate power level suitable for transmission over a wireless medium. Transmitter chain 712 may also include one or more antennas for transmitting the amplified signal over the air or in the wireless medium.

[0134] Figure 8 The diagram illustrates a flowchart of a method 800 according to various aspects of this disclosure. The operation of the method can be implemented by nodes (e.g., UE and / or NE) as described herein. In some embodiments, the node (e.g., UE and / or NE) can execute a set of instructions to control the functional elements of the node (e.g., UE and / or NE) to perform the described functions. It should be noted that the methods described herein describe possible implementations, and the operations and steps can be rearranged or otherwise modified, and other implementations are possible.

[0135] At point 802, the method may include receiving a set of data samples associated with the wireless channel from at least one second node. The operation of point 802 may be performed according to examples as described herein. In some embodiments, aspects of the operation of point 802 may be provided by reference to [reference needed]. Figure 5 The UE described and / or as referenced Figure 7 The NE described is used to execute.

[0136] At 804, the method may include obtaining a subset of data samples from the data sample set, wherein the subset of data samples includes at least a first subset of data samples and a second subset of data samples. The operation of 804 may be performed according to the examples described herein. In some embodiments, aspects of the operation of 804 may be provided by reference to... Figure 5 The UE described and / or as referenced Figure 7 The NE described is used to execute.

[0137] At 806, the method may include determining one or more parameters from a subset of data samples of a machine learning model of the wireless device associated with the first node. The operation of 806 may be performed according to the examples described herein. In some embodiments, aspects of the operation of 806 may be provided by reference to [reference needed]. Figure 5 The UE described and / or as referenced Figure 7 The NE described is used to execute.

[0138] Figure 9 The diagram illustrates a flowchart of a method 900 according to various aspects of this disclosure. The operation of the method can be implemented by nodes (e.g., UE and / or NE) as described herein. In some embodiments, the node (e.g., UE and / or NE) can execute a set of instructions to control the functional elements of the node (e.g., UE and / or NE) to perform the described functions. It should be noted that the methods described herein describe possible implementations, and the operations and steps can be rearranged or otherwise modified, and other implementations are possible.

[0139] At 902, the method may include transmitting a first signaling to a second node, the first signaling indicating a set of data samples associated with the wireless channel. The operation of 902 may be performed according to the examples described herein. In some embodiments, aspects of the operation of 902 may be as described in references... Figure 5 The UE described and / or as referenced Figure 7 The NE described is used to execute.

[0140] At 904, the method may include receiving second signaling from a second node, the second signaling indicating one or more parameters of a machine learning model of a wireless device associated with the first node, said one or more parameters being obtained from a subset of data samples of the data sample set. The operation of 904 may be performed according to the examples described herein. In some embodiments, aspects of the operation of 904 may be provided by reference to [reference needed]. Figure 5 The UE described and / or as referenced Figure 7 The NE described is used to execute.

[0141] The description herein is provided to enable those skilled in the art to make or use this disclosure. Various modifications to this disclosure will be apparent to those skilled in the art, and the general principles defined herein may be applied to other variations without departing from the scope of this disclosure. Therefore, this disclosure is not limited to the examples and designs described herein, but is given the broadest scope consistent with the principles and novel features disclosed herein.

Claims

1. A first node for wireless communication, comprising: At least one memory; as well as At least one processor, coupled to the at least one memory and configured to enable the first node to: Receive a set of data samples associated with the wireless channel from at least one second node; A subset of data samples is obtained from the data sample set, wherein the subset of data samples includes at least a first subset of data samples and a second subset of data samples; and One or more parameters of the machine learning model of the wireless device associated with the first node are determined from the subset of the data samples.

2. The first node of claim 1, wherein the at least one processor is further configured to enable the first node to transmit signaling to the at least one second node indicating the one or more parameters of the machine learning model.

3. The first node of claim 1, wherein the corresponding data sample in the data sample set includes at least one of a first value representing an output corresponding to the wireless channel and a second value representing a message associated with the output corresponding to the wireless channel.

4. The first node according to claim 3, wherein the message corresponds to a transmission from one or more of the at least one second node or at least one third node.

5. The first node of claim 1, wherein, for determining the one or more parameters of the machine learning model, the at least one processor is configured to cause the first node to: In response to a first instance of training the machine learning model using supervised learning techniques based at least in part on the first subset of data samples, a first set of parameters associated with the first subset of data samples is obtained; In response to a second instance of training the machine learning model using the supervised learning technique based at least in part on the second subset of data samples, a second set of parameters associated with the second subset of data samples is obtained; and The one or more parameters are obtained at least in part based on combining the first set of parameters associated with the first subset of data samples and the second set of parameters associated with the second subset of data samples.

6. The first node according to claim 1, wherein the one or more parameters define the machine learning model.

7. The first node according to claim 1, wherein the data sample set includes one or more different data sample types, and wherein a corresponding data sample subset includes at least one data sample of the one or more different data sample types.

8. The first node according to claim 1, wherein the number of corresponding values ​​of data samples in the subset of data samples satisfies a threshold, and wherein the threshold is at least partially based on the total number of values ​​of data samples in the data sample set and the number of values ​​of the subset of data samples.

9. The first node of claim 1, wherein the at least one processor is configured to cause the first node to: The machine learning model is at least partially based on inputting an additional set of data samples associated with the received transmission into the machine learning model to obtain an output associated with at least a portion of the received transmission, wherein the at least portion of the received transmission corresponds to corrupted signaling; and The received transmission is decoded at least in part based on obtaining at least a portion of the received transmission.

10. The first node of claim 1, wherein the at least one processor is configured to cause the first node to: At least in part, based on inputting an additional set of data samples into the machine learning model, an output associated with at least one of a beam selection procedure or an estimation procedure is obtained from the machine learning model, the beam selection procedure or the estimation procedure being associated with the wireless channel; and The beam selection procedure may select a beam or update one or more communication parameters, at least in part, based on the output.

11. The first node of claim 1, wherein the wireless device comprises at least one of a receiver or a decoder.

12. A processor for wireless communication, comprising: At least one controller, coupled to at least one memory and configured to enable the processor to: Receive a set of data samples associated with the wireless channel from at least one second node; A subset of data samples is obtained from the data sample set, wherein the subset of data samples includes at least a first subset of data samples and a second subset of data samples; and One or more parameters of the machine learning model of the wireless device associated with the processor are determined from the subset of the data samples.

13. A method executed by a first node, the method comprising: Receive a set of data samples associated with the wireless channel from at least one second node; A subset of data samples is obtained from the data sample set, wherein the subset of data samples includes at least a first subset of data samples and a second subset of data samples; and One or more parameters of the machine learning model of the wireless device associated with the first node are determined from the subset of the data samples.

14. A first node for wireless communication, comprising: At least one memory; as well as At least one processor, coupled to the at least one memory and configured to enable the first node to: Transmit a first signaling message to the second node, the first signaling message indicating a data sample set associated with the wireless channel; and The second node receives a second signaling message indicating one or more parameters of a machine learning model of a wireless device associated with the first node, wherein the one or more parameters are obtained from a subset of data samples from the data sample set.

15. The first node of claim 14, wherein the corresponding data sample in the data sample set includes at least one of a first value representing an output corresponding to the wireless channel and a second value representing a message associated with the output corresponding to the wireless channel.

16. The first node of claim 14, wherein the one or more parameters of the machine learning model are obtained at least in part based on a combination of corresponding parameter sets associated with the subset of data samples, and wherein the corresponding parameter sets are obtained in response to training corresponding instances of the machine learning model using supervised learning techniques based at least in part on the subset of data samples.

17. The first node of claim 14, wherein the data sample set comprises one or more different data sample types, and wherein a corresponding subset of data samples comprises at least one data sample of the one or more different data sample types.

18. The first node of claim 14, wherein the number of corresponding values ​​of data samples in the subset of data samples satisfies a threshold, and wherein the threshold is at least partially based on the total number of values ​​of data samples in the data sample set and the number of values ​​of the subset of data samples.

19. The first node of claim 14, wherein the at least one processor is further configured to cause the first node to: The machine learning model is at least partially based on inputting an additional set of data samples associated with the received transmission into the machine learning model to obtain an output associated with at least a portion of the received transmission, wherein the at least portion of the received transmission corresponds to corrupted signaling; and The received transmission is decoded at least in part based on obtaining at least a portion of the received transmission.

20. The first node of claim 14, wherein the at least one processor is further configured to cause the first node to: At least in part, based on inputting an additional set of data samples into the machine learning model, an output associated with at least one of a beam selection procedure or an estimation procedure is obtained from the machine learning model, the beam selection procedure or the estimation procedure being associated with the wireless channel; and The beam selection procedure may select a beam or update one or more communication parameters, at least in part, based on the output.

Images