Transmission of known data for collaborative training of artificial neural networks

Abstract

A wireless communication method performed by a receiving device includes sending a request to a first transmitting device in a group of transmitting devices for a first known payload for training an artificial neural network of the receiving device. The group of transmitting devices may be a group of multiple transmit reception points (TRPs) or a group of user equipments (UEs). The first transmitting device may be a serving TRP or one of the UEs in the group. The method also includes receiving the first known payload from the first transmitting device in response to the request. The method also includes updating the artificial neural network at the receiving device based on at least the first known payload.

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims priority to U.S. patent application Ser. No. 17 / 019,125, filed on September 11, 2020, entitled “TRANSMISSION OF KNOWN DATA FORCOOPERATIVE TRAINING OF ARTIFICIAL NEURAL NETWORKS,” the disclosure of which is incorporated herein by reference in its entirety. Technical Field

[0003] Aspects of the present disclosure relate generally to wireless communications and, more particularly, to techniques and apparatus for transmitting known payloads for collaborative training of artificial neural networks. Background Art

[0004] Wireless communication systems are widely deployed to provide a variety of telecommunication services, such as telephony, video, data, messaging, and broadcasting. Typical wireless communication systems may employ multiple access technologies capable of supporting communication with multiple users by sharing available system resources (e.g., bandwidth, transmit power, etc.). Examples of such multiple access technologies include code division multiple access (CDMA) systems, time division multiple access (TDMA) systems, frequency division multiple access (FDMA) systems, orthogonal frequency division multiple access (OFDMA) systems, single-carrier frequency division multiple access (SC-FDMA) systems, time division synchronous code division multiple access (TD-SCDMA) systems, and long term evolution (LTE). LTE / LTE-Advanced is a set of enhancements to the Universal Mobile Telecommunications System (UMTS) mobile standard promulgated by the Third Generation Partnership Project (3GPP).

[0005] A wireless communication network may include multiple base stations (BSs) that can support communication for multiple user equipment (UEs). UEs can communicate with a BS via downlinks and uplinks. A downlink (or forward link) refers to the communication link from the BS to the UE, and an uplink (or reverse link) refers to the communication link from the UE to the BS. As will be described in more detail herein, a BS may be referred to as a Node B, gNB, access point (AP), radio head, transmit receive point (TRP), new radio (NR) BS, fifth generation (5G) Node B, etc.

[0006] The aforementioned multiple access technologies have been adopted in various telecommunications standards to provide a common protocol that enables diverse user devices to communicate at municipal, national, regional, and even global levels. New Radio (NR), also known as 5G, is a set of enhancements to the LTE mobile standard promulgated by the Third Generation Partnership Project (3GPP). NR aims to better support mobile broadband internet access by improving spectral efficiency, reducing costs, and improving service. It utilizes new spectrum and uses orthogonal frequency division multiplexing (CP-OFDM) with a cyclic prefix (CP) on the downlink (DL) and CP-OFDM and / or SC-FDM (e.g., also known as discrete Fourier transform spread OFDM (DFT-s-OFDM)) on the uplink (UL), as well as supporting beamforming, multiple-input, multiple-output (MIMO) antenna technology, and carrier aggregation for better integration with other open standards. However, with the increasing demand for mobile broadband access, further improvements to NR and LTE technologies are needed. These improvements should preferably be applicable to other multiple access technologies and the telecommunications standards that adopt them.

[0007] An artificial neural network may include a group of interconnected artificial neurons (e.g., a neuron model). An artificial neural network may be a computing device or represented as a method executed by a computing device. A convolutional neural network (such as a deep convolutional neural network) is a feedforward artificial neural network. A convolutional neural network may include layers of neurons that may be configured in a tiled receptive field. It is desirable to apply neural network processing to wireless communications to achieve higher efficiency. Summary of the Invention

[0008] According to one aspect of the present disclosure, a wireless communication method performed by a receiving device sends a request to a first transmitting device in a group of transmitting devices for a first known payload for training an artificial neural network of the receiving device. The method also receives the first known payload from the first transmitting device in response to the request. The method also updates the artificial neural network at the receiving device based on at least the first known payload.

[0009] In another aspect of the present disclosure, a wireless communication method performed by at least a first transmitting device in a group of transmitting devices transmits a first unknown payload to a receiving device in a group of receiving devices based on first transmission settings of the first transmitting device. The method also receives a request from the receiving device for a first known payload for training an artificial neural network of the receiving device. The method also transmits the first known payload to the receiving device based on the first transmission settings.

[0010] In another aspect of the present disclosure, an apparatus for wireless communication, executed by a receiving device, includes a processor and a memory coupled to the processor. When executed by the processor, instructions stored in the memory are operable to cause the apparatus to send a request to a first transmitting device in a group of transmitting devices for a first known payload for training an artificial neural network of the receiving device. The apparatus may also receive the first known payload from the first transmitting device in response to the request. The apparatus may also update the artificial neural network at the receiving device based at least on the first known payload.

[0011] In another aspect of the present disclosure, an apparatus for wireless communication, executed by at least a first transmitting device in a group of transmitting devices, includes a processor and a memory coupled to the processor. When executed by the processor, instructions stored in the memory are operable to cause the apparatus to transmit a first unknown payload to a receiving device in a group of receiving devices based on first transmission settings of the first transmitting device. The apparatus may also receive a request from the receiving device for a first known payload for training an artificial neural network of the receiving device. The apparatus may also transmit the first known payload to the receiving device based on the first transmission settings.

[0012] In another aspect of the present disclosure, a receiving device includes means for sending a request to a first transmitting device in a group of transmitting devices for a first known payload for training an artificial neural network of the receiving device. The receiving device also includes means for receiving the first known payload from the first transmitting device in response to the request. The receiving device also includes means for updating the artificial neural network at the receiving device based on at least the first known payload.

[0013] In another aspect of the present disclosure, a transmitting device includes means for transmitting a first unknown payload to a receiving device in a group of receiving devices based on first transmission settings of the first transmitting device. The transmitting device also includes means for receiving, from the receiving device, a request for a first known payload for training an artificial neural network of the receiving device. The transmitting device also includes means for transmitting the first known payload to the receiving device based on the first transmission settings.

[0014] In another aspect of the present disclosure, a non-transitory computer-readable medium having program code recorded thereon is disclosed. The program code is executed by a receiving device and includes program code for sending a request to a first transmitting device in a group of transmitting devices for a first known payload for training an artificial neural network of the receiving device. The receiving device also includes program code for receiving the first known payload from the first transmitting device in response to the request. The receiving device also includes program code for updating the artificial neural network at the receiving device based on at least the first known payload.

[0015] In another aspect of the present disclosure, a non-transitory computer-readable medium having program code recorded thereon is disclosed. The program code is executed by a sending device and includes program code for sending a first unknown payload to a receiving device in a group of receiving devices based on first transmission settings of a first sending device. The sending device also includes program code for receiving, from the receiving device, a request for a first known payload for training an artificial neural network of the receiving device. The sending device also includes program code for sending the first known payload to the receiving device based on the first transmission settings.

[0016] Aspects generally include methods, apparatus, systems, computer program products, non-transitory computer-readable media, user equipment, base stations, wireless communication devices, and processing systems as fully described with reference to and illustrated by the accompanying drawings and description.

[0017] The features and technical advantages of the examples according to the present disclosure have been outlined in a rather broad manner so that the following detailed description may be better understood. Additional features and advantages will be described below. The concepts and specific examples disclosed may be readily used as a basis for modifying or designing other structures for achieving the same purposes of the present disclosure. Such equivalent structures do not depart from the scope of the appended claims. The nature of the concepts disclosed herein, their organization and method of operation, and related advantages will be better understood from the following description when considered in conjunction with the accompanying drawings. Each of the figures is provided for the purpose of illustration and description and not as a definition of limitations of the claims. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to enable a detailed understanding of the above-described features of the present disclosure, a more particular description, briefly summarized above, may be obtained by reference to various aspects, some of which are illustrated in the accompanying drawings. It should be noted, however, that the drawings illustrate only certain typical aspects of the present disclosure and are therefore not to be considered limiting of its scope, as the description may admit to other equally effective aspects. The same reference numerals in different drawings may identify the same or similar elements.

[0019] Figure 1 is a block diagram conceptually illustrating an example of a wireless communication network according to aspects of the present disclosure.

[0020] Figure 2 is a block diagram conceptually illustrating an example of a base station communicating with a user equipment (UE) in a wireless communication network according to aspects of the present disclosure.

[0021] Figure 3 An example implementation of designing a neural network using a system on a chip (SOC), including a general-purpose processor, according to certain aspects of the present disclosure is shown.

[0022] Figure 4A 、 Figure 4B and Figure 4C is a diagram illustrating a neural network according to aspects of the present disclosure.

[0023] Figure 4D is a diagram illustrating an exemplary deep convolutional network (DCN) according to aspects of the present disclosure.

[0024] Figure 5 is a block diagram illustrating an exemplary deep convolutional network (DCN) according to aspects of the present disclosure.

[0025] Figure 6 is a diagram illustrating an example of multi-TRP (mTRP) communication according to aspects of the present disclosure.

[0026] Figure 7 is a block diagram illustrating an example of an artificial neural network according to aspects of the present disclosure.

[0027] Figure 8 and Figure 9 is a timing diagram illustrating an example of sending a known payload according to aspects of the present disclosure.

[0028] Figure 10 is a diagram illustrating an example process performed, for example, by a receiving device, according to aspects of the present disclosure.

[0029] Figure 11 is a diagram illustrating an example process performed, for example, by a transmitting device, according to aspects of the present disclosure. DETAILED DESCRIPTION

[0030] The various aspects of the present disclosure will be described more fully below with reference to the accompanying drawings. However, the present disclosure can be implemented in many different forms and should not be interpreted as being limited to any specific structure or function presented throughout the present disclosure. On the contrary, these aspects are provided to make the present disclosure thorough and complete and to fully present the scope of the present disclosure to those skilled in the art. Based on the teachings herein, those skilled in the art should understand that the scope of the present disclosure is intended to cover any aspect of the present disclosure disclosed herein, whether implemented independently of any other aspect of the present disclosure or implemented in combination with any other aspect of the present disclosure. For example, any number of aspects set forth herein can be used to implement a device or practice method. In addition, the scope of the present disclosure is intended to cover such a device or method, which is based on the various aspects of the present disclosure set forth herein and also uses other structures, functions, or structures and functions to practice, or uses other structures, functions, or structures and functions different from the various aspects of the present disclosure set forth herein to practice. It should be understood that any aspect of the present disclosure can be embodied by one or more elements of the claims.

[0031] Several aspects of telecommunications systems will now be presented with reference to various devices and techniques. These devices and techniques will be described in the following detailed description and illustrated in the accompanying drawings by various blocks, modules, components, circuits, steps, processes, algorithms, etc. (collectively, "elements"). These elements can be implemented using hardware, software, or a combination thereof. Whether these elements are implemented as hardware or software depends on the specific application and the design constraints imposed on the overall system.

[0032] It should be noted that while aspects may be described using terminology generally associated with 5G and later wireless technologies, aspects of the present disclosure may be applicable to other generation-based communication systems, such as and including 3G and / or 4G technologies.

[0033] Figure 1 is a diagram illustrating a network 100 in which aspects of the present disclosure may be practiced. The network 100 may be a 5G or NR network or other wireless network, such as an LTE network. The wireless network 100 may include multiple BSs 110 (shown as BS 110a, BS 110b, BS 110c, and BS 110d) and other network entities. A BS is an entity that communicates with a UE and may also be referred to as a base station, NR BS, Node B, gNB, 5G Node B (NB), access point, transmit receive point (TRP), etc. Each BS may provide communication coverage for a specific geographic area. In 3GPP, the term "cell" may refer to the coverage area of ​​a BS and / or a BS subsystem serving that coverage area, depending on the context in which the term is used.

[0034] A BS may provide communication coverage for macro cells, pico cells, femto cells, and / or other types of cells. A macro cell may cover a relatively large geographic area (e.g., a radius of several kilometers) and may allow unrestricted access to UEs with service subscriptions. A pico cell may cover a relatively small geographic area and may allow unrestricted access to UEs with service subscriptions. A femto cell may cover a relatively small geographic area (e.g., a home) and may allow restricted access to UEs associated with the femto cell (e.g., UEs in a closed subscriber group (CSG)). A BS for a macro cell may be referred to as a macro BS. A BS for a pico cell may be referred to as a pico BS. A BS for a femto cell may be referred to as a femto BS or a home BS. In Figure 1In the example shown, BS 110a may be a macro BS for macro cell 102a, BS 110b may be a pico BS for pico cell 102b, and BS 110c may be a femto BS for femto cell 102c. A BS may support one or more (e.g., three) cells. The terms "eNB," "base station," "NR BS," "gNB," "TRP," "AP," "Node B," "5G NB," and "cell" may be used interchangeably herein.

[0035] In some aspects, the cells are not necessarily stationary, and the geographic area of ​​the cells can move depending on the location of the mobile BS. In some aspects, the BSs can be interconnected with each other and / or with one or more other BSs or network nodes (not shown) in the wireless network 100 via various types of backhaul interfaces, such as direct physical connections using any suitable transport network, virtual networks, etc.

[0036] The wireless network 100 may also include a relay station. A relay station is an entity that can receive data transmissions from an upstream station (e.g., a BS or a UE) and send data transmissions to a downstream station (e.g., a UE or a BS). A relay station may also be a UE that can relay transmissions for other UEs. Figure 1 In the example shown, a relay station 110d may communicate with a macro BS 110a and a UE 120d to facilitate communication between the BS 110a and the UE 120d. A relay station may also be referred to as a relay BS, a relay base station, a relay, or the like.

[0037] The wireless network 100 may be a heterogeneous network including different types of BSs, such as macro BSs, pico BSs, femto BSs, relay BSs, etc. These different types of BSs may have different transmit power levels, different coverage areas, and different impacts on interference in the wireless network 100. For example, a macro BS may have a high transmit power level (e.g., 5 to 40 watts), while a pico BS, a femto BS, and a relay BS may have a lower transmit power level (e.g., 0.1 to 2 watts).

[0038] The network controller 130 may be coupled to a group of BSs and may provide coordination and control for these BSs. The network controller 130 may communicate with the BSs via a backhaul. The BSs may also communicate with each other directly or indirectly, for example, via a wireless or wired backhaul.

[0039] UEs 120 (e.g., 120a, 120b, 120c) may be dispersed throughout the wireless network 100, and each UE may be fixed or mobile. A UE may also be referred to as an access terminal, terminal, mobile station, subscriber unit, station, etc. A UE may be a cellular phone (e.g., a smartphone), a personal digital assistant (PDA), a wireless modem, a wireless communication device, a handheld device, a laptop, a cordless phone, a wireless local loop (WLL) station, a tablet, a camera, a gaming device, a netbook, a smartbook, an ultrabook, a medical device or equipment, a biosensor / device, a wearable device (e.g., a smart watch, smart clothing, smart glasses, a smart wristband, smart jewelry (e.g., a smart ring, a smart bracelet)), an entertainment device (e.g., a music or video device or a satellite radio device), a vehicle component or sensor, a smart meter / sensor, industrial manufacturing equipment, a global positioning system device, or any other suitable device configured to communicate via a wireless or wired medium.

[0040] Some UEs may be considered machine type communication (MTC) or evolved or enhanced machine type communication (eMTC) UEs. MTC and eMTC UEs include, for example, robots, drones, remote devices, sensors, meters, monitors, location tags, etc., which can communicate with a base station, another device (e.g., a remote device), or some other entity. For example, a wireless node may provide connectivity for or to a network (e.g., a wide area network such as the Internet or a cellular network) via a wired or wireless communication link. Some UEs may be considered Internet of Things (IoT) devices and / or may be implemented as narrowband Internet of Things (NB-IoT) devices. Some UEs may be considered customer premises equipment (CPE). UE 120 may be included in a housing that houses components of UE 120 (such as a processor component, a memory component, etc.).

[0041] Generally, any number of wireless networks can be deployed in a given geographic area. Each wireless network can support a specific RAT and operate on one or more frequencies. RATs are also referred to as radio technologies, air interfaces, etc. Frequencies are also referred to as carriers, frequency channels, etc. Each frequency can support a single RAT in a given geographic area to avoid interference between wireless networks of different RATs. In some cases, NR or 5G RAT networks may be deployed.

[0042] In some aspects, two or more UEs 120 (e.g., shown as UE 120a and UE 120e) may communicate directly using one or more sidelink channels (e.g., without using base station 110 as an intermediary for communicating with each other). For example, the UEs 120 may communicate using peer-to-peer (P2P) communication, device-to-device (D2D) communication, vehicle-to-everything (V2X) protocols (e.g., which may include vehicle-to-vehicle (V2V) protocols, vehicle-to-infrastructure (V2I) protocols, etc.), mesh networks, etc. In such cases, the UEs 120 may perform scheduling operations, resource selection operations, and / or other operations described elsewhere herein as being performed by the base station 110.

[0043] As mentioned above, Figure 1 Provided as an example only. Other examples may differ from the Figure 1 Description of what is being done.

[0044] Figure 2 A block diagram shows a design 200 of base station 110 and UE 120, which may be Figure 1 Base station 110 may be equipped with T antennas 234a through 234t, and UE 120 may be equipped with R antennas 252a through 252r. In general, T ≥ 1 and R ≥ 1.

[0045] At base station 110, transmit processor 220 may receive data from data source 212 for one or more UEs, select one or more modulation and coding schemes (MCS) for each UE based at least in part on a channel quality indicator (CQI) received from the UE, process (e.g., encode and modulate) the data for each UE based at least in part on the MCS(s) selected for the UE, and provide data symbols for all UEs. Transmit processor 220 may also process system information (e.g., for semi-static resource partitioning information (SRPI) and control information (e.g., CQI requests, grants, upper layer signaling, etc.) and provide overhead symbols and control symbols. Transmit processor 220 may also generate reference symbols for reference signals (e.g., cell-specific reference signals (CRS)) and synchronization signals (e.g., primary synchronization signal (PSS) and secondary synchronization signal (SSS)). The transmit (TX) multiple-input multiple-output (MIMO) processor 230 may perform spatial processing (e.g., precoding) on ​​data symbols, control symbols, overhead symbols, and / or reference symbols (if applicable), and may provide T output symbol streams to T modulators (MODs) 232a through 232t. Each modulator 232 may process a corresponding output symbol stream (e.g., for OFDM, etc.) to obtain an output sample stream. Each modulator 232 may further process (e.g., convert to analog, amplify, filter, and upconvert) the output sample stream to obtain a downlink signal. The T downlink signals from modulators 232a through 232t may be transmitted via T antennas 234a through 234t, respectively. According to various aspects described in more detail below, position coding may be utilized to generate synchronization signals to convey additional information.

[0046] At UE 120, antennas 252a through 252r may receive downlink signals from base station 110 and / or other base stations and may provide received signals to demodulators (DEMODs) 254a through 254r, respectively. Each demodulator 254 may condition (e.g., filter, amplify, downconvert, and digitize) the received signal to obtain input samples. Each demodulator 254 may further process the input samples (e.g., for OFDM, etc.) to obtain received symbols. A MIMO detector 256 may obtain received symbols from all R demodulators 254a through 254r, perform MIMO detection on the received symbols if applicable, and provide detected symbols. A receive processor 258 may process (e.g., demodulate and decode) the detected symbols, provide decoded data for UE 120 to a data sink 260, and provide decoded control information and system information to a controller / processor 280. The channel processor may determine reference signal received power (RSRP), received signal strength indicator (RSSI), reference signal received quality (RSRQ), channel quality indicator (CQI), etc. In some aspects, one or more components of UE 120 may be included in a housing.

[0047] On the uplink, at UE 120, a transmit processor 264 may receive and process data from a data source 262 and control information (e.g., including reports of RSRP, RSSI, RSRQ, CQI, etc.) from a controller / processor 280. The transmit processor 264 may also generate reference symbols for one or more reference signals. The symbols from the transmit processor 264 may be precoded by a TX MIMO processor 266, if applicable, further processed by modulators 254a through 254r (e.g., for DFT-s-OFDM, CP-OFDM, etc.), and transmitted to the base station 110. At the base station 110, uplink signals from UE 120 and other UEs may be received by antennas 234, processed by demodulators 254, detected by MIMO detector 236 (if applicable), and further processed by receive processor 238 to obtain decoded data and control information transmitted by UE 120. The receive processor 238 may provide decoded data to a data sink 239 and decoded control information to the controller / processor 240. The base station 110 may include a communication unit 244 and communicate with the network controller 130 through the communication unit 244. The network controller 130 may include a communication unit 294, a controller / processor 290, and a memory 292.

[0048] As described below, the controller / processor 240 of the base station 110, the controller / processor 280 of the UE 120, and / or Figure 2Any other component(s) of the system may perform one or more techniques associated with sending or receiving known payloads for collaborative neural network training. For example, the controller / processor 240 of the base station 110, the controller / processor 280 of the UE 120, and / or Figure 2 Any other component(s) of may perform or direct e.g. Figure 10 and Figure 11 The operations of processes 1000, 1100 and / or other processes described herein may be performed. Memories 242 and 282 may store data and program codes for base station 110 and UE 120, respectively. Scheduler 246 may schedule UEs for data transmission on the downlink and / or uplink.

[0049] In some aspects, the UE 120 or base station 110 may include: a component for sending a request for a first known payload for training an artificial neural network of a receiving device to a first transmitting device in a group of transmitting devices; a component for receiving the first known payload from the first transmitting device in response to the request; and a component for updating the artificial neural network at the receiving device based on at least the first known payload.

[0050] In other aspects, the UE 120 or the base station 110 may include: a component for sending a first unknown payload to a receiving device in a group of receiving devices based on a first transmission setting of a first transmitting device; a component for receiving a request for a first known payload for training an artificial neural network of the receiving device from the receiving device; and a component for sending the first known payload to the receiving device based on the first transmission setting.

[0051] These components may include combinations Figure 2 One or more components of UE 120 or base station 110 are described.

[0052] As mentioned above, Figure 2 Provided as an example only. Other examples may differ from the Figure 2 Description of what is being done.

[0053] In some cases, different types of devices supporting different types of applications and / or services can coexist in a cell. Examples of different types of devices include UE handsets, customer premises equipment (CPE), vehicles, and Internet of Things (IoT) devices. Examples of different types of applications include ultra-reliable low-latency communications (URLLC), massive machine-type communications (mMTC), enhanced mobile broadband (eMBB), and vehicle-to-everything (V2X). In addition, in some cases, a single device can support different applications or services simultaneously.

[0054] Figure 3An example implementation of a system on a chip (SOC) 300 according to certain aspects of the present disclosure is shown. The SOC may include a central processing unit (CPU) 302 or a multi-core CPU configured to send or receive a known payload for collaborative neural network training. The SOC 300 may be included in a base station 110 or a UE 120. Variables (e.g., neural signals and synaptic weights), system parameters associated with a computing device (e.g., a neural network with weights), delays, frequency window information, and task information may be stored in a memory block associated with a neural processing unit (NPU) 108, a memory block associated with the CPU 302, a memory block associated with a graphics processing unit (GPU) 304, a memory block associated with a digital signal processor (DSP) 306, a memory block 318, or may be distributed across multiple blocks. Instructions executed at the CPU 302 may be loaded from a program memory associated with the CPU 302 or may be loaded from the memory block 318.

[0055] The SOC 300 may also include additional processing blocks tailored for specific functions, such as a GPU 304, a DSP 306, a connection block 310, which may include fifth generation (5G) connectivity, fourth generation long term evolution (4G LTE) connectivity, Wi-Fi connectivity, USB connectivity, Bluetooth connectivity, etc., and a multimedia processor 312 that may, for example, detect and recognize gestures. In one implementation, the NPU is implemented in the CPU, DSP, and / or GPU. The SOC 300 may also include a sensor processor 314, an image signal processor (ISP) 316, and / or a navigation module 320, which may include a global positioning system.

[0056] The SOC 300 may be based on the ARM instruction set. In one aspect of the present disclosure, the instructions loaded into the general-purpose processor 302 may include: code for sending a request to a first transmitting device in a group of transmitting devices for a first known payload for training an artificial neural network of a receiving device; code for receiving the first known payload from the first transmitting device in response to the request; and code for updating the artificial neural network at the receiving device based on at least the first known payload.

[0057] In another aspect of the present disclosure, the instructions loaded into the general processor 302 may include: code for sending a first unknown payload to a receiving device in a group of receiving devices based on a first transmission setting of a first sending device; code for receiving a request from the receiving device for a first known payload for training an artificial neural network of the receiving device; and code for sending the first known payload to the receiving device based on the first transmission setting.

[0058] Deep learning architectures can perform object recognition tasks by learning to represent input at successively higher levels of abstraction in each layer, thereby building up useful feature representations of the input data. In this way, deep learning addresses a major bottleneck in traditional machine learning. Before the advent of deep learning, machine learning approaches to object recognition problems might rely heavily on human-designed features, perhaps combined with shallow classifiers. For example, a shallow classifier might be a two-class linear classifier, where the weighted sum of the feature vector components is compared to a threshold to predict which class the input belongs to. Human-designed features might be templates or kernels customized for a specific problem domain by engineers with domain expertise. In contrast, deep learning architectures can learn representations similar to those that human engineers might design, but through training. Furthermore, deep networks can learn to represent and recognize new types of features that humans might not have considered.

[0059] Deep learning architectures can learn hierarchies of features. For example, if provided with visual data, the first layer can learn to recognize relatively simple features in the input stream, such as edges. In another example, if provided with auditory data, the first layer can learn to recognize the spectral power of specific frequencies. The second layer, taking the output of the first layer as input, can learn to recognize combinations of features, such as simple shapes in visual data or combinations of sounds in auditory data. For example, higher layers can learn to represent complex shapes in visual data or words in auditory data. Even higher layers can learn to recognize common visual objects or spoken phrases.

[0060] Deep learning architectures can perform particularly well when applied to problems that have a natural hierarchical structure. For example, the classification of motor vehicles might benefit from first learning to recognize wheels, windshields, and other features. These features can then be combined in different ways at higher levels to identify cars, trucks, and airplanes.

[0061] Neural networks can be designed with a variety of connection patterns. In a feedforward network, information is passed from lower layers to higher layers, with each neuron in a given layer communicating with neurons in higher layers. As described above, hierarchical representations can be built up in successive layers of a feedforward network. Neural networks may also have recurrent or feedback (also known as top-down) connections. In a recurrent connection, the output from a neuron in a given layer can be passed to another neuron in the same layer. Recurrent architectures may help recognize patterns that span multiple chunks of input data that are passed sequentially to the neural network. The connections from neurons in a given layer to neurons in lower layers are called feedback (or top-down) connections. Networks with many feedback connections may be helpful when the recognition of high-level concepts can help discern specific low-level features of the input.

[0062] The connections between neural network layers can be fully connected or partially connected. Figure 4A Shown is an example of a fully connected neural network 402. In the fully connected neural network 402, neurons in a first layer may transmit their output to every neuron in a second layer, such that every neuron in the second layer will receive input from every neuron in the first layer. Figure 4B An example of a locally connected neural network 404 is shown. In the locally connected neural network 404, neurons in a first layer may be connected to a limited number of neurons in a second layer. More generally, the locally connected layers of the locally connected neural network 404 may be configured such that each neuron in the layer will have the same or similar connection pattern, but the values ​​of the connection strengths may be different (e.g., 410, 412, 414, and 416). The locally connected connection pattern can produce spatially different receptive fields in higher layers because higher layer neurons in a given area may receive input that is adjusted through training to properties of a limited portion of the total input to the network.

[0063] An example of a locally connected neural network is a convolutional neural network. Figure 4C An example of a convolutional neural network 406 is shown. The convolutional neural network 406 can be configured such that the connection strengths associated with the inputs of each neuron in the second layer are shared (e.g., 408). Convolutional neural networks may be well suited for problems where the spatial location of the input is meaningful.

[0064] One type of convolutional neural network is the deep convolutional network (DCN). Figure 4D A detailed example of a DCN 400 is shown, which is designed to recognize visual features from an image 426 input from an image capture device 430 (such as a vehicle-mounted camera). The DCN 400 of the current example can be trained to recognize traffic signs and numbers on traffic signs. Of course, the DCN 400 can be trained for other tasks, such as recognizing lane markings or identifying traffic lights.

[0065] DCN 400 can be trained using supervised learning. During training, an image, such as image 426 of a speed limit sign, can be presented to DCN 400, and a forward pass can be computed to produce output 422. DCN 400 can include a feature extraction portion and a classification portion. Upon receiving image 426, convolution layer 432 can apply a convolution kernel (not shown) to image 426 to generate a first set of feature maps 418. For example, the convolution kernel of convolution layer 432 can be a 5×5 kernel that generates a 28×28 feature map. In this example, because four different feature maps are generated in first set of feature maps 418, four different convolution kernels are applied to image 426 in convolution layer 432. Convolution kernels can also be referred to as filters or convolution filters.

[0066] The first set of feature maps 418 may be subsampled by a max pooling layer (not shown) to generate a second set of feature maps 420. The max pooling layer reduces the size of the first set of feature maps 418. That is, the size of the second set of feature maps 420 (such as 14×14) is smaller than the size of the first set of feature maps 418 (such as 28×28). The reduced size provides similar information to subsequent layers while reducing memory consumption. The second set of feature maps 420 may be further convolved via one or more subsequent convolutional layers (not shown) to generate one or more subsequent sets of feature maps (not shown).

[0067] exist Figure 4D In the example of , the second set of feature maps 420 is convolved to generate a first feature vector 424. In addition, the first feature vector 424 is further convolved to generate a second feature vector 428. Each feature of the second feature vector 428 can include a number corresponding to a possible feature of the image 426, such as "symbol", "60", and "100". A softmax function (not shown) can convert the numbers in the second feature vector 428 into probabilities. Thus, the output 422 of the DCN 400 is the probability that the image 426 includes one or more features.

[0068] In this example, the probabilities of "symbol" and "60" in output 422 are higher than the probabilities of other values ​​in output 422, such as "30," "40," "50," "70," "80," "90," and "100." Before training, output 422 generated by DCN 400 may be incorrect. Therefore, the error between output 422 and the target output can be calculated. The target output is the ground truth of image 426 (e.g., "symbol" and "60"). The weights of DCN 400 can then be adjusted so that output 422 of DCN 400 is closer to the target output.

[0069] To adjust the weights, the learning algorithm can calculate a gradient vector with respect to the weights. The gradient can indicate the amount by which the error will increase or decrease if the weights are adjusted. At the top layer, the gradient can directly correspond to the value of the weights connecting the activated neurons in the penultimate layer to the neurons in the output layer. In lower layers, the gradient can depend on the value of the weights and the calculated error gradients of higher layers. The weights can then be adjusted to reduce the error. This way of adjusting weights can be referred to as "backpropagation" because it involves a "backward pass" through the neural network.

[0070] In practice, the error gradient of the weights can be calculated over a small number of examples so that the calculated gradient is close to the error gradient of the ground truth. This approximation method can be called stochastic gradient descent. Stochastic gradient descent can be repeated until the achievable error rate of the entire system has stopped decreasing, or until the error rate has reached a target level. After learning, a new image can be presented to the DCN (e.g., image 426 of a speed limit sign), and a forward pass through the network can produce output 422, which can be considered an inference or prediction of the DCN.

[0071] A deep belief network (DBN) is a probabilistic model that includes multiple layers of hidden nodes. A DBN can be used to extract a hierarchical representation of a training dataset. A DBN can be obtained by stacking multiple layers of restricted Boltzmann machines (RBMs). An RBM is an artificial neural network that can learn a probability distribution over a set of inputs. Because RBMs can learn probability distributions without information about the class to which each input should be classified, RBMs are often used for unsupervised learning. Using a hybrid unsupervised and supervised paradigm, the bottom RBM of a DBN can be trained in an unsupervised manner and can be used as a feature extractor, while the top RBM can be trained in a supervised manner (based on the joint distribution of inputs from previous layers and the target class) and can be used as a classifier.

[0072] A deep convolutional network (DCN) is a network of convolutional networks configured with additional pooling and normalization layers. DCNs have achieved state-of-the-art performance on many tasks. DCNs can be trained using supervised learning, where the input and output targets are known for many examples and are used to modify the network's weights using gradient descent.

[0073] A DCN can be a feedforward network. Furthermore, as described above, connections from neurons in the first layer of a DCN to a set of neurons in the next higher layer are shared among the neurons in the first layer. The feedforward and shared connections of a DCN can be used for fast processing. For example, the computational burden of a DCN can be much smaller than that of a similarly sized neural network that includes recurrent or feedback connections.

[0074] The processing of each layer of a convolutional network can be thought of as a spatially invariant template or basis projection. If the input is first decomposed into multiple channels, such as the red, green, and blue channels of a color image, then the convolutional network trained on this input can be thought of as three-dimensional, with two spatial dimensions along the axes of the image and the third dimension capturing color information. The output of the convolutional connection can be thought of as forming a feature map in the subsequent layer, with each element of the feature map (e.g., 220) receiving input from a series of neurons in the previous layer (e.g., feature map 218) and from each of the multiple channels. The values ​​in the feature map can be further processed with nonlinearities (such as rectification, max(0,x)). Values ​​from neighboring neurons can be further pooled, which corresponds to downsampling and can provide additional local invariance and dimensionality reduction. Normalization, corresponding to whitening, can also be applied by lateral inhibition between neurons in the feature map.

[0075] The performance of deep learning architectures can improve as more labeled data points become available or as computing power increases. Modern deep neural networks are often trained with computing resources thousands of times greater than those available to the average researcher 15 years ago. New architectures and training paradigms have the potential to further improve deep learning performance. Rectified linear units can reduce the training problem known as vanishing gradients. New training techniques can reduce overfitting, enabling larger models to achieve better generalization. Encapsulation techniques can extract data within a given receptive field and further improve overall performance.

[0076] Figure 5 is a block diagram illustrating a deep convolutional network 550. Based on connectivity and weight sharing, the deep convolutional network 550 may include multiple different types of layers. Figure 5 As shown, the deep convolutional network 550 includes convolution blocks 554A and 554B. Each convolution block 554A and 554B can be configured with a convolution layer (CONV) 356, a normalization layer (LNorm) 558, and a maximum pooling layer (MAX POOL) 560.

[0077] The convolution layer 556 may include one or more convolution filters that can be applied to the input data to generate a feature map. Although only two convolution blocks 554A and 554B are shown, the present disclosure is not limited thereto, and any number of convolution blocks 554A and 554B may be included in the deep convolutional network 550 based on design preferences. The normalization layer 558 may normalize the output of the convolution filter. For example, the normalization layer 558 may provide whitening or lateral suppression. The maximum pooling layer 560 may provide spatial downsampling aggregation to achieve local invariance and dimensionality reduction.

[0078] For example, the parallel filter banks of the deep convolutional network can be loaded onto the CPU 302 or GPU 304 of the SOC 300 to achieve high performance and low power consumption. In alternative embodiments, the parallel filter banks can be loaded onto the DSP 306 or ISP 316 of the SOC 300. In addition, the deep convolutional network 550 can access other processing blocks that may be present on the SOC 300, such as the sensor processor 314 and the navigation module 320 dedicated to sensors and navigation, respectively.

[0079] The deep convolutional network 550 may also include one or more fully connected layers 562 (FC1 and FC2). The deep convolutional network 550 may also include a logistic regression (LR) layer 564. Between each layer 556, 558, 560, 562, 564 of the deep convolutional network 550 are weights (not shown) to be updated. The output of each layer (e.g., 556, 558, 560, 562, 564) can be used as an input to subsequent layers (e.g., 556, 558, 560, 562, 564) in the deep convolutional network 550 to learn hierarchical feature representations from the input data 552 (e.g., images, audio, video, sensor data, and / or other input data) provided at the first convolutional block 554A. The output of the deep convolutional network 550 is a classification score 566 for the input data 552. The classification score 566 can be a set of probabilities, where each probability is a probability of the input data, including features from a set of features.

[0080] Figure 6 6 is a diagram illustrating an example 600 of multi-TRP (mTRP) communication according to various aspects of the present disclosure. Figure 6 As shown, multiple TRPs 610 (shown as TRP A and TRP B) can communicate with the same UE 620 in a coordinated manner (eg, using coordinated multipoint transmission, etc.) to improve reliability, increase throughput, etc. UE 620 can be a reference Figure 1 The TRP 610 can coordinate communications via the backhaul when the TRP 610 is co-located with the same base station (such as the one shown in FIG. Figure 1 When the TRP 610 is located at a base station 110 as described above, the backhaul may have lower latency and / or higher transmission capacity. The TRP 610 may correspond to different antenna arrays of the same base station. Alternatively, when the TRP 610 is located at a different base station, the latency may increase and / or the transmission capacity may decrease. Figure 6The TRPs 610 (e.g., TRP A and TRP B) can be examples of a multi-TRP group. In some examples, a multi-TRP group can be a group of TRPs 610 that communicate with the same UE, are managed as a group by an access node controller, transmit the same physical downlink shared channel (PDSCH) 640, transmit the respective PDSCHs 640 simultaneously and contemporaneously, etc.

[0081] like Figure 6 As shown, a physical downlink control channel (PDCCH) 630 can schedule downlink communications for one or two TRPs 610. The downlink communications scheduled by the PDCCH 630 can be sent via a physical downlink shared channel (PDSCH) 640. In one configuration, TRP A and TRP B coordinate transmissions via PDSCH 640. That is, TRP A and TRPB can perform the same transmission. In another configuration, TRP A and TRP B perform different transmissions via PDSCH 640. For example, the transmissions can include one or more of different payloads, different modulation and coding schemes, different transmit powers, and / or different repetition schemes. For example, in a first multi-TRP transmission mode (e.g., mode 1), the PDCCH 630 can schedule downlink data communications for two TRPs 610 via PDSCH 640. In this example, two TRPs 610 (e.g., TRP A and TRPB) can send communications to the UE 620 on the same PDSCH 640. In another example, the TRP 610 can be transmitted independently in different (e.g., non-contiguous) resource block (RB) groups and / or different symbol groups. Additionally or alternatively, the TRP 610 can be transmitted independently using different layers (e.g., different multiple-input multiple-output (MIMO) layers). In some aspects, transmissions on different layers can occur in overlapping resource blocks and / or overlapping symbols. As another example, in a second multiple-TRP transmission mode (e.g., Mode 2), multiple PDCCHs 630 can schedule downlink data communications for multiple corresponding PDSCHs 640.

[0082] As mentioned above, Figure 3-Figure 6 are provided as examples. Other examples may differ from the reference Figure 3-Figure 6 described.

[0083] Artificial intelligence (AI) / machine learning (ML) functions can improve wireless communications at base stations and / or user equipment (UE). An AI / ML module (such as a reference Figure 3The SOC 300 described herein can be implemented at the UE, the base station, or used for distributed functions, jointly used on the UE and the base station. More specifically, the AI / ML module can execute a neural network or machine learning model. While the present disclosure may involve an AI / ML module on the UE side, the present disclosure explicitly considers the case of an autoencoder. In the autoencoder scenario, joint training occurs.

[0084] The AI / ML module can be trained to perform one or more tasks. The weights and biases of the neural network or machine learning model can be derived from the training data during the offline training phase. For ease of explanation, the examples provided below refer to machine learning models. The present disclosure also contemplates neural networks or other types of trainable models. Before offline training, the weights and biases can be set to default values ​​(e.g., initial values). However, offline training does not take into account the dynamic real-world environment. Online training, on the other hand, takes into account the dynamic real-world environment. For example, online training can be accomplished using known air transmissions (or signaling) that reflect the wireless environment. For the online training phase (e.g., training program), the weights and biases can be set from the previous offline training phase. Online training can fine-tune the neural network in terms of wireless channel, noise, and / or other environmental characteristics. Online training can be referred to as retraining or updating of the neural network.

[0085] For example, a machine learning model can estimate the symbols of a MIMO demapper. The machine learning model can implement a function such as y = Hx + n, where y is the received vector, x is the transmitted symbol vector, n is the noise vector, and H is the channel matrix. The machine learning model can be trained offline to estimate (estimated symbols). The accuracy of the estimates generated by the machine learning model can be improved by fine-tuning the machine learning model through online training. During the online training phase, the received observations (y) and the estimated channel matrix (H) are input to the machine learning model. In addition, the machine learning model outputs the detected transmitted symbols, such as the estimated symbols In a conventional system, a device (e.g., UE or gNB) may decode y to convert the decoded Serving as ground truth data for a neural network. According to aspects of the present disclosure, ground truth data with known payloads is used to train a neural network. Because the payload is known, the ML model can use the training data to learn how to interpret the symbols in the presence of channel characteristics, noise characteristics, etc.

[0086] The above example is an example of online training using known data transmission. The above example can be implemented in a network with multiple transmit receive points (TRPs) and / or base stations, where multiple TRPs and / or base stations send known payloads as training data to a given UE (given MIMO rank, MCS, etc.), and the UE uses the known data to train the machine learning model. Aspects of the present disclosure are not limited to MIMO demappers. Other types of functions are considered for machine learning models. In the present disclosure, known data can be used interchangeably with known payloads.

[0087] As described above, some machine learning models can be trained using known transmissions, such as reference signals. For example, a demodulation reference signal (DMR) or a channel state information reference signal (CSI-RS) can be a known transmission. However, reference signals may not provide appropriate information for certain types of machine learning models (e.g., neural networks). Certain machine learning models, such as those specified for log-likelihood ratio (LLR) calculations or MIMO demapping, can use known data transmissions in addition to or instead of known transmissions. The known data transmissions can include payloads, such as data messages or control signaling, transmitted on, for example, a physical downlink control channel (PDCCH), a physical uplink control channel (PUCCH), a physical downlink shared channel (PDSCH), and / or a physical uplink shared channel (PUSCH).

[0088] In traditional systems, after the data is decoded, known data transmissions can be considered to be known payloads. In one example, a known data transmission can include, for example, a system information block (SIB) that is known to repeat periodically. For example, when the system information (SI) changes, the SIB can be repeated between boundaries. In addition, in traditional systems, unicast transmissions can be classified as known transmissions in response to passing a cyclic redundancy check (CRC). In traditional systems, decoding transmissions used to train machine learning models increases both memory usage and computational overhead, resulting in high latency. For example, received modulation symbols are stored until decoding is complete, increasing memory usage and computational overhead. Only after decoding is complete can the symbols be used for training.

[0089] Therefore, it is preferable that both the base station and the UE know (e.g., agree) in advance on the payloads transmitted by certain physical channels. In one configuration, the known payloads can be used as ground truth data for online training of a neural network. In one configuration, the known payloads are sent from multiple TRPs and / or base stations. For ease of explanation, the examples provided below describe multiple TRPs. However, aspects of the present disclosure are not limited to multiple TRPs, and multiple TRPs and / or base stations are also contemplated.

[0090] For the purpose of online training, the signaling framework can enable the UE and / or multiple TRPs to train their respective neural networks (e.g., machine learning models). Known payloads known to both parties can be beneficial in many cases because the receiver does not need to fully decode the payload to find the labels (e.g., ground truth) for training.

[0091] Figure 7 is a block diagram illustrating an example of an artificial neural network 700 according to aspects of the present disclosure. Figure 7 In the example of FIG, neural network 700 is implemented on a device such as a UE or a base station. The UE may be Figure 1 As shown in FIG120, the base station may be Figure 1 The base station 110 is shown. The base station may also be referred to as a TRP. The neural network 700 may be implemented during an offline training phase. The neural network 700 may be trained to generate a neural network based on known data (x k ) produces an estimate (y). A set of parameters (w), such as weights and biases of the neural network 700, can be learned based on offline training. The parameters (w) can be used by layers (such as layer 1, layer 2, and layer 3) of the neural network 700. Layer 3 can be a fully connected layer.

[0092] When a device such as a UE is deployed, the UE can update the neural network 700 to take into account changes in one or more characteristics such as channel characteristics and / or noise characteristics. Online training can be performed to take into account one or more changes. That is, one or more parameters (w) can be updated to take into account one or more changes. In one configuration, the parameter (w) is updated based on an error (e.g., an error gradient) determined by a loss function 708. The loss function 708 compares the estimate (y) with a ground truth value (e.g., an expected value). The error is the difference (e.g., loss) between the estimate (y) and the ground truth value. The error is output from the loss function 708 to the neural network 700, and the error is back-propagated through the neural network 700 to update the parameter (w).

[0093] As described above, aspects of the present disclosure use known payloads to improve online training. In one configuration, the device stores an estimate of the known payload (y*) in a memory, such as a memory of the device and / or a memory associated with the neural network 700. For example, the value may be a value obtained in response to receiving the known data (x k ) and the symbols estimated by the MIMO demapper. That is, based on offline training, the neural network 700 can respond to receiving the known data (x k ) to determine the estimated value (y*) that should be generated. The known estimated value (y*) can be called the ground truth data or the ground truth label.

[0094] According to aspects of the present disclosure, during the online training phase, in order to account for changes in one or more characteristics, the device requests one or more transmitters 710 (such as multiple TRPs) to send known data (x k ). Known data (x k ) can be processed by the neural network 700 to produce an estimate (y). In this example, the estimate (y) may differ from the ground truth data (y*) due to the changing characteristics. In this example, the loss function 708 determines the loss between the estimate (y) and the ground truth data (y*). The parameter (w) can be updated based on the loss. Updating the parameter (w) can improve the accuracy of the predictions generated by the neural network 700 based on the unknown data (x u ) generates an estimate (y) of the accuracy. During online training, the known data (x k ) can be used to send unknown data (x u ) is sent using the transmission settings of . After training (e.g., online training and / or offline training), the loss function 708 can be excluded from the process used to generate the estimate (y).

[0095] As described above, multiple TRPs can send known data to the UE. The transmission can be customized for a specific neural network. That is, based on one type of neural network, the receiving device may require different types of data. Therefore, a receiver-specific known payload can be generated to include data for a specific receiver. For example, multiple TRPs can send known data on a specific set of beam pairs using a specific modulation and coding scheme (MCS) and / or one or more specific ranks. Similarly, the UE can send known data to multiple TRPs so that multiple TRPs can train their neural networks. The training data is specifically customized for one or more specific receiving devices.

[0096] The known payload can be generated independently by the UE or multiple TRPs. For example, the known payload can be based on a scrambling seed configured by Radio Resource Control (RRC), or can be any sequence known to both the UE and multiple TRPs.

[0097] In one configuration, a UE request triggers the transmission of a known payload. In this configuration, the known payload is sent in a downlink shared channel (e.g., PDSCH). In another configuration, a request from one or more of multiple TRPs triggers the transmission of a known payload from the UE. The UE may send the known payload on an uplink shared channel (e.g., PUSCH). The UE and base station requests may include specific characteristics such as MCS, beam pairs, rank, etc. The training of a specific neural network will be based on these requested characteristics.

[0098] The transmission of a known payload in the downlink shared channel for training may also be initiated by an indication sent by one or more TRPs. This indication may be sent in a downlink control channel or a media access control (MAC) layer control element (MAC-CE). In other configurations, the transmission of a known payload in the uplink shared channel may be initiated by a UE indication. This indication may be sent in an uplink control channel or a MAC-CE.

[0099] In one configuration, the location of the known payload (e.g., on which physical channel) and the exact time-frequency resources, periodicity (aperiodic, semi-persistent, periodic, etc.), duration, aggregation level (for PDCCH), payload size, etc. of the known payload are explicitly signaled separately from the payload. For example, the base station can send information about the known payload. Explicit signaling can be via RRC, MAC-CE, or via downlink control information (DCI).

[0100] Some non-limiting examples of known data positioning include PDSCH, PUSCH, PUCCH and PDCCH. For downlink control channels (e.g., PDCCH), the known payload can be paired with a known payload sent in a downlink shared channel (e.g., PDSCH) or a known payload sent by the UE via an uplink shared channel (e.g., PUSCH). For downlink and uplink control channels (e.g., PDCCH and PUCCH), the known payload can be periodic, aperiodic or semi-persistent, and as mentioned above, the parameters are signaled separately. These parameters may include the positioning and time (e.g., duration) of the transmission of the known payload. In the case of periodic transmission, semi-persistent scheduling or configuration grants (CGs) may be defined. The known payload may be mapped to semi-persistent scheduling or configuration grants.

[0101] In one configuration, the neural network of the UE is trained based on joint transmissions from multiple TRPs. That is, known payloads from multiple TRPs can be multiplexed and received at the UE. In one configuration, the transmission settings used to send unknown payloads in the deployment phase (e.g., real-world scenarios) are used to send known payloads during the online and / or offline training phase. Unknown payloads may refer to data or control channel transmissions from multiple TRPs to the UE, and data or control channel transmissions from the UE to multiple TRPs. These settings may include, for example, the multiplexing type, beam pair, rank, modulation and coding scheme (MCS), and / or other transmission settings used for the transmission.

[0102] For example, multiple TRPs can have different precodings. In one configuration, the UE knows the precoding. That is, the precoding can be a known precoding. The precoding can be signaled to the UE separately from the known payload. For each TRP, the precoding used for unknown payloads during the deployment phase should be the same as the precoding used to send known payloads during the offline training phase or the online training phase.

[0103] Additionally, multiple TRPs can be multiplexed for transmission during the deployment phase. The multiplexing can be spatial division multiplexing (SDM), time division multiplexing (TDM), or frequency division multiplexing (FDM). In one configuration, the multiplexing type used to send unknown payloads during the deployment phase should be the same as the multiplexing type used to send known payloads during the offline training phase or the online training phase.

[0104] In one example, during the deployment phase, the TRPs use time division multiplexing for transmission. For example, the first TRP may send data, then the second TRP may send data, and then the third TRP may send data. The data from the first TRP, the second TRP, and the third TRP may be multiplexed. In this example, known payloads are also sent according to the time division multiplexing used to send data (e.g., unknown payloads). In one configuration, the ranks across multiple TRPs may be coordinated. For example, the transmission from the first TRP to the UE may be rank 1, while the transmission from the second TRP to the UE may be rank 2. During the offline training phase and the online training phase, the transmission of known payloads may be coordinated between multiple TRPs to follow the rank of the TRPs used in the deployment phase.

[0105] When a known payload is sent on an uplink control channel, parameters related to the known payload may be separately signaled. This signaling may be an RRC message or via MAC-CE. Separately signaled parameters may include, for example, the PUCCH format, uplink control information (UCI) type, and / or payload size.

[0106] Figure 8 8 is a timing diagram illustrating an example 800 for sending a known payload according to aspects of the present disclosure. Figure 8 As shown, at time t1, the artificial neural network of UE 802 is trained. Figure 8 The UE 802 may be one of a group of UEs (eg, a UE group). Figure 8Only one UE 802 is shown in the example. The artificial neural network can be trained offline (e.g., before the UE 802 is deployed). After the UE 802 is deployed, the UE 802 can send a known payload request to the first TRP 804 at time t2a. The request can be sent via a physical uplink control channel (PUCCH) or a media access control layer (MAC) control element (CE). In one configuration, the known payload request indicates whether the payload is requested for self-training (e.g., one UE) or joint training (e.g., a group of UEs). For example, joint training can be requested for a coordinated multi-point (CoMP) group having multiple TRPs and multiple UEs, where one UE can be specified for each TRP. According to aspects of the present disclosure, the known payload is sent in the same manner as the unknown payload in the CoMP group. In one configuration, the first TRP 804 is the serving TRP of the multi-TRP (mTRP) group. Figure 8 In the example of , the first TRP 804 and the second TRP 806 form a multi-TRP group. The multi-TRP group is not limited to two TRPs, and additional TRPs may also be considered. As described above, a known payload request may be sent to receive known data to update the artificial neural network. In one configuration, the known payload request may depend on the goal of the training (e.g., update). In addition, the known payload request may include a list of other TRPs that are requested to send known data. The list of other TRPs may be sorted based on a sorting criterion. For example, the list of other TRPs may be sorted from largest to smallest reference signal received power (RSRP). In addition, the known payload request may include transmission settings specified for training the neural network. The transmission settings may correspond to transmission settings for unknown data. For example, the transmission settings may include the rank of each TRP of the multi-TRP group, the precoding, beam pairs and / or modulation and coding scheme (MCS) of each TRP of the multi-TRP group.

[0107] exist Figure 8 In the example of , the second TRP 806 is included in the list of other TRPs. Figure 8 In the example of FIG, at time t2b, the first TRP 804 forwards the known payload request to the second TRP 806. That is, the first TRP 804 (e.g., the serving TRP) forwards the known payload request to the other TRPs included in the other TRP list. At time t3, the TRPs 804 and 806 send the known payload to the UE 802 according to the transmission settings. Figure 8In the example of , a known payload is sent in response to a known payload request sent by UE 802. In another configuration, the transmission of the known payload is triggered by an indication from one or more TRPs 804 and 806. The indication can be sent via a downlink control channel or MAC-CE. According to aspects of the present disclosure, each TRP 804 and 806 can send a different known payload. In addition, the known payload can be multiplexed according to a multiplexing scheme such as spatial division multiplexing (SDM), time division multiplexing (TDM), or frequency division multiplexing (FDM). In one configuration, before sending the known payload, one or more of the TRPs 804 and 806 can send known payload information, including one or more of the periodicity, time and frequency resources, and payload size of the known payload. The known payload information can be sent via radio resource control (RRC) signaling, MAC-CE, or downlink control information (DCI). The known payload can be sent on a downlink control channel (e.g., PDCCH). In one configuration, each known payload sent on the PDCCH may correspond to another known payload sent on a downlink shared channel (e.g., PDSCH). Alternatively, the known payload may be sent only on the downlink shared channel. In addition, the known payload may be sent according to periodic scheduling, semi-persistent scheduling, or aperiodically. In one configuration, the known payload may be scrambled based on a radio resource control (RRC) scrambling seed. Based on the RRC scrambling seed, the payload may be known to the UE 802, the first TRP 804, and the second TRP 806.

[0108] exist Figure 8 In the example of , at time t4, UE 802 updates the artificial neural network. The update can be an example of retraining the artificial neural network. The update performed at time t4 can be an online update. In one configuration, as with respect to Figure 7 As described in the example of FIG, an artificial neural network generates ground truth values ​​based on known payloads received from TRPs 804 and 806. In one configuration, the artificial neural network processes each known payload to generate an estimated value. The weights and parameters of the artificial neural network can be updated by comparing the estimated value with the corresponding ground truth value.

[0109] Figure 9 is a timing diagram illustrating an example 900 of sending known data according to aspects of the present disclosure. Figure 9 As shown, at time t1a, the artificial neural network of the first TRP 904 is trained. In addition, at time t1b, the artificial neural network of the second TRP 906 is trained. The artificial neural network can be trained offline. The first TRP 904 can be the service TRP of a multi-TRP (mTRP) group. Figure 9 In the example shown in FIG2 , a first TRP 904 and a second TRP 906 form a multi-TRP group. A multi-TRP group is not limited to two TRPs, and additional TRPs are also contemplated. Once deployed (e.g., online), one or more of TRPs 904 and 906 may send a known payload request to UE 902 at time t2. The known payload request may be unicast to one UE 902 or multicast to a group of UEs.

[0110] Figure 9 The UE 902 may be one of a group of UEs (eg, a UE group). Figure 9 Only one UE 902 is shown in the example of FIG. The request may be sent via a physical downlink control channel (PUCCH) or a medium access control layer (MAC) control element (CE). As described above, a known payload request may be sent to receive known data to update the artificial neural network. In one configuration, the known payload request may depend on the goal of the training (e.g., update). In addition, the known payload request may include transmission settings specified for training the neural network. The transmission settings may correspond to transmission settings for unknown data. For example, the transmission settings may include beam pairs and / or modulation and coding schemes (MCS).

[0111] At time t3, UE 902 sends a known payload to TRPs 904 and 906 according to the transmission settings. UE 902 may send the same known payload to TRPs 904 and 906, or may send a different known payload to each TRP 904 and 906. Figure 9In the example of , a known payload is sent in response to a known payload request sent by one or more TRPs 904 and 906. In another configuration, the known payload transmission can be triggered by an indication from UE 902. The notification can be sent via an uplink control channel or MAC-CE. In one configuration, before sending the known payload, UE 902 can send known payload information, including one or more of the periodicity, time and frequency resources, and payload size of the known payload. The known payload information can be sent via radio resource control (RRC) signaling or MAC-CE. The known payload can be sent on an uplink control channel (e.g., PUCCH). In one configuration, the known payload sent on the PUCCH can correspond to a known payload sent on an uplink shared channel (e.g., PUSCH). Alternatively, the known payload can be sent only on an uplink shared channel. In addition, the known payload can be sent according to periodic scheduling, semi-persistent scheduling, or aperiodically. In one configuration, the known payload may be scrambled based on a radio resource control (RRC) scrambling seed. Based on the RRC scrambling seed, the payload may be known to the UE 902, the first TRP 904, and the second TRP 906.

[0112] exist Figure 9 In the example of FIG. 1 , at time t4a, the first TRP 904 updates the artificial neural network. Additionally, at time t4b, the second TRP 906 updates the artificial neural network. The update may be an example of retraining the artificial neural network. The updates performed at times t4a and t4b may be online updates. In one configuration, as described with respect to FIG. Figure 7 As described in the example of FIG, an artificial neural network generates ground truth values ​​based on known payloads received from UE 902. In one configuration, the artificial neural network processes each known payload to generate an estimated value. The weights and parameters of the artificial neural network can be updated by comparing the estimated value with the corresponding ground truth value.

[0113] Figure 10 is a diagram illustrating an example process 1000, such as performed by a receiving device, in accordance with various aspects of the present disclosure. The example process 1000 is an example of requesting a known payload, such as for training an artificial neural network.

[0114] like Figure 10As shown, in some aspects, process 1000 may include sending a request to a first transmitting device in a group of transmitting devices for a first known payload for training an artificial neural network of a receiving device (block 1002). For example, the receiving device (e.g., using antenna 252, DEMOD / MOD 254, TX MIMO processor 266, transmit processor 264, controller / processor 280, and / or memory 282) may send a request to a first transmitting device in the group of transmitting devices for a first known payload for training an artificial neural network of the receiving device. Process 1000 may also include receiving the first known payload from the first transmitting device in response to the request (block 1004). For example, the receiving device (e.g., using antenna 252, DEMOD / MOD 254, MIMO detector 256, receive processor 258, controller / processor 280, and / or memory 282) may receive the first known payload from the first transmitting device in response to the request. Process 1000 may include updating the artificial neural network at the receiving device based on at least the first known payload (block 1006). For example, a receiving device (e.g., using antenna 252, DEMOD / MOD 254, MIMO detector 256, TX MIMO processor 266, receive processor 258, transmit processor 264, controller / processor 280, and / or memory 282) may update an artificial neural network at the receiving device based on at least the first known payload.

[0115] Figure 11 1 is a diagram illustrating an example process 1100, for example, performed by a sending device, according to various aspects of the present disclosure. The example process 1100 is an example of sending known data for training an artificial neural network.

[0116] like Figure 11As shown, in some aspects, process 1100 may include transmitting a first unknown payload to a receiving device in a group of receiving devices based on a first transmission setting of a first transmitting device (block 1102). For example, the transmitting device (e.g., using antenna 252, DEMOD / MOD 254, TX MIMO processor 266, transmit processor 264, controller / processor 280, and / or memory 282) may transmit the first unknown payload to a receiving device in a group of receiving devices based on the first transmission setting of the first transmitting device. Process 1100 may also include receiving a request from a receiving device for a first known payload for training an artificial neural network of the receiving device (block 1104). For example, the transmitting device (e.g., using antenna 252, DEMOD / MOD 254, MIMO detector 256, receive processor 258, controller / processor 280, and / or memory 282) may receive a request from a receiving device for a first known payload for training an artificial neural network of the receiving device. Process 1100 may also include transmitting the first known payload to the receiving device based on the first transmission setting (block 1106). For example, a transmitting device (e.g., using antenna 252, DEMOD / MOD 254, TX MIMO processor 266, transmit processor 264, controller / processor 280, and / or memory 282) may transmit a first known payload to a receiving device based on a first transmission setting.

[0117] The foregoing disclosure provides illustration and description, but is not intended to be exhaustive or to limit the aspects to the precise forms disclosed. Modifications and variations are possible in light of the above disclosure or may be acquired from practice of these aspects.

[0118] As used herein, the term "component" is intended to be broadly interpreted as hardware, firmware, and / or a combination of hardware and software. As used herein, a processor is implemented in hardware, firmware, and / or a combination of hardware and software.

[0119] Some aspects are described herein in conjunction with thresholds. As used herein, depending on the context, satisfying a threshold may mean a value is greater than a threshold, greater than or equal to a threshold, less than a threshold, less than or equal to a threshold, equal to a threshold, not equal to a threshold, etc.

[0120] It will be apparent that the systems and / or methods described herein can be implemented in various forms of hardware, firmware, and / or a combination of hardware and software. The actual dedicated control hardware or software code used to implement these systems and / or methods is not limited in these respects. Thus, the operation and behavior of the systems and / or methods are described herein without reference to specific software code—it is understood that software and hardware can be designed to implement the systems and / or methods based, at least in part, on the description herein.

[0121] Although the specific combination of features is recorded in the claims and / or disclosed in the specification, these combinations are not intended to limit the disclosure of various aspects. In fact, many of these features can be combined in a manner that is not specifically recorded in the claims and / or not specifically disclosed in the specification. Although each dependent claim listed below can be directly subordinate to only one claim, the disclosure of various aspects includes that each dependent claim is combined with each other claim in the claim set. The phrase "at least one" mentioned in the list of items refers to any combination of those items, including single members. For example, "at least one of a, b or c" is intended to cover a, b, c, ab, ac, bc and abc, as well as any combination with multiples of the same elements (for example, aa, aaa, aab, aac, abb, acc, bb, bbb, bbc, cc and ccc, or any other sorting of a, b and c).

[0122] Unless explicitly described as such, the elements, actions or instructions used herein should not be interpreted as key or necessary. In addition, as used herein, the articles "a" and "an" are intended to include one or more projects and can be used interchangeably with "one or more". In addition, as used herein, the terms "set" and "group" are intended to include one or more projects (e.g., related projects, unrelated projects, combinations of related and unrelated projects, etc.) and can be used interchangeably with "one or more". When only one project is meant, the phrase "only one" or similar language is used. In addition, as used herein, the terms "has", "have", "having" etc. are open terms. In addition, the phrase "based on" is intended to mean "based at least in part on", unless otherwise explicitly stated.

Claims

1. A wireless communication method performed by a receiving device, comprising: sending a request to a first sending device in a group of sending devices for a first known payload from the first sending device and a second known payload from a second sending device in the group of sending devices, the first known payload and the second known payload being used to train an artificial neural network of the receiving device; receiving the first known payload from the first sending device in response to the request; receiving, from the second sending device in response to the request, a second known payload, the second known payload being multiplexed with the first known payload; as well as updating the artificial neural network at the receiving device based on at least the first known payload and the second known payload, The first known payload is known to both the receiving device and the first sending device, and the second known payload is known to both the receiving device and the second sending device.

2. The method according to claim 1, wherein the first sending device is a service sending receiving point TRP, the second sending device is a second TRP, the group of sending devices includes a group of TRPs of a multi-TRP group, and the receiving device is a user equipment UE. The method according to claim 2 , wherein the UE is one UE in a UE group.

4. The method according to claim 2, wherein: The request also includes a first transmission setting for the first known payload and the second known payload; and The first transmission setting includes at least one of a multiplexing scheme, a rank of each TRP of the multi-TRP group, a precoding of each TRP of the multi-TRP group, a beam pair, a modulation and coding scheme MCS, or a combination thereof.

5. The method of claim 4, wherein the first transmission setting of the first known payload and the second known payload is the same as the second transmission setting of the first unknown payload from the serving TRP and the second unknown payload from the second TRP. The method of claim 4 , wherein the first known payload is different from the second known payload. 7 . The method according to claim 4 , wherein the multiplexing scheme comprises space division multiplexing (SDM), time division multiplexing (TDM), or frequency division multiplexing (FDM).

8. The method of claim 4, further comprising receiving information including periodicity, time and frequency resources, and payload size of the first known payload and the second known payload, wherein the information is received via radio resource control (RRC) signaling, a media access control layer (MAC) control element, or downlink control information (DCI).

9. The method of claim 4 , further comprising receiving the first known payload and the second known payload on a physical downlink control channel (PDCCH), each of the first known payload and the second known payload being paired with a payload received on a physical downlink shared channel (PDSCH).

10. The method of claim 4, further comprising receiving the first known payload and the second known payload according to a periodic schedule, a semi-persistent schedule, or aperiodically.

11. The method of claim 4, wherein the first known payload and the second known payload are scrambled based on a radio resource control (RRC) scrambling seed.

12. The method of claim 11, wherein the first known payload and the second known payload are known to the UE, the serving TRP, and the second TRP based on the RRC scrambling seed.

13. The method of claim 1, wherein the request occurs via a physical uplink control channel (PUCCH) or a medium access control layer (MAC) control element (CE).

14. The method of claim 13, wherein the first known payload is received in a physical downlink shared channel (PDSCH).

15. The method of claim 1, further comprising training the artificial neural network during an offline training phase, wherein updating the artificial neural network comprises retraining the artificial neural network during an online training phase.

16. The method of claim 1, further comprising generating ground truth values ​​for training the artificial neural network from the first known payload.

17. The method according to claim 16, further comprising: processing the first known payload through the artificial neural network to produce an estimate; as well as The weights and parameters of the artificial neural network are updated by comparing the estimated values ​​with the ground truth values.

18. The method of claim 1, wherein: The first transmitting device is a first user equipment (UE), the second transmitting device is a second UE, and the receiving device is a transmitting receiving point (TRP) of a multi-TRP group; and The request includes transmission settings including at least one of a multiplexing scheme, a modulation and coding scheme (MCS), or a combination thereof. The method of claim 18 , wherein the group of transmitting devices comprises a group of UEs.

20. The method of claim 19, further comprising updating the artificial neural network based on the first known payload from the first UE and the second known payload from the second UE in the group of UEs.

21. The method of claim 18, further comprising receiving the first known payload via a physical uplink control channel (PUCCH).

22. The method of claim 21, further comprising receiving a second known payload via a physical uplink shared channel (PUSCH), the first known payload being paired with the second known payload.

23. The method of claim 18, wherein the request is sent via a physical downlink control channel (PDCCH) or a medium access control layer (MAC) element (CE).

24. An apparatus for wireless communication at a receiving device, the apparatus comprising: at least one memory, the at least one memory comprising instructions; as well as at least one processor configured to execute instructions to cause the apparatus to: sending a request to a first sending device in a group of sending devices for a first known payload from the first sending device and a second known payload from a second sending device in the group of sending devices, the first known payload and the second known payload being used to train an artificial neural network of the receiving device; receiving the first known payload from the first sending device in response to the request; receiving, from the second sending device in response to the request, a second known payload, the second known payload being multiplexed with the first known payload; as well as updating the artificial neural network at the receiving device based on at least the first known payload and the second known payload, The first known payload is known to both the receiving device and the first sending device, and the second known payload is known to both the receiving device and the second sending device.

25. An apparatus according to claim 24, wherein the first sending device is a serving sending receiving point (TRP), the second sending device is a second TRP, the group of sending devices includes a group of TRPs of a multi-TRP group, and the receiving device is a user equipment (UE). The apparatus according to claim 25 , wherein the UE is one UE in a group of UEs.

27. The apparatus of claim 25, wherein: The request also includes a first transmission setting for the first known payload and the second known payload; and The first transmission setting includes at least one of a multiplexing scheme, a rank of each TRP of the multi-TRP group, a precoding of each TRP of the multi-TRP group, a beam pair, a modulation and coding scheme (MCS), or a combination thereof.

28. The apparatus of claim 27, wherein the first transmission settings of the first known payload and the second known payload are the same as the second transmission settings of the first unknown payload from the serving TRP and the second unknown payload from the second TRP.

29. The apparatus of claim 27, wherein the first known payload is different from the second known payload.

30. The apparatus of claim 27, wherein the multiplexing scheme comprises space division multiplexing (SDM), time division multiplexing (TDM), or frequency division multiplexing (FDM).

31. The apparatus of claim 27, wherein the at least one processor is further configured to cause the apparatus to receive information comprising periodicity, time and frequency resources, and payload size of the first known payload and the second known payload, wherein the information is received via radio resource control (RRC) signaling, a media access control layer (MAC) control element, or downlink control information (DCI).

32. The apparatus of claim 27, wherein the at least one processor is further configured to cause the apparatus to receive the first known payload and the second known payload on a physical downlink control channel (PDCCH), each of the first known payload and the second known payload being paired with a payload received on a physical downlink shared channel (PDSCH).

33. The apparatus of claim 27, wherein the at least one processor is further configured to cause the apparatus to receive the first known payload and the second known payload according to a periodic schedule, a semi-persistent schedule, or aperiodically.

34. The apparatus of claim 27, wherein the first known payload and the second known payload are scrambled based on a radio resource control (RRC) scrambling seed.

35. The apparatus of claim 34, wherein the first known payload and the second known payload are known to the UE, the serving TRP, and the second TRP based on the RRC scrambling seed.

36. The apparatus of claim 24, wherein the request occurs via a physical uplink control channel (PUCCH) or a medium access control layer (MAC) element (CE).

37. The apparatus of claim 36, wherein the first known payload is received in a physical downlink shared channel (PDSCH).

38. The apparatus of claim 24, wherein the at least one processor is further configured to cause the apparatus to train the artificial neural network during an offline training phase, wherein the instructions for causing the apparatus to update the artificial neural network include instructions for causing the apparatus to retrain the artificial neural network during an online training phase.

39. The apparatus of claim 24, wherein the at least one processor is further configured to cause the apparatus to generate ground truth values ​​for training the artificial neural network from the first known payload.

40. The apparatus of claim 39, wherein the at least one processor is further configured to cause the apparatus to: processing the first known payload through the artificial neural network to produce an estimate; and The weights and parameters of the artificial neural network are updated by comparing the estimated values ​​with the ground truth values.

41. The device of claim 24, wherein The first transmitting device is a first user equipment (UE), the second transmitting device is a second UE, and the receiving device is a transmitting receiving point (TRP) of a multi-TRP group; and The request includes transmission settings including at least one of a multiplexing scheme, a modulation and coding scheme (MCS), or a combination thereof.

42. The apparatus of claim 41, wherein the group of transmitting devices comprises a group of UEs.

43. The apparatus of claim 42, wherein the at least one processor is further configured to cause the apparatus to update the artificial neural network based on the first known payload from the first UE and the second known payload from the second UE in the group of UEs.

44. The apparatus of claim 41, wherein the at least one processor is further configured to cause the apparatus to receive the first known payload via a physical uplink control channel (PUCCH).

45. The apparatus of claim 44, wherein the at least one processor is further configured to cause the apparatus to receive a second known payload via a physical uplink shared channel (PUSCH), the first known payload being paired with the second known payload.

46. ​​The apparatus of claim 41, wherein the request is sent via a physical downlink control channel (PDCCH) or a medium access control layer (MAC) element (CE).

47. A receiving device for wireless communication, the receiving device comprising: means for sending, to a first sending device in a group of sending devices, a request for a first known payload from the first sending device and a second known payload from a second sending device in the group of sending devices, the first known payload and the second known payload being used to train an artificial neural network of the receiving device; means for receiving the first known payload from the first sending device in response to the request; means for receiving, in response to the request, from the second sending device, a second known payload, the second known payload being multiplexed with the first known payload; as well as means for updating, at the receiving device, the artificial neural network based on at least the first known payload and the second known payload, The first known payload is known to both the receiving device and the first sending device, and the second known payload is known to both the receiving device and the second sending device.

48. A receiving device according to claim 47, wherein the first sending device is a serving sending receiving point (TRP), the second sending device is a second TRP, the group of sending devices includes a group of TRPs of a multi-TRP group, and the receiving device is a user equipment (UE).

49. The receiving device according to claim 48, wherein: The request also includes a first transmission setting for the first known payload and the second known payload; and The first transmission setting includes at least one of a multiplexing scheme, a rank of each TRP of the multi-TRP group, a precoding of each TRP of the multi-TRP group, a beam pair, a modulation and coding scheme (MCS), or a combination thereof.

50. The receiving device of claim 49, wherein the first transmission setting of the first known payload and the second known payload is the same as the second transmission setting of the first unknown payload from the serving TRP and the second unknown payload from the second TRP.

51. The receiving device of claim 49, wherein the first known payload is different from the second known payload.

52. The receiving device of claim 47, further comprising means for training the artificial neural network during an offline training phase, wherein the means for updating the artificial neural network comprises means for retraining the artificial neural network during an online training phase.

53. The receiving device of claim 47, further comprising means for generating ground truth values ​​for training the artificial neural network from the first known payload.

54. The receiving device according to claim 53, further comprising: means for processing said first known payload through said artificial neural network to produce an estimate; as well as Means for updating weights and parameters of the artificial neural network by comparing the estimated values ​​with the ground truth values.

55. The receiving device of claim 47, wherein: The first transmitting device is a first user equipment (UE), the second transmitting device is a second UE, and the receiving device is a transmitting receiving point (TRP) of a multi-TRP group; and The request includes transmission settings including at least one of a multiplexing scheme, a modulation and coding scheme (MCS), or a combination thereof.

56. The receiving device of claim 55, wherein the group of transmitting devices comprises a group of UEs.

57. The receiving device of claim 56, further comprising means for updating the artificial neural network based on the first known payload from the first UE and the second known payload from the second UE in the group of UEs.

58. A computer-readable medium having instructions stored thereon, which, when executed by a processor, cause the processor to perform the method of any one of claims 1 to 23.

59. A wireless communication method performed by at least a first transmitting device in a group of transmitting devices, comprising: sending a first unknown payload to a receiving device in a group of receiving devices based on a first transmission setting of the first sending device; receiving, from the receiving device, a request for a first known payload for training an artificial neural network of the receiving device; as well as transmitting the first known payload to the receiving device based on the first transmission setting, wherein the first known payload is multiplexed with a second known payload from a second transmitting device of the group of transmitting devices, The first known payload is known to both the receiving device and the first sending device, and the second known payload is known to both the receiving device and the second sending device.

60. A method according to claim 59, wherein the receiving device is a user equipment (UE), the first sending device is a serving sending receiving point (TRP), the second sending device is a second TRP, and the group of sending devices includes a multi-TRP group.

61. The method of claim 60, wherein: The group of receiving devices comprises a UE group; and Sending the first known payload includes sending the first known payload to each UE in the group of UEs.

62. The method of claim 60, wherein: The request further includes a request for a second known payload of the second TRP from the multi-TRP group and a second transmission setup for the first known payload and the second known payload; The second transmission setting includes at least one of a multiplexing scheme, a rank of each TRP of the multi-TRP group, a precoding of each TRP of the multi-TRP group, a beam pair, a modulation and coding scheme (MCS), or a combination thereof; and The method also includes sending the second known payload from the second TRP.

63. The method of claim 62, wherein the second transmission settings for the first known payload and the second known payload are the same as the first transmission settings for the first unknown payload and the second unknown payload from the second TRP.

64. The method of claim 62, wherein the first known payload is different from the second known payload.

65. The method of claim 62, wherein the multiplexing scheme comprises space division multiplexing (SDM), time division multiplexing (TDM), or frequency division multiplexing (FDM).

66. The method of claim 62, further comprising sending information including periodicity, time and frequency resources, and payload size of the first known payload and the second known payload, wherein the information is sent via radio resource control (RRC) signaling, a media access control layer (MAC) control element, or downlink control information (DCI).

67. The method of claim 62, wherein the first known payload and the second known payload are transmitted on a physical downlink control channel (PDCCH), each of the first known payload and the second known payload being paired with a corresponding payload transmitted on a physical downlink shared channel (PDSCH).

68. The method of claim 62, further comprising sending information including periodicity, time and frequency resources, and payload size of the first known payload, wherein the information is received via radio resource control (RRC) signaling, a media access control layer (MAC) control element, or downlink control information (DCI).

69. The method of claim 62, wherein the first known payload and the second known payload are sent according to a periodic schedule, a semi-persistent schedule, or aperiodically.

70. The method of claim 62, further comprising scrambling the first known payload and the second known payload based on a radio resource control (RRC) scrambling seed.

71. The method of claim 70, wherein the first known payload and the second known payload are known to the UE, the serving TRP, and the second TRP based on the RRC scrambling seed.

72. The method of claim 59, wherein receiving the request comprises receiving the request via a physical uplink control channel (PUCCH) or a medium access control layer (MAC) element (CE).

73. The method of claim 72, further comprising transmitting the first known payload via a physical downlink shared channel (PDSCH).

74. The method of claim 59, wherein: The first transmitting device is a first user equipment (UE), and the receiving device is a transmitting receiving point (TRP) of a multi-TRP group; The request includes a second transmission setting for the first known payload; The second transmission setting comprises at least one of a rank, a beam pair, a modulation and coding scheme (MCS), or a combination thereof; Sending the first known payload includes sending the first known payload to each TRP in the multi-TRP group.

75. The method of claim 74, wherein: The group of transmitting devices includes a group of UEs; The request further includes a request for a second known payload from a second UE of the group of UEs; and The second transmission settings include transmission settings for the second known payload.

76. The method of claim 74, wherein the second transmission setting of the first known payload is the same as the first transmission setting of the first known payload.

77. The method of claim 74, further comprising transmitting the first known payload via a physical uplink control channel (PUCCH).

78. The method of claim 77, further comprising transmitting a second known payload via a physical uplink shared channel (PUSCH), the first known payload being paired with the second known payload.

79. The method of claim 74, wherein receiving the request comprises receiving the request via a physical downlink control channel (PDCCH) or a medium access control layer (MAC) control element (CE).

80. An apparatus for wireless communication at a first transmitting device of a group of transmitting devices, comprising: at least one memory, the at least one memory comprising instructions; as well as at least one processor configured to execute instructions to cause the apparatus to: sending a first unknown payload to a receiving device in a group of receiving devices based on a first transmission setting of the first sending device; receiving, from the receiving device, a request for a first known payload for training an artificial neural network of the receiving device; as well as transmitting the first known payload to the receiving device based on the first transmission setting, wherein the first known payload is multiplexed with a second known payload from a second transmitting device of the group of transmitting devices, The first known payload is known to both the receiving device and the first sending device, and the second known payload is known to both the receiving device and the second sending device.

81. An apparatus according to claim 80, wherein the receiving device is a user equipment (UE), the first sending device is a serving sending receiving point (TRP), the second sending device is a second TRP, and the group of sending devices includes a multi-TRP group.

82. The apparatus of claim 81 , wherein: The group of receiving devices comprises a UE group; and The at least one processor is further configured to cause the apparatus to transmit the first known payload to each UE in the group of UEs.

83. The apparatus of claim 81 , wherein: The request further includes a request for a second known payload of the second TRP from the multi-TRP group and a second transmission setup for the first known payload and the second known payload; and The second transmission setting includes at least one of a multiplexing scheme, a rank of each TRP of the multi-TRP group, a precoding of each TRP of the multi-TRP group, a beam pair, a modulation and coding scheme (MCS), or a combination thereof.

84. The apparatus of claim 83, wherein the second transmission settings for the first known payload and the second known payload are the same as the first transmission settings for the first unknown payload and the second unknown payload from the second TRP.

85. The apparatus of claim 83, wherein the first known payload is different from the second known payload.

86. The apparatus of claim 83, wherein the multiplexing scheme comprises spatial division multiplexing (SDM), time division multiplexing (TDM), or frequency division multiplexing (FDM).

87. The apparatus of claim 83, wherein the at least one processor is further configured to cause the apparatus to send information comprising periodicity, time and frequency resources, and payload size of the first known payload and the second known payload, and the information is sent via radio resource control (RRC) signaling, a media access control layer (MAC) control element, or downlink control information (DCI).

88. The apparatus of claim 83, wherein the first known payload and the second known payload are transmitted on a physical downlink control channel (PDCCH), each of the first known payload and the second known payload being paired with a corresponding payload transmitted on a physical downlink shared channel (PDSCH).

89. The apparatus of claim 83, wherein the at least one processor is further configured to cause the apparatus to send information comprising periodicity, time and frequency resources, and payload size of the first known payload, and the information is received via radio resource control (RRC) signaling, a media access control layer (MAC) control element, or downlink control information (DCI).

90. The apparatus of claim 83, wherein the first known payload and the second known payload are sent according to a periodic schedule, a semi-persistent schedule, or aperiodically.

91. The apparatus of claim 83, wherein the at least one processor is further configured to cause the apparatus to scramble the first known payload and the second known payload based on a radio resource control (RRC) scrambling seed.

92. The apparatus of claim 91, wherein the first known payload and the second known payload are known to the UE, the serving TRP, and the second TRP based on the RRC scrambling seed.

93. The apparatus of claim 80, wherein the request is received via a physical uplink control channel (PUCCH) or a medium access control layer (MAC) element (CE).

94. The apparatus of claim 93, wherein the at least one processor is further configured to cause the apparatus to transmit the first known payload via a physical downlink shared channel (PDSCH).

95. The device of claim 80, wherein The first transmitting device is a first user equipment (UE), and the receiving device is a transmitting receiving point (TRP) of a multi-TRP group; The request includes a second transmission setting for the first known payload; The second transmission setting comprises at least one of a rank, a beam pair, a modulation and coding scheme (MCS), or a combination thereof; Sending the first known payload includes sending the first known payload to each TRP in the multi-TRP group.

96. The apparatus of claim 95, wherein: The group of transmitting devices includes a group of UEs; The request further includes a request for a second known payload from a second UE of the group of UEs; and The second transmission settings include transmission settings for the second known payload.

97. The apparatus of claim 95, wherein the second transmission setting for the first known payload is the same as the first transmission setting for the first known payload.

98. The apparatus of claim 95, wherein the at least one processor is further configured to cause the apparatus to transmit the first known payload via a physical uplink control channel (PUCCH).

99. The apparatus of claim 98, wherein the at least one processor is further configured to cause the apparatus to transmit a second known payload via a physical uplink shared channel (PUSCH), the first known payload being paired with the second known payload.

100. The apparatus of claim 95, wherein the request is received via a physical downlink control channel (PDCCH) or a medium access control layer (MAC) control element (CE).

101. A first transmitting device in a group of transmitting devices for wireless communication, comprising: means for transmitting a first unknown payload to a receiving device in a group of receiving devices based on first transmission settings of the first transmitting device; means for receiving, from the receiving device, a request for a first known payload for training an artificial neural network of the receiving device; as well as means for transmitting the first known payload to the receiving device based on the first transmission setting, wherein the first known payload is multiplexed with a second known payload from a second transmitting device of the group of transmitting devices, The first known payload is known to both the receiving device and the first sending device, and the second known payload is known to both the receiving device and the second sending device.

102. A first sending device according to claim 101, wherein the receiving device is a user equipment (UE), the first sending device is a serving sending receiving point (TRP), the second sending device is a second TRP, and the group of sending devices includes a multi-TRP group.

103. The first sending device according to claim 102, wherein: The group of receiving devices comprises a UE group; and The means for sending the first known payload includes means for sending the first known payload to each UE in the group of UEs.

104. The first sending device according to claim 102, wherein: The request further includes a request for a second known payload of the second TRP from the multi-TRP group and a second transmission setup for the first known payload and the second known payload; The second transmission setting includes at least one of a multiplexing scheme, a rank of each TRP of the multi-TRP group, a precoding of each TRP of the multi-TRP group, a beam pair, a modulation and coding scheme (MCS), or a combination thereof.

105. A first sending device according to claim 104, wherein the second transmission setting of the first known payload and the second known payload is the same as the first transmission setting of the first unknown payload and the second unknown payload from the second TRP.

106. The first sending device according to claim 101, wherein: The first transmitting device is a first user equipment (UE), and the receiving device is a transmitting receiving point (TRP) of a multi-TRP group; The request includes a second transmission setting for the first known payload; The second transmission setting comprises at least one of a rank, a beam pair, a modulation and coding scheme (MCS), or a combination thereof; and The means for sending the first known payload includes means for sending the first known payload to each TRP in the multi-TRP group.

107. The first sending device according to claim 106, wherein: The group of transmitting devices includes a group of UEs; The request further includes a request for a second known payload from a second UE of the group of UEs; and The second transmission settings include transmission settings for the second known payload.

108. The first sending device of claim 107, wherein the second transmission setting of the first known payload is the same as the first transmission setting of the first known payload.

109. A computer readable medium having instructions stored thereon, which, when executed by a processor, cause the processor to perform the method of any one of claims 59 to 79.

110. A computer program product comprising computer instructions which, when executed by a processor, cause the processor to perform the method of any one of claims 1 to 23 and 59 to 79.

Images