Joint learning of client-specific neural network parameter generation for wireless communication

Abstract

Various aspects of this disclosure generally relate to wireless communication. In some aspects, a client may use a conditioning network and determine a client-specific set of parameters based at least in part on observed environment vectors. The client may use a client-side autoencoder and determine a latent vector based at least in part on both the client-specific parameter set and a shared parameter set. The client may transmit the observed environment vector and the latent vector to a server. Numerous other aspects are provided.

Description

[0001] Cross-reference to related applications

[0002] This patent application claims U.S. Provisional Patent Application No. 62 / 706,465, filed August 18, 2020, entitled "FEDERATED LEARNING FOR CLIENT-SPECIFIC NEURAL NETWORK PARAMETER GENERATION FOR WIRELESS COMMUNICATION"; U.S. Provisional Patent Application No. 62 / 706,605, filed August 27, 2020, entitled "FEDERATED LEARNING FOR CLIENT-SPECIFIC NEURAL NETWORK PARAMETER GENERATION FOR WIRELESS COMMUNICATION"; and U.S. Provisional Patent Application No. 62 / 706,605, filed July 29, 2021, entitled "FEDERATED LEARNING FOR CLIENT-SPECIFIC NEURAL NETWORK PARAMETER GENERATION FOR WIRELESS COMMUNICATION". Priority is claimed in U.S. non-provisional patent application No. 17 / 444,028, entitled "Joint learning for client-specific neural network parameter generation for wireless communication," which is hereby expressly incorporated by reference.

[0003] introduction

[0004] Various aspects of this disclosure generally relate to wireless communication, and more particularly to techniques and apparatus for neural networks.

[0005] Wireless communication systems are widely deployed to provide a variety of telecommunications services such as telephone, video, data, messaging, and broadcasting. Typical wireless communication systems employ multiple access technologies that can support communication with multiple users by sharing available system resources (e.g., bandwidth, transmit power). 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 an enhancement set of the Universal Mobile Telecommunications System (UMTS) mobile standard issued by the 3rd Generation Partnership Project (3GPP).

[0006] A wireless network may include several base stations (BSs) capable of supporting communication between several user equipments (UEs). UEs may communicate with the BS via downlink and uplink. 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, the BS may be referred to as a B-node, gNB, access point (AP), radio headend, transmit / receive point (TRP), new radio (NR) BS, 5G B-node, etc.

[0007] The multiple access technologies mentioned above have been adopted in various telecommunications standards to provide a common protocol enabling different user equipment to communicate at the city, country, region, and even global levels. NR (which can also be referred to as 5G) is an enhancement set of the LTE mobile standard issued by 3GPP. NR is designed to better support mobile broadband Internet access by improving spectrum efficiency, reducing costs, improving service, utilizing new spectrum, and better integrating with other open standards that support beamforming, multiple-input multiple-output (MIMO) antenna technologies and carrier aggregation, using Orthogonal Frequency Division Multiplexing (OFDM) with a Cyclic Prefix (CP) on the downlink (DL) (CP-OFDM), and CP-OFDM and / or SC-FDM (e.g., also known as Discrete Fourier Transform Extended OFDM (DFT-s-OFDM)) on the uplink (UL). However, with the continued growth in demand for mobile broadband access, there is a need for further improvements to LTE and NR technologies. Preferably, these improvements should be applicable to other multiple access technologies and telecommunications standards that employ these technologies.

[0008] Overview

[0009] In some aspects, a wireless communication method is performed by a client and includes: determining a client-specific set of encoder parameters using a conditioning network, at least in part, based on an observed environment vector. The method may include: determining a latent vector using an autoencoder, also at least in part, based on the client-specific set of encoder parameters. The method may further include: transmitting the observed environment vector and the latent vector.

[0010] In some aspects, a wireless communication method performed by a server includes: receiving from a client an observed environment vector associated with one or more features, the one or more features being associated with the client's environment. The method may further include: receiving from the client a latent vector based at least in part on a client-specific set of encoder parameters. The method may include: determining an observed wireless communication vector based at least in part on the latent vector and the observed environment vector. The method may further include: performing a wireless communication action based at least in part on the determined observed wireless communication vector.

[0011] In some aspects, a client for wireless communication includes: a memory and one or more processors coupled to the memory. The memory and the one or more processors may be configured to: determine a client-specific set of encoder parameters using a conditioning network, at least in part, based on an observed environment vector. The memory and the one or more processors may be further configured to: determine a latent vector using an autoencoder, at least in part, based on the client-specific set of encoder parameters. The memory and the one or more processors may also be configured to: transmit the observed environment vector and the latent vector.

[0012] In some aspects, a server for wireless communication includes: a memory and one or more processors coupled to the memory. The memory and the one or more processors may be configured to: receive from a client an observed environment vector associated with one or more features, the one or more features being associated with the client's environment. The memory and the one or more processors may be further configured to: receive from the client a latent vector based at least in part on a client-specific set of encoder parameters. The memory and the one or more processors may also be configured to: determine an observed wireless communication vector based at least in part on the latent vector and the observed environment vector. The memory and the one or more processors may be configured to: perform a wireless communication action based at least in part on the determined observed wireless communication vector.

[0013] In some aspects, a non-transient computer-readable medium storing a set of instructions for wireless communication, the set of instructions comprising one or more instructions, which, when executed by one or more processors of a client, cause the client to: determine a client-specific set of encoder parameters using a conditioning network based at least in part on an observed environment vector. The one or more instructions may also cause the client to: determine a latent vector using an autoencoder and based at least in part on the client-specific set of encoder parameters. The one or more instructions may also cause the client to: transmit the observed environment vector and the latent vector.

[0014] In some aspects, a non-transient computer-readable medium storing a set of instructions for wireless communication, the set of instructions comprising one or more instructions, which, when executed by one or more processors of a server, cause the server to: receive from a client an observed environment vector associated with one or more features, the one or more features being associated with the client's environment. The one or more instructions may cause the server to: receive from the client a latent vector based at least in part on a client-specific set of encoder parameters. The one or more instructions may cause the server to: determine an observed wireless communication vector based at least in part on the latent vector and the observed environment vector. The one or more instructions may also perform a wireless communication action based at least in part on the determination of the observed wireless communication vector.

[0015] In some aspects, an apparatus for wireless communication includes: means for determining a client-specific set of encoder parameters using an conditioning network, at least in part, based on observed environment vectors. The apparatus may further include: means for determining a latent vector using an autoencoder, also at least in part, based on the client-specific set of encoder parameters. The apparatus may include: means for transmitting the observed environment vector and the latent vector.

[0016] In some aspects, an apparatus for wireless communication includes: means for receiving from a client an observed environment vector associated with one or more features, the one or more features being associated with the client's environment. The apparatus may further include: means for receiving from the client a latent vector based at least in part on a client-specific set of encoder parameters. The apparatus may further include: means for determining an observed wireless communication vector based at least in part on the latent vector and the observed environment vector. The apparatus may further include: means for performing a wireless communication action based at least in part on the determined observed wireless communication vector.

[0017] The aspects generally include, as described substantially with reference to the accompanying drawings and description and explained as such, methods, apparatus, systems, computer program products, non-transient computer-readable media, clients, servers, user equipment, base stations, wireless communication equipment, and / or processing systems.

[0018] The foregoing has broadly outlined the features and technical advantages of the examples according to this disclosure in an effort to facilitate a better understanding of the following detailed description. Additional features and advantages will be described thereafter. The disclosed concepts and specific examples can be readily used as the basis for modifying or designing other structures for implementing the same purposes as this disclosure. Such equivalent constructions do not depart from the scope of the appended claims. The characteristics of the concepts disclosed herein, in both their organization and manner of operation, and their associated advantages, will be better understood by considering the following description in conjunction with the accompanying drawings. Each drawing is provided for illustrative and descriptive purposes and does not define any limitation on the claims. Brief description of the attached diagram

[0020] To gain a more detailed understanding of the features described above in this disclosure, reference can be made to various aspects of the above brief overview, some of which are illustrated in the accompanying drawings. However, it should be noted that the drawings illustrate only certain typical aspects of this disclosure and should not be considered as limiting its scope, as other equivalent aspects are permissible in this description. Identical reference numerals in different drawings may identify the same or similar elements.

[0021] Figure 1 This is a diagram illustrating an example of a wireless network according to this disclosure.

[0022] Figure 2 This is a diagram illustrating an example of communication between a base station and a user equipment (UE) in a wireless network according to this disclosure.

[0023] Figure 3 This is a diagram illustrating an example of wireless communication using an autoencoder and an associated conditioning network according to this disclosure.

[0024] Figure 4 This is a diagram illustrating an example of a directional graphical model corresponding to an autoencoder and an associated conditioning network according to the present disclosure.

[0025] Figure 5 This is a diagram illustrating an example of an autoencoder and an associated conditioning network according to this disclosure.

[0026] Figure 6 This is a diagram illustrating an example of wireless communication using an autoencoder and an associated conditioning network according to this disclosure.

[0027] Figure 7 This is a diagram illustrating an example of joint learning of an autoencoder and an associated regulatory network according to the present disclosure.

[0028] Figure 8-13 This is a diagram illustrating an example of joint learning of an autoencoder and an associated regulatory network according to the present disclosure.

[0029] Figure 14 and 15 This is a diagram illustrating an example process associated with joint learning of an autoencoder and an associated regulatory network according to this disclosure.

[0030] Figure 16 and 17 This is an example of an apparatus for wireless communication according to this disclosure.

[0031] Figure 18 and 19 These are illustrations illustrating an example of the hardware implementation of an apparatus employing a processing system according to this disclosure.

[0032] Figure 20 and 21 These are illustrations illustrating examples of the implementation of code and circuitry systems for wireless communication according to this disclosure.

[0033] Detailed description

[0034] Clients operating within a network can measure reference signals and other parameters to report to a server. For example, a client may measure reference signals during beam management to achieve channel state feedback (CSF), measure the received power of reference signals from serving cells and / or neighboring cells, measure the signal strength of networks between radio access technologies (e.g., WiFi), measure sensor signals used to detect the location of one or more objects in the environment, and so on. However, reporting this information to the server can consume communication and / or network resources.

[0035] In some aspects described herein, a client (e.g., a UE, base station, transmit-receive point (TRP), network equipment, low Earth orbit (LEO) satellite, medium Earth orbit (MEO) satellite, geostationary orbit (GEO) satellite, highly elliptical orbit (HEO) satellite, etc.) may use one or more neural networks that can be trained to learn the dependence of the measured quality on individual parameters, to isolate the measured quality through the various layers (also referred to as "operations") of the one or more neural networks, and to compress the measurement in a manner that limits compression loss. The client may transmit the compressed measurement to a server (e.g., a TRP, another UE, base station, etc.).

[0036] The server can decode compressed measurements using one or more decompression and reconstruction operations associated with one or more neural networks. These decompression and reconstruction operations may be at least partially based on a feature set of the compressed dataset to produce reconstructed measurements. The server may then perform wireless communication actions based at least partially on the reconstructed measurements.

[0037] In some respects, clients and servers can use conditioning networks and autoencoders to compress and reconstruct information. In other cases, conditioning networks and autoencoders can be trained using joint learning. Joint learning is a machine learning technique that enables multiple clients to collaboratively learn a neural network model without the server collecting that data from these clients. Typically, joint learning techniques involve training a single global neural network model from data stored on multiple clients. For example, in a joint averaging algorithm, the server sends a neural network model to the clients. Each client uses its own data to train the received neural network model and sends the updated neural network model back to the server. The server averages the updated neural network models from those clients to obtain a new neural network model.

[0038] However, in some scenarios, some clients may operate in environments different from other clients (e.g., indoors / outdoors, stationary in a coffee shop / mobile on a highway, etc.). In some cases, different clients may suffer from different implementation aspects (e.g., different form factors, different RF impairments, etc.). As a result, from the perspective of physical layer link performance, finding a single neural network model that is well-suited for all devices may be difficult.

[0039] Based on various aspects of the techniques and apparatus described herein, partially federated learning and / or fully federated learning techniques can be used to provide and train autoencoders and conditioning networks. The autoencoder parameter set may include client-specific parameter sets and shared parameter sets. The conditioning network computes client-specific parameters of the autoencoder applicable to the respective client. In this way, various aspects can contribute to better physical layer link performance.

[0040] Based on aspects of the techniques and apparatus described herein, partially joint learning and / or fully joint learning techniques can be used to train conditioning networks and autoencoders. In this way, aspects can contribute to better physical layer link performance. In some aspects, the client can use the conditioning network and determine a client-specific parameter set, etc., based at least in part on observed environment vectors. The client can use a client-side autoencoder and determine a latent vector based at least in part on the client-specific parameter set and a shared parameter set. The client can transmit the observed environment vector and the latent vector to the server. The server can use the conditioning network to determine client-specific parameters, at least in part on the observed environment vector, and use a decoder corresponding to the autoencoder to recover the observed wireless communication vector, at least in part on the client-specific parameters, the shared parameters, and the latent vector. Aspects of the techniques described herein can be used for any number of cross-node machine learning challenges, including facilitating channel state feedback, facilitating client localization, learning modulation and / or waveforms for wireless communication, etc.

[0041] The various aspects of this disclosure are described more fully below with reference to the accompanying drawings. However, this disclosure may be implemented in many different forms and should not be construed as being limited to any specific structure or function given throughout this disclosure. Rather, these aspects are provided so that this disclosure will be thorough and complete, and will fully convey the scope of this disclosure to those skilled in the art. Based on the teachings herein, those skilled in the art will appreciate that the scope of this disclosure is intended to cover any aspect of this disclosure disclosed herein, whether implemented independently of or in combination with any other aspect of this disclosure. For example, any number of aspects set forth herein may be used to implement an apparatus or practice. Furthermore, the scope of this disclosure is intended to cover such apparatuses or methods practiced using additional structures, functionalities, or structures and functionalities that complement or supplement the various aspects of this disclosure set forth herein. It should be understood that any aspect of this disclosure disclosed herein may be implemented by one or more elements of the claims.

[0042] Several aspects of a telecommunications system will now be described with reference to various devices and techniques. These devices and techniques will be described in the following detailed description and explained 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 such elements are implemented as hardware or software depends on the specific application and the design constraints imposed on the overall system.

[0043] It should be noted that although the aspects herein may be described using terms commonly associated with 5G or NR radio access technology (RAT), the aspects of this disclosure may be applied to other RATs, such as 3G RAT, 4G RAT, and / or RATs after 5G (e.g., 6G).

[0044] Figure 1 This is a diagram illustrating an example of a wireless network 100 according to this disclosure. The wireless network 100 may be a 5G (NR) network and / or an LTE network, etc., or may include its elements. The wireless network 100 may include several base stations 110 (shown as BS110a, BS 110b, BS 110c, and BS 110d) and other network entities. A base station (BS) is an entity that communicates with a user equipment (UE) and may also be referred to as an NR BS, B-node, gNB, 5G B-node (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 the BS subsystem serving that coverage area, depending on the context in which the term is used.

[0045] A BS can provide communication coverage for macrocells, picocells, femtocells, and / or another type of cell. Macrocells can cover a relatively large geographic area (e.g., a radius of several kilometers) and allow unrestricted access by UEs with a service subscription. Picocells can cover a relatively small geographic area and allow unrestricted access by UEs with a service subscription. Femtocells can cover a relatively small geographic area (e.g., a residential area) and allow restricted access by UEs associated with that femtocell (e.g., UEs in a Closed Subscriber Group (CSG)). A BS used for macrocells may be referred to as a macro BS. A BS used for picocells may be referred to as a pico BS. A BS used for femtocells may be referred to as a femto BS or a home BS. Figure 1 In the example shown, BS 110a can be a macro BS for macro cell 102a, BS 110b can be a pico BS for pico cell 102b, and BS 110c can be a femto BS for femto cell 102c. A BS can support one or more (e.g., three) cells. The terms “eNB,” “base station,” “NR BS,” “gNB,” “TRP,” “AP,” “B node,” “5G NB,” and “cell” are used interchangeably herein.

[0046] In some examples, the cell may not be stationary, and the geographical area of ​​the cell may move depending on the location of the mobile BS. In some examples, BSs may interconnect with each other and / or interconnect to 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 or virtual networks, using any suitable transport network).

[0047] The wireless network 100 may also include a relay station. A relay station is an entity capable of receiving data transmissions from an upstream station (e.g., a BS or a UE) and transmitting those data transmissions to a downstream station (e.g., a UE or a BS). A relay station may also be a UE capable of relaying transmissions for other UEs. Figure 1 In the example shown, relay BS 110d can communicate with macro BS 110a and UE 120d to facilitate communication between BS 110a and UE 120d. A relay BS can also be referred to as a relay station, relay base station, relay, etc.

[0048] In some aspects, wireless network 100 may include one or more non-terrestrial network (NTN) deployments, wherein non-terrestrial wireless communication devices may include UEs (which are interchangeably referred to herein as “non-terrestrial UEs”), BSs (which are interchangeably referred to herein as “non-terrestrial BSs” and “non-terrestrial base stations”), relay stations (which are interchangeably referred to herein as “non-terrestrial relay stations”), and so on. As used herein, “NTN” may refer to a network to which access is facilitated by non-terrestrial UEs, non-terrestrial BSs, non-terrestrial relay stations, and so on.

[0049] Wireless Network 100 may include any number of non-terrestrial wireless communication devices. Non-terrestrial wireless communication devices may include satellites, unmanned aerial vehicle (UAS) platforms, etc. Satellites may include low Earth orbit (LEO) satellites, medium Earth orbit (MEO) satellites, geostationary orbit (GEO) satellites, highly elliptical orbit (HEO) satellites, etc. UAS platforms may include high-altitude platform stations (HAPS) and may include balloons, airships, aircraft, etc. Non-terrestrial wireless communication devices may be part of an NTN separate from Wireless Network 100. Alternatively, the NTN may be part of Wireless Network 100. Satellites may use satellite communications to communicate directly and / or indirectly with other entities in Wireless Network 100. Other entities may include UEs (e.g., terrestrial UEs and / or non-terrestrial UEs), one or more other satellites in one or more NTN deployments, other types of BSs (e.g., stationary or terrestrial BSs), relay stations, one or more components and / or devices included in the core network of Wireless Network 100, etc.

[0050] Wireless network 100 can be a heterogeneous network comprising different types of Base Stations (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 effects on interference in wireless network 100. For example, macro BSs may have high transmit power levels (e.g., 5 to 40 watts), while pico BSs, femto BSs, and relay BSs may have lower transmit power levels (e.g., 0.1 to 2 watts).

[0051] Network controller 130 can be coupled to a set of BSs and can provide coordination and control over these BSs. Network controller 130 can communicate with each BS via backhaul. These BSs can also communicate with each other directly or indirectly, for example, via wireless or wired backhaul.

[0052] UE 120 (e.g., 120a, 120b, 120c) may be distributed throughout the wireless network 100, and each UE may be stationary or mobile. UE may also be referred to as an access terminal, terminal, mobile station, subscriber unit, station, etc. UE may be a cellular phone (e.g., a smartphone), personal digital assistant (PDA), wireless modem, wireless communication device, handheld device, laptop computer, cordless phone, wireless local loop (WLL) station, tablet, camera, gaming device, netbook, smartbook, ultrabook, medical device or equipment, biometric sensor / device, wearable device (smartwatch, smart clothing, smart glasses, smart wristband, smart jewelry (e.g., smart ring, smart bracelet)), entertainment device (e.g., music or video device, or satellite radio), vehicle component or sensor, smart meter / sensor, industrial manufacturing equipment, GPS device, or any other suitable device configured to communicate via wireless or wired media.

[0053] Some UEs may be considered machine-type communication (MTC) devices, or evolved or enhanced machine-type communication (eMTC) UEs. MTC and eMTC UEs include, for example, robots, drones, remote devices, sensors, instruments, monitors, and / or location tags that can communicate with a base station, another device (e.g., a remote device), or some other entity. Wireless nodes may provide connectivity to or to a network (e.g., a wide area network, such as the Internet or a cellular network) via wired or wireless communication links, for example. Some UEs may be considered Internet of Things (IoT) devices, and / or may be implemented as NB-IoT (Narrowband Internet of Things) devices. Some UEs may be considered customer premises equipment (CPE). UE 120 may be included within a housing that houses components of UE 120, such as processor components and / or memory components. In some aspects, the processor components and memory components may be coupled together. For example, the processor components (e.g., one or more processors) and memory components (e.g., memory) may be operatively coupled, communicatively coupled, electronically coupled, and / or electrically coupled.

[0054] Generally, any number of wireless networks can be deployed in a given geographical area. Each wireless network can support a specific RAT and can operate on one or more frequencies. A RAT can also be referred to as a radio technology, air interface, etc. A frequency can also be referred to as a carrier, frequency channel, etc. Each frequency can support a single RAT in a given geographical area to avoid interference between wireless networks using different RATs. In some cases, NR or 5G RAT networks can be deployed.

[0055] 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 example, 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 or vehicle-to-infrastructure (V2I) protocols), and / or mesh networks. In some aspects, UEs 120 may perform scheduling operations, resource selection operations, and / or other operations described elsewhere herein as performed by base station 110.

[0056] Devices in the wireless network 100 can communicate using the electromagnetic spectrum, which can be subdivided into various categories, bands, channels, etc., based on frequency or wavelength. For example, devices in the wireless network 100 can communicate using an operating band with a first frequency range (FR1) and / or an operating band with a second frequency range (FR2), where the first frequency range (FR1) spans from 410 MHz to 7.125 GHz and the second frequency range (FR2) spans from 24.25 GHz to 52.6 GHz. The frequencies between FR1 and FR2 are sometimes referred to as intermediate frequency bands. Although a portion of FR1 is greater than 6 GHz, FR1 is often referred to as the "sub-6 GHz band." Similarly, although different from the extremely high frequency (EHF) band (30 GHz–300 GHz) designated as the "millimeter wave" band by the International Telecommunication Union (ITU), FR2 is often referred to as the "millimeter wave" band. Therefore, unless otherwise stated, it should be understood that, if used herein, the term "sub-6GHz," etc., can broadly refer to frequencies less than 6GHz, frequencies within FR1, and / or intermediate frequency band frequencies (e.g., greater than 7.125GHz). Similarly, unless otherwise stated, it should be understood that, if used herein, the term "millimeter wave," etc., can broadly refer to frequencies within the EHF band, frequencies within FR2, and / or intermediate frequency band frequencies (e.g., less than 24.25GHz). It is conceivable that the frequencies included in FR1 and FR2 can be modified, and the techniques described herein are applicable to those modified frequency ranges.

[0057] like Figure 1 As shown, UE 120 may include a first communication manager 140. As described in more detail elsewhere herein, the first communication manager 140 may use an adjustment network to determine a client-specific encoder parameter set based at least in part on an observed environment vector; use an autoencoder and determine a latent vector based at least in part on the client-specific encoder parameter set; and transmit the observed environment vector and the latent vector. Additionally or alternatively, the first communication manager 140 may perform one or more other operations described herein.

[0058] In some aspects, base station 110 may include a second communication manager 150. As described in more detail elsewhere herein, the second communication manager 150 may receive from a client an observed environment vector associated with one or more features, the one or more features being associated with the client's environment; receive from the client a latent vector based at least in part on a client-specific set of encoder parameters; determine an observed wireless communication vector based at least in part on the latent vector and the observed environment vector; and perform a wireless communication action based at least in part on the determined observed wireless communication vector. Additionally or alternatively, the second communication manager 150 may perform one or more other operations described herein.

[0059] As indicated above, Figure 1 This is provided merely as an example. Other examples may differ from those provided. Figure 1 The example described.

[0060] Figure 2 This is a diagram illustrating an example 200 of communication between a base station 110 and a UE 120 in a wireless network 100 according to this disclosure. The base station 110 may be equipped with T antennas 234a to 234t, while the UE 120 may be equipped with R antennas 252a to 252r, wherein generally T≥1 and R≥1.

[0061] At base station 110, transmit processor 220 can receive data destined for one or more UEs from data source 212, select one or more modulation and coding schemes (MCS) for each UE based at least in part on channel quality indicators (CQI) received from each UE, process (e.g., encode and modulate) the data for each UE based at least in part on the MCS selected for each UE, and provide data symbols for all UEs. Transmit processor 220 can also process system information (e.g., semi-static resource allocation information (SRPI)) and control information (e.g., CQI requests, grants, and / or upper-layer signaling), and provide overhead symbols and control symbols. Transmit processor 220 can also generate reference symbols for reference signals (e.g., cell-specific reference signals (CRS) or demodulation reference signals (DMRS)) and synchronization signals (e.g., primary synchronization signal (PSS) or secondary synchronization signal (SSS)). The transmit (TX) multiple-input multiple-output (MIMO) processor 230 can perform spatial processing (e.g., precoding) on ​​data symbols, control symbols, overhead symbols, and / or reference symbols, where applicable, and can provide T output symbol streams to T modulators (MODs) 232a to 232t. Each modulator 232 can process its respective output symbol stream (e.g., for OFDM) to obtain an output sample stream. Each modulator 232 can further process (e.g., convert to analog, amplify, filter, and up-convert) the output sample stream to obtain a downlink signal. The T downlink signals from modulators 232a to 232t can be transmitted via T antennas 234a to 234t, respectively.

[0062] At UE 120, antennas 252a to 252r can receive downlink signals from base station 110 and / or other base stations and can provide the received signals to demodulators (DEMODs) 254a to 254r respectively. Each demodulator 254 can condition (e.g., filter, amplify, downconvert, and digitize) the received signal to obtain an input sample. Each demodulator 254 can further process the input sample (e.g., for OFDM) to obtain received symbols. MIMO detector 256 can obtain the received symbols from all R demodulators 254a to 254r, perform MIMO detection on these received symbols where applicable, and provide detected symbols. Receiver processor 258 can process (e.g., demodulate and decode) these detected symbols, provide the decoded data for UE 120 to data sink 260, and provide the decoded control information and system information to controller / processor 280. The term "controller / processor" can refer to one or more controllers, one or more processors, or a combination thereof. The channel processor can determine the Reference Signal Received Power (RSRP) parameter, the Received Signal Strength Indicator (RSSI) parameter, the Reference Signal Received Quality (RSRQ) parameter, and / or the CQI parameter. In some respects, one or more components of the UE 120 may be included in the housing.

[0063] Network controller 130 may include communication unit 294, controller / processor 290, and memory 292. Network controller 130 may include one or more devices, such as those in a core network. Network controller 130 may communicate with base station 110 via communication unit 294.

[0064] Antennas (e.g., antennas 234a to 234t and / or antennas 252a to 252r) may include, or be included therein, one or more antenna panels, antenna groups, antenna element sets, and / or antenna arrays. Antenna panels, antenna groups, antenna element sets, and / or antenna arrays may include one or more antenna elements. Antenna panels, antenna groups, antenna element sets, and / or antenna arrays may include coplanar antenna element sets and / or non-coplanar antenna element sets. Antenna panels, antenna groups, antenna element sets, and / or antenna arrays may include antenna elements within a single housing and / or multiple antenna elements within housings. Antenna panels, antenna groups, antenna element sets, and / or antenna arrays may include elements coupled to one or more transmission and / or reception components (such as...). Figure 2 One or more antenna elements (one or more components).

[0065] On the uplink, at UE 120, transmit processor 264 can receive and process data from data source 262 and control information from controller / processor 280 (e.g., reports including RSRP, RSSI, RSRQ, and / or CQI). Transmit processor 264 can also generate reference symbols for one or more reference signals. Symbols from transmit processor 264 may be pre-encoded by TX MIMO processor 266 where applicable, further processed by modulators 254a to 254r (e.g., for DFT-s-OFDM or CP-OFDM), and transmitted to base station 110. In some aspects, modulators and demodulators (e.g., MOD / DEMOD 254) of UE 120 may be included in the modem of UE 120. In some aspects, UE 120 includes a transceiver. The transceiver may include any combination of antennas 252, modulators and / or demodulators 254, MIMO detectors 256, receiver processors 258, transmitter processors 264, and / or TX MIMO processors 266. The transceiver may be used by a processor (e.g., controller / processor 280) and memory 282 to perform aspects of any of the methods described herein.

[0066] At base station 110, uplink signals from UE 120 and other UEs can be received by antenna 234, processed by demodulator 232, detected by MIMO detector 236 where applicable, and further processed by receiver processor 238 to obtain decoded data and control information transmitted by UE 120. Receiver processor 238 can provide the decoded data to data sink 239 and the decoded control information to controller / processor 240. Base station 110 may include communication unit 244 and communicate with network controller 130 via communication unit 244. Base station 110 may include scheduler 246 to schedule UE 120 for downlink and / or uplink communications. In some aspects, the modulator and demodulator (e.g., MOD / DEMOD 232) of base station 110 may be included in the modem of base station 110. In some aspects, base station 110 includes a transceiver. The transceiver may include any combination of antennas 234, modulators and / or demodulators 232, MIMO detectors 236, receiver processors 238, transmitter processors 220, and / or TX MIMO processors 230. The transceiver may be used by a processor (e.g., controller / processor 240) and memory 242 to perform aspects of any of the methods described herein.

[0067] The controller / processor 240 of base station 110, the controller / processor 280 of UE 120, and / or Figure 2Any other component may perform one or more techniques associated with aspects of the process, which is related to joint learning for client-specific neural network parameter generation for wireless communication, as described elsewhere in this document in more detail. For example, the controller / processor 240 of base station 110, the controller / processor 280 of UE 120, and / or Figure 2 Any other component may execute or direct, for example Figure 8 The process 800 Figure 9 The process 900 Figure 10 Process 1000 Figure 11 Process 1100 Figure 12 Process 1200 Figure 13 Process 1300 Figure 14 Process 1400 Figure 15 The operation of process 1500 and / or other processes as described herein. Memory 242 and 282 may store data and program code for base station 110 and UE 120, respectively. In some aspects, memory 242 and / or memory 282 may include: a non-transitory computer-readable medium storing one or more instructions (e.g., code and / or program code) for wireless communication. For example, when executed by one or more processors of base station 110 and / or UE 120 (e.g., direct execution, or execution after compilation, transformation, and / or interpretation), the one or more processors, UE 120, and / or base station 110 may perform or direct, for example... Figure 8 The process 800 Figure 9 The process 900 Figure 10 Process 1000 Figure 11 Process 1100 Figure 12 Process 1200 Figure 13 Process 1300 Figure 14 Process 1400 Figure 15 The operation of process 1500, and / or other processes described herein. In some aspects, the execution instructions may include run instructions, translate instructions, compile instructions, and / or interpret instructions, etc.

[0068] In some aspects, the client (e.g., UE 120) may include means for determining a client-specific set of encoder parameters using an conditioning network based at least in part on observed environment vectors, means for determining a latent vector using an autoencoder based at least in part on the client-specific set of encoder parameters, means for transmitting the observed environment vector and the latent vector, etc. Additionally or alternatively, the client may include means for performing one or more other operations described herein. In some aspects, such means may include a communication manager 140. Additionally or alternatively, such means may include a combination of... Figure 2 One or more other components of the described UE 120, such as controller / processor 280, transmit processor 264, TX MIMO processor 266, MOD 254, antenna 252, DEMOD 254, MIMO detector 256, receive processor 258, etc.

[0069] In some aspects, the server (e.g., base station 110) may include means for receiving from a client an observed environment vector associated with one or more features, the one or more features being associated with the client's environment; means for receiving from the client a latent vector based at least in part on a client-specific set of encoder parameters; means for determining an observed wireless communication vector based at least in part on the latent vector and the observed environment vector; means for performing a wireless communication action based at least in part on the determined observed wireless communication vector; and so on. Additionally or alternatively, the server may include means for performing one or more other operations described herein. In some aspects, such means may include a communication manager 150. In some aspects, such means may include a combination of... Figure 2 One or more other components of the described base station 110, such as antenna 234, DEMOD 232, MIMO detector 236, receiver processor 238, controller / processor 240, transmitter processor 220, TX MIMO processor 230, MOD 232, antenna 234, etc.

[0070] although Figure 2 The boxes in the diagram are interpreted as different components, but the functions described above with respect to these boxes can be implemented by a single hardware component, software component, or combination of components. For example, the functions described with respect to transmit processor 264, receive processor 258, and / or TX MIMO processor 266 can be performed by controller / processor 280 or under the control of controller / processor 280.

[0071] As indicated above, Figure 2 This is provided merely as an example. Other examples may differ from those provided. Figure 2 The example described.

[0072] As indicated above, a client operating within a network can measure reference signals, etc., to report to a server. However, reporting this information to the server can consume communication and / or network resources. In some aspects described herein, the client can use one or more neural networks, which can be trained to learn the dependence of measured quality on individual parameters, isolate the measured quality through the layers of the one or more neural networks, and compress the measurement in a manner that limits compression loss. The client can then transmit the compressed measurement to the server.

[0073] The server can decode compressed measurements using one or more decompression and reconstruction operations associated with one or more neural networks. These decompression and reconstruction operations may be at least partially based on a feature set of the compressed dataset to produce reconstructed measurements. The server may then perform wireless communication actions based at least partially on the reconstructed measurements.

[0074] In some respects, clients and servers can use conditioning networks and autoencoders to compress and reconstruct information. In other cases, conditioning networks and autoencoders can be trained using joint learning. Typically, joint learning techniques involve a single global neural network model trained from data stored on multiple clients. For example, in a joint averaging algorithm, the server sends a neural network model to the clients. Each client uses its own data to train the received neural network model and sends the updated model back to the server. The server then averages the updated neural network models from those clients to obtain a new neural network model.

[0075] However, in some cases, some clients may operate in environments different from other clients. In other cases, different clients may suffer from different implementation aspects. As a result, from the perspective of physical layer link performance, finding a single neural network model that is well-suited for all devices can be difficult. Based on the aspects of the techniques and devices described in this paper, joint learning techniques can be used to provide and train an autoencoder and a conditioning network that compute client-specific parameters for the autoencoder appropriate for the respective client. In this way, aspects can contribute to better physical layer link performance.

[0076] Based on aspects of the techniques and apparatus described herein, joint learning techniques can be used to train conditioning networks and autoencoders. In this way, aspects can contribute to better physical layer link performance. In some aspects, the client can use the conditioning network and determine a client-specific parameter set, etc., based at least in part on observed environment vectors. The client can use a client-side autoencoder and determine a latent vector, based at least in part on the client-specific parameter set and a shared parameter set. The client can transmit the observed environment vector and the latent vector to the server. The server can use the conditioning network to determine client-specific parameters, at least in part on the observed environment vector, and use a decoder corresponding to the autoencoder to recover the observed wireless communication vector, based at least in part on the client-specific parameters, the shared parameters, and the latent vector. Aspects of the techniques described herein can be used for any number of cross-node machine learning challenges, including facilitating channel state feedback, facilitating client localization, learning modulation and / or waveforms for wireless communication, etc.

[0077] Figure 3 This is a diagram illustrating an example 300 of wireless communication using an autoencoder and an associated conditioning network according to this disclosure. As shown, client 302 and server UE 304 can communicate with each other. In some aspects, client 302 and server 304 can communicate via a wireless network (e.g., Figure 1 The wireless networks 100 shown communicate with each other. In some respects, more than one client 302 and / or more than one server 304 can communicate with each other.

[0078] In some aspects, client 302 may be, similar to, include, or be included in a wireless communication device (e.g., UE 120, base station 110, Integrated Access and Backhaul (IAB) node, etc.). In some aspects, server 304 may be, similar to, include, or be included in a wireless communication device (e.g., UE 120, base station 110, IAB node, etc.). For example, in some aspects, client 302 may be UE 120 and server 304 may be a base station, and client 302 and server 304 may communicate via an access link. In some aspects, client 302 and server 304 may be UE 120 communicating via a sidelink.

[0079] As shown, client 302 may include first communication manager 306 (e.g., Figure 1 The first communication manager 140 shown can be configured to perform one or more wireless communication tasks using a client conditioning network (shown as "conditioning network") 308 and a client autoencoder 310. In some aspects, the client conditioning network 308 may be, similar to, include, or be included in. Figure 5 The conditioning network 510 shown and described below. In some aspects, the client autoencoder 310 may be, similar to, include, or be included in. Figure 5 As shown and in the autoencoder 520 described below.

[0080] The client-side conditioning network is a conditioning network implemented at the client-side, and the server-side conditioning network is a conditioning network implemented at the server-side. The client-side conditioning network may be referred to herein as a "conditioning network," as it is clear from the context that the conditioning network is implemented at the client-side. The server-side conditioning network may be referred to herein as a "conditioning network," as it is clear from the context that the conditioning network is implemented at the server-side. The client-side autoencoder is an autoencoder implemented at the client-side, and the server-side autoencoder is an autoencoder implemented at the server-side. The client-side autoencoder may be referred to herein as an "autoencoder," as it is clear from the context that the autoencoder is implemented at the client-side. The server-side autoencoder may be referred to herein as an "autoencoder," as it is clear from the context that the autoencoder is implemented at the server-side.

[0081] As shown, the client conditioning network 308 can be configured to receive an observed environment vector f as input and provide a conditioning vector c as output. The conditioning vector c may include a client-specific encoder parameter set, a client-specific decoder parameter set, and so on. In some aspects, the client conditioning network 308 may include: a first subnetwork configured to determine a client-specific encoder parameter set, a second subnetwork configured to determine a client-specific decoder parameter set, and so on.

[0082] As shown, the client-side autoencoder 310 may include an encoder 312 configured to receive an observed wireless communication vector x as input and provide a latent vector h as output. This encoder may include at least one encoder layer configured to use a client-specific set of encoder parameters determined by the client conditioning network 308. The client 302 can determine the latent vector by using this at least one encoder layer to map the observed wireless communication vector x to the latent vector h. The at least one encoder layer may include: a first set of layers using a first subset of the client-specific encoder parameter set, a second set of layers using a first subset of a shared encoder parameter set, a third set of layers using a second subset of the client-specific encoder parameter set and a second subset of the shared encoder parameter set, and so on.

[0083] The client-side autoencoder 310 may further include a decoder 314 configured to receive a latent vector h as input and provide an observed wireless communication vector x as output. The decoder may include at least one decoder layer configured to use a client-specific set of decoder parameters. The client 302 can determine the observed wireless communication vector x by using the at least one decoder layer to map the latent vector h to the observed wireless communication vector x. The client 302 can use the decoder parameter set to map the latent vector h to the observed wireless communication vector x. The decoder parameter set may include a client-specific set of decoder parameters and a shared set of decoder parameters. The at least one decoder layer may include: a first set of layers using a first subset of the client-specific set of decoder parameters, a second set of layers using a first subset of the shared set of decoder parameters, a third set of layers using a second subset of the client-specific set of decoder parameters and a second subset of the shared set of decoder parameters, and so on.

[0084] like Figure 3 As shown, server 304 may include a second communication manager 316 (e.g., second communication manager 150) configured to perform one or more wireless communication tasks using server conditioning network (shown as "conditioning network") 318. For example, in some aspects, server conditioning network 318 may correspond to client conditioning network 308. In some aspects, server conditioning network 318 may be a copy of client conditioning network 308. In some aspects, server conditioning network 318 may be, similar to, include, or be included in... Figure 5 As shown and in the regulation network 510 described below.

[0085] As shown, the server conditioning network 318 can be configured to receive an observed environment vector f as input and provide a conditioning vector c as output. The conditioning vector c may include a client-specific encoder parameter set, a client-specific decoder parameter set, and so on. In some aspects, the server conditioning network 318 may include: a first subnetwork configured to determine a client-specific encoder parameter set, a second subnetwork configured to determine a client-specific decoder parameter set, and so on.

[0086] In some aspects, server 304 may include server autoencoder 320. Server autoencoder 320 may include encoder 322 configured to receive an observed wireless communication vector x as input and provide a latent vector h as output. Server autoencoder 320 may also include decoder 324 configured to receive the latent vector h as input and provide the observed wireless communication vector x as output. Encoder 322 may include at least one encoder layer configured to use a client-specific encoder parameter set. Decoder 324 may include at least one decoder layer configured to use a client-specific decoder parameter set.

[0087] Decoder 324 is configured to determine the observed wireless communication vector x by mapping the latent vector h to the observed wireless communication vector x using the at least one decoder layer. Server 304 may use a decoder parameter set to map the latent vector h to the observed wireless communication vector x. The decoder parameter set may include a client-specific decoder parameter set and a shared decoder parameter set. The at least one decoder layer may include: a first layer set that uses a first subset of the client-specific decoder parameter set, a second layer set that uses a first subset of the shared decoder parameter set, a third layer set that uses a second subset of the client-specific decoder parameter set and a second subset of the shared decoder parameter set, and so on.

[0088] like Figure 3 As shown, client 302 may include a transceiver (shown as "Tx / Rx") 326 that facilitates wireless communication with transceiver 328 of server 304. As indicated by reference numeral 330, server 304 may, for example, use transceiver 328 to transmit wireless communication to client 302. In some aspects, the wireless communication may include reference signals (such as channel state information reference signals (CSI-RS)). Transceiver 326 of client 302 may receive the wireless communication. First communication manager 306 may determine the observed wireless communication vector x based at least in part on the wireless communication. For example, in the aspect where the wireless communication is CSI-RS, the observed wireless communication vector x may include channel state information (CSI).

[0089] As shown, the first communication manager 306 can obtain the observed environmental vector f and provide it to the client conditioning network 308. The first communication manager 306 can obtain the observed environmental vector f from memory, from one or more sensors, etc. As shown, the client conditioning network can determine the conditioning vector c based at least in part on the observed environmental vector f. The first communication manager 306 can load client-specific autoencoder parameters from the conditioning vector c into the client autoencoder 310. The first communication manager 306 can provide the observed wireless communication vector x as input to the encoder 312 of the client autoencoder 310. The encoder 312 of the client autoencoder 310 can determine the latent vector h based at least in part on the observed wireless communication vector x.

[0090] As indicated by reference numeral 332, the first communication manager 306 can provide the observed environment vector f and the latent vector h to the transceiver 326 for transmission. As indicated by reference numeral 334, the transceiver 326 can transmit and the transceiver 328 of the server 304 can receive the observed environment vector f and the latent vector h. As shown, the second communication manager 316 of the server 304 can provide the observed environment vector f as input to the server conditioning network 318, which can determine the conditioning vector c. The second communication manager 316 can load client-specific decoder parameters from the conditioning vector c into one or more layers of the decoder 324 of the server autoencoder 320.

[0091] The second communication manager 316 can provide the latent vector h as input to the decoder 324 of the server autoencoder 320. The decoder 324 can determine (e.g., reconstruct) the observed wireless communication vector x based at least in part on the latent vector h. In some aspects, the server 304 can perform wireless communication actions based at least in part on the observed wireless communication vector x. For example, where the observed wireless communication vector x includes CSI, the second communication manager 316 of the server 304 can use CSI for communication grouping, beamforming, etc.

[0092] As indicated above, Figure 3 This is provided merely as an example. Other examples may differ from those provided. Figure 3 The example described.

[0093] Figure 4 Figure 400 illustrates an example of an oriented graphical model corresponding to an autoencoder and an associated conditioning network according to this disclosure. In some aspects, for example, the oriented graphical model may correspond to... Figure 3 The client-side conditioning network 308 and client-side autoencoder 310 shown are... Figure 3The server conditioning network 318 and server self-encoder 320 shown are among the examples.

[0094] In some respects, the data distribution at client s can be determined by... Figure 4 The directional graphical model shown is used to represent this. This model provides an example of the relationship between the client s; the observed wireless communication vector x; the latent vector h; the observed environment vector f; and the conditioning vector c.

[0095] In some aspects, the observed wireless communication vector x and the latent vector h can be associated with a wireless communication task. The observed wireless communication vector x may include an array of observed values ​​associated with one or more measurements obtained in conjunction with the wireless communication. In some aspects, the wireless communication task may include: determining channel state feedback (CSF), determining location information associated with a client, determining modulation associated with the wireless communication, determining waveforms associated with the wireless communication, and so on. The latent vector h is the output of an autoencoder that takes the observed wireless communication vector x as input. The latent vector h may include an array of hidden values ​​associated with one or more aspects of the observed communication vector x.

[0096] For example, in some aspects, an autoencoder can be used to compress the CSF to feed the CSI back to the server. In some aspects, the observed wireless communication vector x may include the propagation channel estimated by the client (e.g., UE 120) at least in part based on the received CSI-RS. The latent vector h may include the compressed CSF to be fed back to the server (e.g., base station 110).

[0097] In some aspects, the observed environment vector f may include one or more variables that can be observed to facilitate the learning of the client s's environment. In some aspects, the conditioning vector c may be determined at least in part based on the observed environment vector f, such that the conditioning vector c can be used to customize the output of the autoencoder's encoder and / or decoder with respect to the environment associated with the client.

[0098] In some respects, the observed environment vector f may include the client (e.g., Figure 3The client 302 shown can obtain any number of different types of information about its environment. This information may include information about the client itself (e.g., device information, configuration information, capability information, etc.), information about the state associated with the client (e.g., operating state, power state, activation state, etc.), information about the client's location (e.g., location information, orientation information, geographic information, motion information, etc.), information about the client's surrounding environment (e.g., weather information, information about obstacles to wireless signals around the client, information about materials near the client, etc.), and so on. The observed environment vector f can be formed by concatenating one or more information indicators (such as those listed above).

[0099] In some aspects, for example, the observed environment vector f may include a client identifier (ID), client antenna configuration, large-scale channel characteristics, CSI-RS configuration, images obtained by an imaging device, a portion of the estimated propagation channel, and so on. In some aspects, for example, large-scale channel characteristics may indicate channel-associated delay spread, channel-associated power delay profile, channel-associated Doppler measurements, channel-associated Doppler spectrum, channel-associated signal-to-noise ratio (SNR), channel-associated signal-to-interference-plus-noise ratio (SINR), reference signal received power (RSRP), received signal strength indicator (RSSI), and so on.

[0100] In some respects, based on the aspects discussed in this article, Figure 4 The graphical model shown can be used to derive probabilistic expressions related to the use of a conditioning network associated with an autoencoder. For example, in some aspects, the graphical model can indicate the following probability density function:

[0101] p(x,h,c,f|s)=p(x|h,c)·p(h|c)·p(c|f)·p(f|s),

[0102] Where p is the probability density function, x is the observed wireless communication vector, f is the observed environment vector, h is the latent vector, c is the conditioning vector at least partially based on f, and s is the client. In some respects, the probability density function indicated by the oriented graphical model can be used to configure the training of the autoencoder.

[0103] As indicated above, Figure 4 This is provided merely as an example. Other examples may differ from those provided. Figure 4 The example described.

[0104] Figure 5This is a diagram illustrating an example 500 of an autoencoder and an associated conditioning network according to this disclosure. Aspects of example 500 can be configured by a client (e.g., Figure 3 The client 302 shown), server (e.g., Figure 3 This is achieved through methods such as the server 304 shown in the diagram. Figure 5 The autoencoder and associated conditioning network in the code can be used to achieve... Figure 4 The directional graphical model shown relates to various aspects.

[0105] As shown, a conditioning network (represented as "conditioning network") 510 may be configured to receive an observed environment vector f as input and provide a conditioning vector c as output. In some aspects, conditioning network 510 may be or include one or more neural networks. The conditioning vector c may include client-specific parameters that may be loaded into one or more layers of autoencoder 520. In some aspects, autoencoder 520 may be or include one or more neural networks. Autoencoder 520 may be a regular autoencoder or a variational autoencoder. Autoencoder 520 may include encoder 530, which is configured to receive an observed wireless communication vector x as input and provide a latent vector h as output. Autoencoder 520 may also include decoder 540, which is configured to receive the latent vector h as input and provide (e.g., recover) the observed wireless communication vector x as output.

[0106] Depending on various aspects, the conditioning network 510 and the autoencoder 520 may be trained prior to inference. In some aspects, training the conditioning network 510 and / or the autoencoder 520 may include determining a set of neural network parameters w, θ, and φ that maximize the variational lower bound function (which may be interchangeably referred to as the evidence lower bound (ELBO) function) corresponding to the conditioning network 510 and the autoencoder 520.

[0107] In some respects, in order to establish the evidence function log p w，θ Finding the ELBO on (x, f|s) allows us to introduce the variational distribution q. φ，w (h, c|x, f) = q φ (h|x,c)p w (c|f), and the ELBO can be written as:

[0108]

[0109] Where φ represents the shared parameters used for encoder 530, θ represents the shared parameters used for decoder 540, and w represents the shared parameters used to adjust the network. ELBO These can be used during training to optimize the neural network parameters w, θ, and φ. ELBO It can be simplified to:

[0110]

[0111] The last term, log p(f|s), may be unimportant for neural network optimization. Accordingly, ELBO... It can be rewritten as:

[0112]

[0113] in And P w (c|f) is parameterized by the regulation network 510, q φ( h|x, c) are parameterized by encoder 530 of autoencoder 520, p θ (x|h, c) is parameterized by the decoder 540 of the autoencoder 520, p θ (h|c) is parameterized by the previous network to train the autoencoder 520. It is the reconstruction loss of the autoencoder 520, and KL(q) φ (h|x,c)||p θ (h|c)) is the regularization term used for the autoencoder 520.

[0114] In some aspects, training the conditioning network 510 and the autoencoder 520 may include finding ways to make ELBO Maximize the neural network parameters w, θ, and φ. In some respects, the negative variational lower bound function can correspond to the loss function associated with the autoencoder 520. Here, by The given negative ELBO (also known as variational free energy) can be used as the loss function during training, and the stochastic gradient descent (SGD) algorithm can be used to optimize the shared neural network parameters w, θ, and φ. The adjustment variable c can be considered as a client-specific parameter for the autoencoder 520. Thus, when adjusting network 510, it provides client-specific parameters for the autoencoder 520. It can be considered as a loss function for the autoencoder 520.

[0115] In some respects, the conditioning vector c from the conditioning network 510 can adjust the autoencoder 520 to perform well in the observed environment. The conditioning vector c may include client-specific parameters φ for the encoder. c And the client-specific parameter θ used for the decoder. c Therefore, the encoder parameter set can include a shared parameter φ and a parameter φ that varies from client to client. c The decoder parameter set may include a shared parameter θ and client-specific parameters θ. c The neural network parameter set can include parameters used for the encoder, decoder, and network conditioning.

[0116] In some respects, the conditioning network 510 and / or the autoencoder 520 can be trained using unsupervised learning procedures. The conditioning network 510 and / or the autoencoder 520 can be trained using joint learning procedures (e.g., such as...). Figure 8-13 (as shown in the image).

[0117] As indicated above, Figure 5 This is provided merely as an example. Other examples may differ from those provided. Figure 5 The example described.

[0118] Figure 6 This is a diagram illustrating an example 600 of wireless communication using an autoencoder and an associated conditioning network according to the present disclosure. As shown, client 605 and server UE 610 can communicate with each other. In some aspects, client 605 may be, similar to, include, or be included in... Figure 3 In the client 302 shown. In some respects, server 610 may be, similar to, include, or be included in. Figure 3 The server 304 shown is shown.

[0119] As indicated by reference numeral 615, server 610 can transmit and client 605 can receive wireless communications. In some aspects, the wireless communications may include reference signals (e.g., CSI-RS, etc.), data communications, control communications, etc. In some aspects, the wireless communications may be carried using a physical downlink control channel (PDCCH), a physical downlink shared channel (PDSCH), a physical sidelink control channel (PSCCH), a physical sidelink shared channel (PSSCH), etc.

[0120] As indicated by reference numeral 620, client 605 can determine a conditioning vector associated with one or more features, which are related to the environment of client 605. Client 605 can determine the conditioning vector based at least in part on observed environment vectors. In some aspects, client 605 may use a conditioning network (e.g., Figure 5 The regulation network 510 shown Figure 3 The client-side conditioning network 308, etc., shown in the figure, determines the conditioning vector. As indicated by reference numeral 625, the client 605 can determine the latent vector. For example, in some aspects, the client can use an autoencoder (e.g., Figure 5 The self-encoder 520 shown Figure 3 The client autoencoder 310 shown in the figure determines the latent vector.

[0121] As indicated by reference numeral 630, client 605 can transmit and server 610 can receive the observed environment vector and the latent vector. In some aspects, the observed environment vector and the latent vector may be carried in the Physical Uplink Control Channel (PUCCH), Physical Uplink Shared Channel (PUSCH), etc. As indicated by reference numeral 635, server 610 can determine the observed wireless communication vector. In some aspects, server 610 can determine the observed wireless communication vector based at least in part on the latent vector and the adjustment vector.

[0122] For example, in some aspects, server 610 can use server-side network regulation (e.g., Figure 5 The regulation network 510 shown Figure 3 The server conditioning network 318 shown (etc.) determines the conditioning vector at least in part based on the observed environment vector. This conditioning vector may include client-specific decoder parameters, which the server 610 may load into the server autoencoder (e.g., Figure 5 The self-encoder 520 shown Figure 3 The server 610 can use the decoder (such as the server autoencoder 320 shown) to determine the observed wireless communication vector. As indicated by reference numeral 640, the server can perform wireless communication actions based at least in part on the determined observed wireless communication vector.

[0123] As indicated above, Figure 6 This is provided merely as an example. Other examples may differ from those provided. Figure 6 The example described.

[0124] Figure 7 This is a diagram illustrating example 700 of the joint learning of the autoencoder and associated conditioning network according to this disclosure. As shown, server 705, first client 710, and second client 715 can communicate with each other. In some aspects, server 705 can be, similar to, include, or be included in... Figure 6 The server 610 shown Figure 3 The server 304, etc., shown in the example. In some respects, client 710 and / or client 715 may be, similar to, include, or be included in. Figure 6 The client 605 shown Figure 3 As shown in client 302, etc. In some respects, any number of additional clients can communicate with server 705.

[0125] In some respects, joint learning may include jointly training a regulatory network (e.g., Figure 5 The conditioning network 510 shown) and the autoencoder (e.g., Figure 5 (The autoencoder 520 shown). In joint learning, the conditioning network and the autoencoder can be trained by a client. In some aspects, joint learning may include alternating between training the conditioning network and training the autoencoder. In some aspects, the mapping to the conditioning vector c may change slowly because the client features selected for the observed vector f may be relatively static or change slowly. As a result, in some aspects, alternating between training the conditioning vector and training the autoencoder may include: performing a first set of training iterations associated with the conditioning vector according to a first training frequency; and performing a second set of training iterations associated with the autoencoder according to a second training frequency higher than the first training frequency.

[0126] As indicated by reference numeral 720, server 705 may transmit a shared neural network parameter set (e.g., parameters w, θ, and φ, as described above) to a set of clients (e.g., clients 710 and 715). Figure 5 (As described). The neural network parameters w may correspond to the conditioning network, and the neural network parameters θ and φ may correspond to the autoencoder. As shown by reference numeral 725, the first client 710 may determine a locally updated set of shared neural network parameters. As shown by reference numeral 730, the first client 710 may transmit, and the server 705 may receive, the locally updated set of shared neural network parameters.

[0127] As indicated by reference numeral 735, the second client 715 can also determine a locally updated set of shared neural network parameters. As indicated by reference numeral 740, the second client 715 can transmit, and the server 705 can receive, the locally updated set of shared neural network parameters. As indicated by reference numeral 745, the server 705 can determine a “final” server-updated set of shared neural network parameters. The server 705 can determine the server-updated set of shared neural network parameters by averaging the locally updated shared neural network parameters received from clients 710 and 715. The server 705 can use the server-updated set of shared neural network parameters to update the server conditioning network and the server autoencoder.

[0128] As indicated above, Figure 7 This is provided merely as an example. Other examples may differ from those provided. Figure 7 The example described.

[0129] Figure 8 This is a diagram illustrating an example process 800 of joint learning of the autoencoder and associated conditioning network according to this disclosure. Process 800 can be, for example, performed by a client (e.g., Figure 7 The client 710 shown Figure 7 The client 715 shown Figure 6The client 605 shown Figure 3 The aspects of the fully federated learning process performed by the client (302, etc.) shown in the figure.

[0130] Process 800 may include receiving a shared neural network parameter set from a server (box 805). In some aspects, the shared neural network parameter set may include a shared conditioning network parameter set w and a shared autoencoder parameter set (φ, θ). The process may further include obtaining an observed environment training vector f and an observed wireless communication training vector x (box 810). The process may include inputting the observed environment training vector into the conditioning network to determine a conditioning training vector (box 815). The conditioning training vector may include a client-specific encoder training parameter set and a client-specific decoder training parameter set.

[0131] As shown, process 800 may include determining the loss of the client autoencoder (box 820). For example, in some aspects, the client may determine a first loss corresponding to the autoencoder by inputting the observed wireless communication training vector x into the encoder of the autoencoder to determine a hidden training vector h. The client may input the hidden training vector h into the decoder of the autoencoder to determine the training output of the autoencoder. The training output may include a reconstruction of the observed wireless communication training vector x. The client may determine the loss associated with the autoencoder based at least in part on the training output. In some aspects, the client may load at least one layer of the encoder of the autoencoder using a client-specific set of encoder training parameters from the adjustment vector, load at least one layer of the decoder of the autoencoder using a client-specific set of decoder training parameters from the adjustment vector, and so on. The loss may be associated with a shared set of neural network parameters. In some aspects, the client may determine a regularization term corresponding to the autoencoder. The client may determine the loss based at least in part on a second regularization term.

[0132] Process 800 may further include updating the shared neural network parameter set (box 825). In some aspects, the client may update the shared neural network parameter set by determining multiple gradients of the loss relative to the shared neural network parameter set, and at least in part based on those multiple gradients.

[0133] In some aspects, process 800 may further include repeating operations 810-825 a specified number of times (box 830). For example, the client may update the shared neural network parameter set a specified number of times to determine a locally updated shared neural network parameter set. Process 800 may include providing the locally updated shared neural network parameter set to the server (box 835). For example, the client may transmit the locally updated shared neural network parameter set to the server.

[0134] As indicated above, Figure 8 This is provided merely as an example. Other examples may differ from those provided. Figure 8 The example described.

[0135] Figure 9 This is a diagram illustrating an example process 900 for the joint learning of an autoencoder and an associated conditioning network according to this disclosure. Example process 900 can be, for example, provided by a server (e.g., Figure 7 The server 705 shown Figure 6 The server 610 shown Figure 3 The aspects of the joint learning process performed by the server (e.g., 304 error shown) are illustrated.

[0136] Process 900 may include selecting a set of clients from which to receive updates (box 905). For example, the server may select a set of clients from which to receive updates associated with the conditioning network and the autoencoder. Process 900 may include transmitting a shared neural network parameter set to the set of clients (box 910). For example, the server may transmit a shared neural network parameter set (w, φ, θ) to the set of clients.

[0137] Process 900 may include receiving multiple locally updated shared neural network parameter sets from the client set (box 915). For example, the server may receive locally updated shared neural network parameter sets (w, φ, θ) associated with the conditioning network and the autoencoder. Process 900 may further include determining a server-updated shared neural network parameter set (box 920). For example, the server may determine the server-updated shared neural network parameter set based at least in part on the multiple locally updated shared neural network parameter sets. As shown, process 900 may include returning to box 905 (box 925) and repeating process 900 using one or more additional client sets.

[0138] As indicated above, Figure 9 This is provided merely as an example. Other examples may differ from those provided. Figure 9 The example described.

[0139] Figure 10 This is a diagram illustrating an example process 1000 of joint learning of an autoencoder and an associated conditioning network according to this disclosure. Process 1000 can be, for example, performed by a client (e.g., Figure 7 The client 710 shown Figure 7 The client 715 shown Figure 6 The client 605 shown Figure 3 The aspects of the fully federated learning process performed by the client (302, etc.) shown in the figure.

[0140] Process 1000 may include receiving a shared neural network parameter set from a server (box 1005). In some aspects, the shared neural network parameter set may include a shared neural network parameter set (w, φ, θ) associated with the conditioning network and the autoencoder. The process may further include obtaining an observed environmental training vector f and an observed wireless communication training vector x (box 1010). The process may include inputting the observed environmental training vector into the conditioning network to determine a conditioning training vector (box 1015). The conditioning training vector may include a client-specific encoder training parameter set and a client-specific decoder training parameter set.

[0141] As shown, process 1000 may include determining the loss of the client autoencoder (box 1020). For example, in some aspects, the client may determine a first loss corresponding to the client autoencoder by inputting the observed wireless communication training vector x into the encoder of the client autoencoder to determine a hidden training vector h. The client may input the hidden training vector h into the decoder of the client autoencoder to determine the training output of the client autoencoder. The training output may include a reconstruction of the observed wireless communication training vector x. The client may determine the loss associated with the client autoencoder based at least in part on the training output. In some aspects, the client may load at least one layer of the encoder of the client autoencoder using a client-specific set of encoder training parameters from the adjustment vector; load at least one layer of the decoder of the client autoencoder using a client-specific set of decoder training parameters from the adjustment vector, and so on. The loss may be associated with a shared set of neural network parameters. In some aspects, the client may determine a regularization term corresponding to the client autoencoder. The client may determine the loss based at least in part on a second regularization term.

[0142] Process 1000 may further include updating the shared tuning network parameter set (box 1025). In some aspects, the client may update the shared tuning network parameters by determining multiple gradients of the loss relative to the shared tuning network parameter set, and update the shared tuning network parameter set at least in part based on the multiple gradients.

[0143] In some aspects, process 1000 may further include repeating operations 1010-1025 a specified number of times (box 1030). For example, the client may update the shared tunable network parameter set a specified number of times to determine a locally updated shared tunable network parameter set. Process 1000 may include providing the locally updated shared tunable network parameter set to the server (box 1035). For example, the client may transmit the locally updated shared tunable network parameter set to the server.

[0144] As indicated above, Figure 10This is provided merely as an example. Other examples may differ from those provided. Figure 10 The example described.

[0145] Figure 11 This is a diagram illustrating an example process 1100 of joint learning of an autoencoder and an associated conditioning network according to this disclosure. Process 1100 can be, for example, performed by a client (e.g., Figure 7 The client 710 shown Figure 7 The client 715 shown Figure 6 The client 605 shown Figure 3 The aspects of the fully federated learning process performed by the client (302, etc.) shown in the figure.

[0146] Process 1100 may include receiving a shared neural network parameter set from a server (box 1105). In some aspects, the shared neural network parameter set may include a shared neural network parameter set (w, φ, θ) associated with the conditioning network and the autoencoder. The process may further include obtaining an observed environmental training vector f and an observed wireless communication training vector x (box 1110). The process may include inputting the observed environmental training vector into the conditioning network to determine a conditioning training vector (box 1115). The conditioning training vector may include a client-specific encoder training parameter set and a client-specific decoder training parameter set.

[0147] As shown, process 1100 may include determining the loss of the client autoencoder (box 1120). For example, in some aspects, the client may determine a first loss corresponding to the client autoencoder by inputting the observed wireless communication training vector x into the encoder of the client autoencoder to determine a hidden training vector h. The client may input the hidden training vector h into the decoder of the client autoencoder to determine the training output of the client autoencoder. The training output may include a reconstruction of the observed wireless communication training vector x. The client may determine the loss associated with the client autoencoder based at least in part on the training output. In some aspects, the client may load at least one layer of the encoder of the client autoencoder using a client-specific set of encoder training parameters from the adjustment vector; load at least one layer of the decoder of the client autoencoder using a client-specific set of decoder training parameters from the adjustment vector, and so on. The loss may be associated with a shared set of neural network parameters. In some aspects, the client may determine a regularization term corresponding to the client autoencoder. The client may determine the loss based at least in part on a second regularization term.

[0148] Process 1100 may further include updating the shared autoencoder parameter set (box 1125). In some aspects, the client may update the shared autoencoder parameters by determining multiple gradients of the loss relative to the shared autoencoder parameter set, and update the shared autoencoder parameter set at least in part based on those multiple gradients.

[0149] In some aspects, process 1100 may further include repeating operations 1110-1125 a specified number of times (box 1130). For example, a client may update the shared autoencoder parameter set a specified number of times to determine a locally updated shared autoencoder parameter set. Process 1000 may include providing the locally updated shared autoencoder parameter set to the server (box 1135). For example, the client may transmit the locally updated shared autoencoder parameter set to the server.

[0150] As indicated above, Figure 11 This is provided merely as an example. Other examples may differ from those provided. Figure 11 The example described.

[0151] Figure 12 This is a diagram illustrating an example process 1200 for the joint learning of an autoencoder and an associated conditioning network according to this disclosure. Example process 1200 can be, for example, provided by a server (e.g., Figure 7 The server 705 shown Figure 6 The server 610 shown Figure 3 The aspects of the joint learning process performed by the server (e.g., 304 error shown) are illustrated.

[0152] Process 1200 may include selecting a set of clients from which to receive updates (box 1205). For example, the server may select a set of clients from which to receive updates associated with the conditioning network and the autoencoder. Process 1200 may include transmitting a shared neural network parameter set to the set of clients (box 1210). For example, the server may transmit a shared neural network parameter set (w, φ, θ) to the set of clients.

[0153] Process 1200 may include receiving multiple locally updated shared neural network parameter sets from the client set (box 1215). For example, the server may receive locally updated shared neural network parameter sets (w, φ, θ) associated with the conditioning network and the autoencoder. Process 1200 may further include determining a server-updated shared conditioning network parameter set (box 1220). For example, the server may determine the server-updated shared conditioning network parameter set based at least in part on the multiple locally updated shared conditioning network parameter sets. As shown, process 1200 may include returning to box 1205 (box 1225) and repeating process 1200 using one or more additional client sets.

[0154] As indicated above, Figure 12 This is provided merely as an example. Other examples may differ from those provided. Figure 12 The example described.

[0155] Figure 13 This is a diagram illustrating an example process 1300 for joint learning of an autoencoder and an associated conditioning network according to this disclosure. Example process 1300 can be, for example, provided by a server (e.g., Figure 7 The server 705 shown Figure 6 The server 610 shown Figure 3 The aspects of the joint learning process performed by the server (e.g., 304 error shown) are illustrated.

[0156] Process 1300 may include selecting a set of clients from which to receive updates (box 1305). For example, the server may select a set of clients from which to receive updates associated with the conditioning network and the autoencoder. Process 1300 may include transmitting a shared neural network parameter set to the set of clients (box 1310). For example, the server may transmit a shared neural network parameter set (w, φ, θ) to the set of clients.

[0157] Process 1300 may include receiving multiple locally updated shared neural network parameter sets from the client set (box 1315). For example, the server may receive locally updated shared neural network parameter sets (w, φ, θ) associated with the conditioning network and the autoencoder. Process 1300 may further include determining a server-updated shared autoencoder parameter set (box 1320). For example, the server may determine the server-updated shared autoencoder parameter set based at least in part on the multiple locally updated shared autoencoder network parameter sets. As shown, process 1300 may include returning to box 1305 (box 1325) and repeating process 1300 using one or more additional client sets.

[0158] As indicated above, Figure 13 This is provided merely as an example. Other examples may differ from those provided. Figure 13 The example described.

[0159] Figure 14 This is a diagram illustrating an example process 1400 performed by a client, for example, according to this disclosure. Example process 1400 is a client (e.g., Figure 7 The client 710 shown Figure 7 The client 715 shown Figure 6 The client 605 shown Figure 3 The example shown is a client 302, etc., performing operations associated with joint learning of neural network parameters generated for wireless communication, which vary from client to client.

[0160] like Figure 14 As shown, in some aspects, process 1400 may include using a conditioning network to determine a client-specific set of encoder parameters based at least in part on observed environment vectors (box 1410). For example, the client (e.g., using...) Figure 16 The communication manager 1604 described herein can use the conditioning network to determine the client-specific set of encoder parameters based at least in part on the observed environment vectors, as described above.

[0161] like Figure 14 As further shown, in some aspects, process 1400 may include determining a latent vector using an autoencoder and at least in part based on a client-specific set of encoder parameters (box 1420). For example, the client (e.g., using an autoencoder) may determine a latent vector using an autoencoder and at least in part based on a client-specific set of encoder parameters. Figure 16 The communication manager 1604 described herein can determine the latent vector using an autoencoder and at least in part based on an encoder parameter set that varies from client to client, as described above.

[0162] like Figure 14 As further shown, in some aspects, process 1400 may include transmitting the observed environment vector and the implicit vector (box 1430). For example, the client (e.g., using...) Figure 16 The transmission component 1606 described herein can transmit the observed environment vector and the implicit vector, as described above.

[0163] Process 1400 may include additional aspects, such as any single aspect or any combination of aspects described below and / or in conjunction with one or more other processes described elsewhere herein.

[0164] In a first aspect, the conditioning network is configured to receive an observed environment vector as input and provide a conditioning vector as output, wherein the conditioning vector includes: a client-specific set of encoder parameters and a client-specific set of decoder parameters.

[0165] In a second aspect, either alone or in combination with the first aspect, the adjustment network includes: a first subnetwork configured to determine a client-specific set of encoder parameters, and a second subnetwork configured to determine a client-specific set of decoder parameters.

[0166] In a third aspect, either alone or in combination with one or more of the first and second aspects, the autoencoder includes: an encoder configured to receive an observed wireless communication vector as input and provide the latent vector as output; and a decoder configured to receive the latent vector as input and provide the observed wireless communication vector as output.

[0167] In a fourth aspect, either alone or in combination with one or more of the first to third aspects, the encoder includes at least one encoder layer configured to use a client-specific set of encoder parameters, and determining the latent vector includes using the at least one encoder layer to map the observed wireless communication vector to the latent vector.

[0168] In the fifth aspect, mapping the observed wireless communication vector to the latent vector, either alone or in combination with one or more of the first to fourth aspects, includes: using an encoder parameter set to map the observed wireless communication vector to the latent vector, wherein the encoder parameter set includes: a client-specific encoder parameter set and a shared encoder parameter set.

[0169] In the sixth aspect, alone or in combination with one or more of the first to fifth aspects, the at least one encoder layer comprises: a first layer set that uses a first subset of a client-specific encoder parameter set, a second layer set that uses a first subset of a shared encoder parameter set, and a third layer set that uses a second subset of both the client-specific encoder parameter set and the shared encoder parameter set.

[0170] In the seventh aspect, either alone or in combination with one or more of the first to sixth aspects, the decoder includes at least one decoder layer configured to use a client-specific set of decoder parameters, wherein the decoder is configured to determine the observed wireless communication vector by mapping the latent vector to the observed wireless communication vector using the at least one decoder layer.

[0171] In the eighth aspect, mapping a latent vector to an observed wireless communication vector, either alone or in combination with one or more of the first to seventh aspects, includes using a decoder parameter set to map the latent vector to the observed wireless communication vector, wherein the decoder parameter set includes: a client-specific decoder parameter set and a shared decoder parameter set.

[0172] In the ninth aspect, alone or in combination with one or more of the first to eighth aspects, the at least one decoder layer comprises: a first layer set that uses a first subset of a client-specific decoder parameter set, a second layer set that uses a first subset of a shared decoder parameter set, and a third layer set that uses a second subset of both the client-specific decoder parameter set and the shared decoder parameter set.

[0173] In the tenth aspect, either alone or in combination with one or more of the first to ninth aspects, the observed environment vector includes one or more characteristic components, wherein the one or more characteristic components indicate: client vendor identifier, client antenna configuration, large-scale channel characteristics, CSI-RS configuration, image obtained by an imaging device, a portion of the estimated propagation channel, or a combination thereof.

[0174] In the eleventh aspect, alone or in combination with one or more of the first to tenth aspects, the large-scale channel characteristics indicate: channel-related delay spread, channel-related power delay profile, channel-related Doppler measurement, channel-related Doppler spectrum, channel-related SNR, channel-related SINR, RSRP, RSSI, etc.

[0175] In the twelfth aspect, the latent vector is associated with the wireless communication task, either alone or in combination with one or more of the first to eleventh aspects.

[0176] In the thirteenth aspect, alone or in combination with one or more of the first to twelfth aspects, the wireless communication task includes: determining the CSF, determining location information associated with the client, determining the modulation associated with the wireless communication, determining the waveform associated with the wireless communication, or a combination thereof.

[0177] In the fourteenth aspect, alone or in combination with one or more of the first to thirteenth aspects, the wireless communication task includes determining the CSF, and the process 1400 includes: receiving the CSI-RS, determining the CSI-RS at least in part based on the CSI-RS, and providing the CSI as an input to the autoencoder.

[0178] In the fifteenth aspect, either alone or in combination with one or more of the first to fourteenth aspects, the implicit vector includes compressed channel state feedback.

[0179] In the sixteenth aspect, transmitting the observed environment vector and the latent vector, alone or in combination with one or more of the first to fifteenth aspects, includes transmitting the observed environment vector and the latent vector using a physical uplink control channel, a physical uplink shared channel, or a combination thereof.

[0180] In the seventeenth aspect, alone or in combination with one or more of the first to sixteenth aspects, the autoencoder includes a variational autoencoder.

[0181] In the eighteenth aspect, alone or in combination with one or more of the first to seventeenth aspects, process 1400 includes: training at least one of the conditioning network or the autoencoder.

[0182] In the nineteenth aspect, training at least one of the conditioning network or the autoencoder, either alone or in combination with one or more of the first to eighteenth aspects, comprises using an unsupervised learning procedure.

[0183] In the twentieth aspect, either alone or in combination with one or more of the first to nineteenth aspects, process 1400 includes: using a joint learning procedure to collaboratively train the conditioning network and the autoencoder.

[0184] In the twenty-first aspect, training the conditioning network and the autoencoder collaboratively, either alone or in combination with one or more of the first to twentieth aspects, includes determining a shared set of neural network parameters that maximizes the variational lower bound function corresponding to the conditioning network and the autoencoder.

[0185] In aspect twenty-two, either alone or in combination with one or more of aspects one through twenty-one, the negative variational lower bound function corresponds to the loss function.

[0186] In the twenty-third aspect, the loss function includes the reconstruction loss of the autoencoder, either alone or in combination with one or more of the first to twenty-second aspects.

[0187] In the twenty-fourth aspect, either alone or in combination with one or more of the first to twenty-third aspects, the loss function includes a regularization term for the autoencoder.

[0188] In aspect twenty-five, alone or in combination with one or more of aspects one through twenty-four, the autoencoder is a regular autoencoder, wherein the negative variational lower bound function does not include a regularization term.

[0189] In the twenty-sixth aspect, the joint learning procedure is performed, either alone or in combination with one or more of the first to twenty-fifth aspects, including jointly training the conditioning network and the autoencoder.

[0190] In aspect twenty-seven, the joint learning procedure is performed, either alone or in combination with one or more of aspects one through twenty-six, including alternating between training the regulatory network and training the autoencoder.

[0191] In the twentieth aspect, alone or in combination with one or more of the first to twenty-seventh aspects, alternating between training the conditioning network and training the autoencoder includes: performing a first plurality of training iterations associated with the conditioning network according to a first training frequency, and performing a second plurality of training iterations associated with the autoencoder according to a second training frequency, the second training frequency being different from the first training frequency.

[0192] In the twenty-ninth aspect, performing a joint learning procedure, either alone or in combination with one or more of the first to twenty-eighth aspects, includes: receiving from a server a shared neural network parameter set corresponding to the conditioning network and the autoencoder; obtaining observed environment training vectors; inputting the observed environment training vectors into the conditioning network to determine conditioning training vectors, which include client-specific encoder training parameter sets and client-specific decoder training parameter sets; loading at least one layer of the encoder of the autoencoder using the client-specific encoder training parameter set; loading at least one layer of the decoder of the autoencoder using the client-specific decoder training parameter set; obtaining observed wireless communication training vectors; inputting the observed wireless communication training vectors into the encoder to determine training latent vectors; inputting the training latent vectors into the decoder to determine training output of the autoencoder, wherein the training output includes a reconstruction of the observed wireless communication training vectors; and determining a loss associated with the autoencoder based at least in part on the training output.

[0193] In the thirtieth aspect, alone or in combination with one or more of the first to twenty-ninth aspects, process 1400 includes: determining a regularization term corresponding to the autoencoder, wherein determining the loss includes: determining the loss based at least in part on the regularization term.

[0194] In the thirty-first aspect, alone or in combination with one or more of the first to thirtieth aspects, process 1400 includes: determining a plurality of gradients of the loss relative to a shared neural network parameter set; and updating the shared neural network parameter set at least in part based on the plurality of gradients.

[0195] In the thirty-second aspect, alone or in combination with one or more of the first to thirty-first aspects, process 1400 includes: updating the shared neural network parameter set a specified number of times to determine a locally updated shared neural network parameter set.

[0196] In the thirty-third aspect, either alone or in combination with one or more of the first to thirty-two aspects, process 1400 includes: transmitting a locally updated set of shared neural network parameters to the server.

[0197] In the thirty-fourth aspect, alone or in combination with one or more of the first to thirty-third aspects, process 1400 includes: determining multiple gradients of the loss relative to a subset of a shared neural network parameter set, wherein the subset includes a shared regulation network parameter set, and updating the shared regulation network parameter set at least in part based on the multiple gradients.

[0198] In the thirty-fifth aspect, alone or in combination with one or more of the first to thirty-fourth aspects, process 1400 includes: updating the shared regulation network parameter set a specified number of times to determine a locally updated shared regulation network parameter set.

[0199] In the thirty-sixth aspect, alone or in combination with one or more of the first to thirty-fifth aspects, process 1400 includes: transmitting a locally updated set of shared adjustment network parameters to the server.

[0200] In the thirty-seventh aspect, alone or in combination with one or more of the first to thirty-sixth aspects, process 1400 includes: determining multiple gradients of the loss relative to a subset of a shared neural network parameter set, wherein the subset includes a shared autoencoder parameter set, and updating the shared autoencoder parameter set at least in part based on the multiple gradients.

[0201] In the thirty-eighth aspect, alone or in combination with one or more of the first to thirty-seventh aspects, process 1400 includes: updating the shared autoencoder parameter set a specified number of times to determine a locally updated shared autoencoder parameter set.

[0202] In the thirty-ninth aspect, alone or in combination with one or more of the first to thirty-eighth aspects, process 1400 includes: transmitting a locally updated set of shared autoencoder parameters to the server.

[0203] In the fortieth aspect, alone or in combination with one or more of the first to thirty-ninth aspects, the regulation network corresponds to a server regulation network implemented at the server.

[0204] In the forty-first aspect, alone or in combination with one or more of the first to forty aspects, the regulation network is a copy of the client implementation of the server regulation network.

[0205] In aspect 42, either alone or in combination with one or more of aspects 1 to 41, the autoencoder corresponds to a server autoencoder implemented at the server.

[0206] In aspect 43, alone or in combination with one or more of aspects 1 to 42, the autoencoder is a copy of the client implementation of the server autoencoder.

[0207] although Figure 14 An example box of process 1400 is shown, but in some respects, process 1400 may include... Figure 14 The boxes depicted in the process are compared to additional boxes, fewer boxes, different boxes, or boxes arranged differently. Additionally or alternatively, two or more boxes in process 1400 can be executed in parallel.

[0208] Figure 15 This is a diagram illustrating an example process 1500 performed by a server, for example, according to this disclosure. Example process 1500 is a server (e.g., Figure 7 The server 705 shown Figure 6 The server 610 shown Figure 3 The example shown is of a server (304, etc.) performing operations associated with joint learning for generating neural network parameters that vary from client to client for wireless communication.

[0209] like Figure 15 As shown, in some aspects, process 1500 may include receiving from a client an observed environment vector associated with one or more features, which are associated with the client's environment (box 1510). For example, the server (e.g., using...) Figure 18 The receiving component 1802 described herein can receive from the client an observed environment vector associated with one or more features, which are associated with the client's environment, as described above.

[0210] like Figure 15 As further shown, in some aspects, process 1500 may include receiving from the client an implicit vector based at least in part on a client-specific set of encoder parameters (box 1520). For example, the server (e.g., using...) Figure 18 The receiving component 1802 described herein can receive, as described above, an implicit vector based at least in part on a set of encoder parameters that varies from client to client.

[0211] like Figure 15 As further shown, in some aspects, process 1500 may include determining the observed wireless communication vector based at least in part on the latent vector and the observed environment vector (box 1530). For example, the server (e.g., using...) Figure 18 The communication manager 1804 described herein can determine the observed wireless communication vector based at least in part on the implicit vector and the observed environment vector, as described above.

[0212] like Figure 15 As further shown, in some aspects, process 1500 may include performing wireless communication actions at least in part based on determining observed wireless communication vectors (box 1540). For example, the server (e.g., using...) Figure 18 The communication manager 1804 described herein can perform wireless communication actions, at least in part, based on the determination of observed wireless communication vectors, as described above.

[0213] Process 1500 may include additional aspects, such as any single aspect or any combination of aspects described below and / or in conjunction with one or more other processes described elsewhere herein.

[0214] In the first aspect, the observed environment vector includes one or more feature components, wherein the one or more feature components indicate: client vendor identifier, client antenna configuration, large-scale channel characteristics, CSI-RS configuration, image obtained by an imaging device, a portion of the estimated propagation channel, or a combination thereof.

[0215] In the second aspect, alone or in combination with the first aspect, this large-scale channel characteristic indicates: channel-associated delay spread, channel-associated power delay profile, channel-associated Doppler measurement, channel-associated Doppler spectrum, channel-associated SNR, channel-associated SINR, RSRP, RSSI, etc.

[0216] In the third aspect, the latent vector is associated with the wireless communication task, either alone or in combination with one or more of the first and second aspects.

[0217] In a fourth aspect, alone or in combination with one or more of the first to third aspects, the wireless communication task includes: receiving CSF, receiving location information associated with the client, receiving modulation associated with wireless communication, receiving waveforms associated with wireless communication, or combinations thereof.

[0218] In a fifth aspect, alone or in combination with one or more of the first to fourth aspects, the wireless communication task includes receiving a CSF, and process 1500 includes transmitting a CSI-RS, wherein the CSF corresponds to the CSI-RS.

[0219] In the sixth aspect, either alone or in combination with one or more of the first to fifth aspects, the latent vector includes a compressed CSF.

[0220] In the seventh aspect, either alone or in combination with one or more of the first to sixth aspects, the observed environment vector and the latent vector are carried in PUCCH, PUSCH, or a combination thereof.

[0221] In the eighth aspect, determining the observed wireless communication vector, either alone or in combination with one or more of the first to seventh aspects, comprises: inputting the observed environment vector into a server conditioning network to determine a conditioning vector including a client-specific decoder parameter set; using the client-specific decoder parameter set to load at least one layer of the decoder of the server autoencoder; and inputting the implicit vector into the decoder to determine the observed wireless communication vector.

[0222] In the ninth aspect, alone or in combination with one or more of the first to eighth aspects, the server autoencoder includes a regular autoencoder or a variational autoencoder.

[0223] In the tenth aspect, either alone or in combination with one or more of the first to ninth aspects, the server conditioning network is configured to receive an observed environment vector as input and provide a conditioning vector as output, wherein the conditioning vector includes: a client-specific set of decoder parameters and a client-specific set of encoder parameters.

[0224] In the eleventh aspect, alone or in combination with one or more of the first to tenth aspects, the server conditioning network includes: a first subnetwork configured to determine a client-specific set of encoder parameters, and a second subnetwork configured to determine a client-specific set of decoder parameters.

[0225] In the twelfth aspect, alone or in combination with one or more of the first to eleventh aspects, the server autoencoder includes: an encoder configured to receive an observed wireless communication vector as input and provide the latent vector as output, and a decoder configured to receive the latent vector as input and provide the observed wireless communication vector as output.

[0226] In the thirteenth aspect, alone or in combination with one or more of the first to twelfth aspects, the encoder includes at least one encoder layer configured to use a set of encoder parameters that varies from client to client.

[0227] In the fourteenth aspect, alone or in combination with one or more of the first to thirteenth aspects, the decoder includes at least one decoder layer configured to use a client-specific set of decoder parameters, wherein the decoder is configured to determine the observed wireless communication vector by mapping the latent vector to the observed wireless communication vector using the at least one decoder layer.

[0228] In the fifteenth aspect, mapping a latent vector to an observed wireless communication vector, either alone or in combination with one or more of the first to fourteenth aspects, includes using a decoder parameter set to map the latent vector to the observed wireless communication vector, wherein the decoder parameter set includes: a client-specific decoder parameter set and a shared decoder parameter set.

[0229] In the sixteenth aspect, alone or in combination with one or more of the first to fifteenth aspects, the at least one decoder layer comprises: a first layer set that uses a first subset of a client-specific decoder parameter set, a second layer set that uses a first subset of a shared decoder parameter set, and a third layer set that uses a second subset of both the client-specific decoder parameter set and the shared decoder parameter set.

[0230] In the seventeenth aspect, alone or in combination with one or more of the first to sixteenth aspects, the server regulation network corresponds to the regulation network implemented at the client.

[0231] In the eighteenth aspect, alone or in combination with one or more of the first to seventeenth aspects, the server regulating network is a copy of the server implementation of the regulating network.

[0232] In the nineteenth aspect, alone or in combination with one or more of the first to eighteenth aspects, the server autoencoder corresponds to the autoencoder implemented at the client.

[0233] In the twentieth aspect, alone or in combination with one or more of the first to nineteenth aspects, the server autoencoder is a copy of the server implementation of the autoencoder.

[0234] In the twenty-first aspect, alone or in combination with one or more of the first to twentieth aspects, the server autoencoder includes a variational autoencoder.

[0235] In the twenty-second aspect, alone or in combination with one or more of the first to twenty-first aspects, process 1500 includes: updating at least one of the server conditioning network or the server autoencoder.

[0236] In aspect 23, process 1500 includes, alone or in combination with one or more of aspects 1 to 22, collaboratively updating the server conditioning network and the server autoencoder using a joint learning procedure.

[0237] In aspect twenty-four, the use of joint learning procedures, either alone or in combination with one or more of aspects one through twenty-three, includes: jointly updating the server conditioning network and the server autoencoder.

[0238] In aspect 25, the use of joint learning procedures, either alone or in combination with one or more of aspects 1 to 24, includes alternating between updating the server conditioning network and updating the server autoencoder.

[0239] In the twentieth aspect, either alone or in combination with one or more of the first to twenty-fifth aspects, alternating between updating the server conditioning network and updating the server autoencoder includes: performing a first plurality of update iterations associated with the server conditioning network according to a first update frequency, and performing a second plurality of update iterations associated with the server autoencoder according to a second update frequency, the second update frequency being different from the first update frequency.

[0240] In the twenty-seventh aspect, updating the server conditioning network and the server autoencoder collaboratively, either alone or in combination with one or more of the first to twenty-sixth aspects, comprises: selecting a set of clients from which to receive updates associated with the server conditioning network and the server autoencoder, wherein the set of clients includes the client and the at least one additional client; transmitting a shared neural network parameter set corresponding to the server conditioning network and the server autoencoder to the set of clients; receiving multiple locally updated shared neural network parameter sets from the set of clients; and determining a server-updated shared neural network parameter set based at least in part on the multiple locally updated shared neural network parameter sets.

[0241] In the twenty-eighth aspect, alone or in combination with one or more of the first to twenty-seventh aspects, process 1500 includes: transmitting a shared neural network parameter set updated by the server to the client set.

[0242] In aspect twenty-nine, determining the server-updated shared neural network parameter set, either alone or in combination with one or more of aspects one through twenty-eight, includes averaging multiple locally updated shared neural network parameter sets.

[0243] In the thirtieth aspect, updating the server conditioning network and the server autoencoder collaboratively, either alone or in combination with one or more of the first to twenty-ninth aspects, comprises: selecting a set of clients from which to receive updates associated with the server conditioning network, wherein the set of clients includes the client and the at least one additional client; transmitting a shared neural network parameter set corresponding to the server conditioning network and the server autoencoder to the set of clients; receiving multiple locally updated shared conditioning network parameter sets from the set of clients; and determining a server-updated shared conditioning network parameter set based at least in part on the multiple locally updated shared conditioning network parameter sets.

[0244] In the thirty-first aspect, alone or in combination with one or more of the first to thirtieth aspects, process 1500 includes: transmitting a server-updated set of shared adjustment network parameters to the client set.

[0245] In aspect thirty-two, determining the shared regulation network parameter set for server updates, either alone or in combination with one or more of aspects one through thirty-one, includes averaging the shared regulation network parameter sets for multiple local updates.

[0246] In the thirty-third aspect, updating the server conditioning network and the server autoencoder collaboratively, either alone or in combination with one or more of the first to thirty-two aspects, comprises: selecting a set of clients from which to receive updates associated with the server autoencoder, wherein the set of clients includes the client and the at least one additional client; transmitting a shared neural network parameter set corresponding to the server conditioning network and the server autoencoder to the set of clients; receiving a plurality of locally updated shared autoencoder parameter sets from the set of clients; and determining a server-updated shared autoencoder parameter set based at least in part on the plurality of locally updated shared autoencoder parameter sets.

[0247] In the thirty-fourth aspect, alone or in combination with one or more of the first to thirty-third aspects, process 1500 includes: transmitting a shared autoencoder parameter set updated by the server to the client set.

[0248] In aspect thirty-five, determining the shared autoencoder parameter set updated by the server, either alone or in combination with one or more of aspects one through thirty-four, includes averaging the shared autoencoder parameter sets updated locally.

[0249] although Figure 15 An example box of process 1500 is shown, but in some respects, process 1500 may include... Figure 15 The boxes depicted in the process are compared to additional boxes, fewer boxes, different boxes, or boxes arranged differently. Additionally or alternatively, two or more boxes in process 1500 can be executed in parallel.

[0250] Figure 16 This is a block diagram of an example device 1600 for wireless communication according to this disclosure. Device 1600 may be, similar to, include, or be included in a client (e.g., Figure 7 The client 710 shown Figure 7 The client 715 shown Figure 6 The client 605 shown Figure 3 (e.g., client 302 shown). In some aspects, equipment 1600 includes a receiving component 1602, a communication manager 1604, and a transmitting component 1606, which can communicate with each other (e.g., via one or more buses). As shown, equipment 1600 can use the receiving component 1606 and the transmitting component 1602 to communicate with another equipment 1608 (such as a client, server, UE, base station, or another wireless communication device).

[0251] In some respects, Equipment 1600 can be configured to perform the actions described in this article. Figure 3-7 One or more operations described herein. Additionally or alternatively, equipment 1600 may be configured to perform one or more processes described herein, such as... Figure 8 The process 800 Figure 10 Process 1000 Figure 11 Process 1100 Figure 14 Process 1400, or a combination thereof. In some aspects, equipment 1600 may include the above combinations. Figure 2 One or more components of the first UE as described.

[0252] Receiver 1602 may provide means for receiving communications (such as reference signals, control information, data communications, or combinations thereof) from equipment 1608. Receiver 1602 may provide the received communications to one or more other components of equipment 1600 (such as communications manager 1604). In some aspects, receiver 1602 may provide means for performing signal processing on the received communications (such as filtering, amplification, demodulation, analog-to-digital conversion, demultiplexing, deinterleaving, demapping, equalization, interference cancellation, or decoding, and other examples), and may provide the processed signals to one or more other components. In some aspects, receiver 1602 may include a combination of the above. Figure 2 The first UE described includes one or more antennas, demodulators, MIMO detectors, receiver processors, controllers / processors, memory, or combinations thereof.

[0253] The transmission component 1606 may provide means for transmitting communications (such as reference signals, control information, data communications, or combinations thereof) to the equipment 1608. In some aspects, the communication manager 1604 may generate communications and transmit the generated communications to the transmission component 1606 for transmission to the equipment 1608. In some aspects, the transmission component 1606 may provide means for performing signal processing (such as filtering, amplification, modulation, digital-to-analog conversion, multiplexing, interleaving, mapping, or encoding, etc.) on the generated communications and may transmit the processed signals to the equipment 1608. In some aspects, the transmission component 1606 may include combinations of the above. Figure 2 The first UE described includes one or more antennas, modulators, transmit MIMO processors, transmit processors, controllers / processors, memory, or combinations thereof. In some aspects, the transmit component 1606 may coexist with the receive component 1602 in a transceiver.

[0254] In some aspects, the communication manager 1604 may provide means for determining a client-specific set of encoder parameters using an adjustment network, at least in part, based on an observed environment vector; means for determining a latent vector using an autoencoder, also at least in part, based on a client-specific set of encoder parameters; and means for transmitting the observed environment vector and the latent vector. In some aspects, the communication manager 1604 may include a combination of the above. Figure 2 The first UE described includes a controller / processor, memory, or a combination thereof. In some aspects, the communication manager 1604 may include a receiving component 1602, a transmitting component 1606, etc. In some aspects, the means provided by the communication manager 1604 may include or be included in the means provided by the receiving component 1602, the transmitting component 1606, etc.

[0255] In some respects, the communication manager 1604 and / or one or more components of the communication manager 1604 may include or may be implemented within hardware (e.g., in combination with...). Figure 20 (One or more of the circuit systems described). In some aspects, the communication manager 1604 and / or one or more of its components may include or may be combined with the above. Figure 2 The described UE 120 is implemented within the controller / processor, memory, or a combination thereof.

[0256] In some respects, the communication manager 1604 and / or one or more components of the component set can be implemented in code (e.g., as software or firmware stored in memory), such as in combination with Figure 20 The described code. For example, the communication manager 1604 and / or a component (or a portion thereof) may be implemented as instructions or code stored in a non-transient computer-readable medium and executable by a controller or processor to perform the functions or operations of the communication manager 1604 and / or the component. If implemented in code, the functionality of the communication manager 1604 and / or the component can be achieved by combining the above. Figure 2 The controller / processor, memory, scheduler, communication unit, or a combination thereof of the described UE 120 are used to perform this action.

[0257] Figure 16 The number and arrangement of components shown are provided as an example. In practice, different arrangements may exist. Figure 16 The components shown are compared to additional components, fewer components, different components, or components arranged differently. Furthermore, Figure 16 The two or more components shown can be implemented within a single component, or Figure 16 The single component shown can be implemented as multiple distributed components. Additionally or alternatively, Figure 16 The component set shown (e.g., one or more components) can perform actions described as being performed by Figure 16 The other set of components shown performs one or more functions.

[0258] Figure 17 This is a diagram illustrating an example 1700 of the hardware implementation of device 1702 employing processing system 1704. Device 1702 may be, similar to, include, or be included in. Figure 16 In the equipment 1600 shown. In some aspects, equipment 1702 may be, similar to, include, or be included in the client (e.g., Figure 7 The client 710 shown Figure 7 The client 715 shown Figure 6 The client 605 shown Figure 3 (e.g., client 302 shown).

[0259] Processing system 1704 can be implemented with a bus architecture generally represented by bus 1706. Depending on the specific application and overall design constraints of processing system 1704, bus 1706 may include any number of interconnect buses and bridges. Bus 1706 links together various circuits including one or more processors and / or hardware components (represented by processor 1708, the illustrated components, and computer-readable medium / memory 1710). Bus 1706 may also link various other circuits, such as timing sources, peripheral devices, voltage regulators, power management circuits, etc.

[0260] Processing system 1704 may be coupled to transceiver 1712. Transceiver 1712 is coupled to one or more antennas 1714. Transceiver 1712 provides means for communicating with various other equipment via a transmission medium. Transceiver 1712 receives signals from one or more antennas 1714, extracts information from the received signals, and provides the extracted information to processing system 1704 (specifically, receiving component 1716). Additionally, transceiver 1712 receives information from processing system 1704 (specifically, transmission component 1718) and generates signals to be applied to the one or more antennas 1714, at least in part, based on the received information.

[0261] Processor 1708 is coupled to computer-readable medium / memory 1710. Processor 1708 is responsible for general processing, including the execution of software stored on computer-readable medium / memory 1710. When executed by processor 1708, the software causes processing system 1704 to perform the various functions described herein in conjunction with a client. Computer-readable medium / memory 1710 may also be used to store data manipulated by processor 1708 during software execution. Processing system 1704 may include communication manager 1720, which may be, or similar to, [other types of communication management systems]. Figure 16 The communication manager 1604 and / or shown Figure 1 and 2 The first communication manager 140 is shown. The processing system 1704 may include... Figure 17 Any number of additional components not described herein. The described and / or undescribed components may be software modules running in processor 1708, software modules residing in / stored in computer-readable medium / memory 1710, one or more hardware modules coupled to processor 1708, or some combination thereof.

[0262] In some aspects, processing system 1704 may be a component of UE 120 and may include memory 282 and / or at least one of the following: TX MIMO processor 266, RX processor 258, and / or controller / processor 280. In some aspects, apparatus 1702 for wireless communication provides means for determining a client-specific set of encoder parameters using an conditioning network based at least in part on observed environmental vectors, means for determining a latent vector using an autoencoder based at least in part on the client-specific set of encoder parameters, means for transmitting the observed environmental vector and the latent vector, etc. The aforementioned means may be one or more components of processing system 1704 of apparatus 1702 configured to perform the functions described by the aforementioned means. As described elsewhere herein, processing system 1704 may include TX MIMO processor 266, RX processor 258, and / or controller / processor 280. In one configuration, the aforementioned apparatus may be a TX MIMO processor 266, an RX processor 258, and / or a controller / processor 280 configured to perform the functions and / or operations described herein.

[0263] Figure 17 This is provided as an example. Other examples may differ from this combination. Figure 17 The example described.

[0264] Figure 18 This is a block diagram of an example device 1800 for wireless communication according to this disclosure. Device 1800 may be, similar to, include, or be included in a server (e.g., Figure 7 The server 705 shown Figure 6 The server 610 shown Figure 3 (e.g., server 304 shown). In some aspects, equipment 1800 includes a receiving component 1802, a communication manager 1804, and a transmitting component 1806, which can communicate with each other (e.g., via one or more buses). As shown, equipment 1800 can use the receiving component 1806 and the transmitting component 1802 to communicate with another equipment 1808 (such as a client, server, UE, base station, or another wireless communication device).

[0265] In some respects, Equipment 1800 can be configured to perform the actions described in this article. Figure 3-7 One or more operations described herein. Additionally or alternatively, equipment 1800 may be configured to perform one or more processes described herein (such as...). Figure 9 The process 900 Figure 12 Process 1200 Figure 13 Process 1300 Figure 15 (Process 1500, etc.). In some aspects, equipment 1800 may include the above combinations. Figure 2 One or more components of the base station described.

[0266] Receiver 1802 may provide means for receiving communications (such as reference signals, control information, data communications, or combinations thereof) from equipment 1808. Receiver 1802 may provide the received communications to one or more other components of equipment 1800 (such as communication manager 1804). In some aspects, receiver 1802 may provide means for performing signal processing on the received communications (such as filtering, amplification, demodulation, analog-to-digital conversion, demultiplexing, deinterleaving, demapping, equalization, interference cancellation, or decoding, and other examples), and may provide the processed signals to one or more other components. In some aspects, receiver 1802 may include a combination of the above. Figure 2 The described base station includes one or more antennas, demodulators, MIMO detectors, receiver processors, controllers / processors, memory, or combinations thereof.

[0267] The transmission component 1806 may provide means for transmitting communications (such as reference signals, control information, data communications, or combinations thereof) to the equipment 1808. In some aspects, the communication manager 1804 may generate communications and transmit the generated communications to the transmission component 1806 for transmission to the equipment 1808. In some aspects, the transmission component 1806 may provide means for performing signal processing (such as filtering, amplification, modulation, digital-to-analog conversion, multiplexing, interleaving, mapping, or encoding, etc.) on the generated communications and transmit the processed signals to the equipment 1808. In some aspects, the transmission component 1806 may include combinations of the above. Figure 2 The described base station includes one or more antennas, modulators, transmit MIMO processors, transmit processors, controllers / processors, memory, or combinations thereof. In some aspects, the transmit component 1806 may coexist with the receive component 1802 in a transceiver.

[0268] The communication manager 1804 may provide means for receiving from a client an observed environment vector associated with one or more features, the one or more features being associated with the client's environment; means for receiving from the client a latent vector based at least in part on a client-specific set of encoder parameters; means for determining an observed wireless communication vector based at least in part on the latent vector and the observed environment vector; means for performing a wireless communication action based at least in part on the determined observed wireless communication vector, etc. In some aspects, the communication manager 1804 may include combinations of the above. Figure 2 The described base station includes a controller / processor, memory, scheduler, communication unit, or a combination thereof. In some aspects, the communication manager 1804 may include a receiving component 1802, a transmitting component 1806, etc. In some aspects, the means provided by the communication manager 1804 may include or be included in the means provided by the receiving component 1802, the transmitting component 1806, etc.

[0269] In some respects, the communication manager 1804 and / or one or more of its components may include or be implemented within hardware (e.g., in conjunction with...). Figure 21 (One or more of the circuit systems described). In some aspects, the communication manager 1804 and / or one or more of its components may include or may be combined with the above. Figure 2 The BS 110 described is implemented within a controller / processor, memory, or a combination thereof.

[0270] In some respects, the communication manager 1804 and / or one or more of its components can be implemented in code (e.g., as software or firmware stored in memory), such as in combination with... Figure 21 The described code. For example, the communication manager 1804 and / or a component (or a portion thereof) may be implemented as instructions or code stored in a non-transient computer-readable medium and executable by a controller or processor to perform the functions or operations of the communication manager 1804 and / or the component. If implemented in code, the functionality of the communication manager 1804 and / or the component can be achieved by combining the above. Figure 2 The controller / processor, memory, scheduler, communication unit, or a combination thereof of the BS 110 described herein shall be used to perform the execution.

[0271] Figure 18 The number and arrangement of components shown are provided as an example. In practice, different arrangements may exist. Figure 18 The components shown are compared to additional components, fewer components, different components, or components arranged differently. Furthermore, Figure 18 The two or more components shown can be implemented within a single component, or Figure 18The single component shown can be implemented as multiple distributed components. Additionally or alternatively, Figure 18 The component set shown (e.g., one or more components) can be executed as described by Figure 18 The other set of components shown performs one or more functions.

[0272] Figure 19 This is a diagram illustrating an example 1900 of the hardware implementation of device 1902 employing processing system 1904. Device 1902 may be, similar to, include, or be included in. Figure 18 As shown in equipment 1800. In some respects, equipment 1902 may be, similar to, include, or be included in the server (e.g., Figure 7 The server 705 shown Figure 6 The server 610 shown Figure 3 (e.g., server 304 shown).

[0273] Processing system 1904 can be implemented with a bus architecture generally represented by bus 1906. Depending on the specific application and overall design constraints of processing system 1904, bus 1906 may include any number of interconnect buses and bridges. Bus 1906 links together various circuits including one or more processors and / or hardware components (represented by processor 1908, the illustrated components, and computer-readable medium / memory 1910). Bus 1906 may also link various other circuits, such as timing sources, peripheral devices, voltage regulators, power management circuits, etc.

[0274] Processing system 1904 may be coupled to transceiver 1912. Transceiver 1912 is coupled to one or more antennas 1914. Transceiver 1912 provides means for communicating with various other equipment via a transmission medium. Transceiver 1912 receives signals from one or more antennas 1914, extracts information from the received signals, and provides the extracted information to processing system 1904 (specifically, receiving component 1916). Additionally, transceiver 1912 receives information from processing system 1904 (specifically, transmitting component 1918) and generates signals to be applied to the one or more antennas 1914, at least in part, based on the received information.

[0275] Processor 1908 is coupled to computer-readable medium / memory 1910. Processor 1908 is responsible for general processing, including the execution of software stored on computer-readable medium / memory 1910. When executed by processor 1908, the software causes processing system 1904 to perform the various functions described herein in conjunction with a server. Computer-readable medium / memory 1910 may also be used to store data manipulated by processor 1908 during software execution. Processing system 1904 may include communication manager 1920, which may be, or similar to, [other types of communication management systems]. Figure 18 The communication manager 1804 and / or shown Figure 1 and 2 The first communication manager 150 is shown. The processing system 1904 may include... Figure 19 Any number of additional components not described herein. The described and / or undescribed components may be software modules running in processor 1908, software modules residing in / stored in computer-readable medium / memory 1910, one or more hardware modules coupled to processor 1908, or some combination thereof.

[0276] In some aspects, processing system 1904 may be a component of UE 120 and may include memory 282 and / or at least one of the following: TX MIMO processor 266, RX processor 258, and / or controller / processor 280. In some aspects, apparatus 1902 for wireless communication provides means for receiving from a client an observed environment vector associated with one or more features, said one or more features being associated with the client's environment; means for receiving from the client a latent vector at least partially based on a client-specific set of encoder parameters; means for determining an observed wireless communication vector at least partially based on the latent vector and the observed environment vector; means for performing a wireless communication action at least partially based on the determined observed wireless communication vector, etc. The aforementioned means may be one or more of the aforementioned components of processing system 1904 of apparatus 1902 configured to perform the functions described by the aforementioned means. As described elsewhere herein, processing system 1904 may include TX MIMO processor 266, RX processor 258, and / or controller / processor 280. In one configuration, the aforementioned apparatus may be a TXMIMO processor 266, an RX processor 258, and / or a controller / processor 280 configured to perform the functions and / or operations described herein.

[0277] Figure 19 This is provided as an example. Other examples may differ from this combination. Figure 19 The example described.

[0278] Figure 20This is a diagram illustrating an example 2000 of the code and circuitry implementation of a device 2002 for wireless communication. Device 2002 may be, similar to, include, or be included in... Figure 17 The equipment shown is 1702. Figure 16 As shown in equipment 1600, etc. In some aspects, equipment 2002 may be, similar to, include, or be included in the client (e.g., Figure 7 The client 710 shown Figure 7 The client 715 shown Figure 6 The client 605 shown Figure 3 (e.g., client 302 shown). Equipment 2002 may include processing system 2004, which may include a bus 2012 coupling one or more components (e.g., processor 2006, computer-readable medium / memory 2008, transceiver 2010, etc.). As shown, transceiver 2012 may be coupled to one or more antennas 2014.

[0279] like Figure 20 As further shown, apparatus 2002 may include a circuit system (circuit system 2016) for determining a client-specific set of encoder parameters using an adjustment network based at least in part on observed environmental vectors. For example, apparatus 2002 may include circuit system 2016 for causing apparatus 2002 to determine a client-specific set of encoder parameters using an adjustment network based at least in part on observed environmental vectors.

[0280] like Figure 20 As further shown, apparatus 2002 may include a circuit system (circuit system 2018) for determining a latent vector using an autoencoder and at least in part based on a client-specific set of encoder parameters. For example, apparatus 2002 may include a circuit system 2018 for determining a latent vector using an autoencoder and at least in part based on a client-specific set of encoder parameters.

[0281] like Figure 20 As further shown, the apparatus 2002 may include a circuit system (circuit system 2020) for transmitting the observed environment vector and the latent vector. For example, the apparatus 2002 may include a circuit system 2020 for enabling the apparatus to transmit the observed environment vector and the latent vector.

[0282] like Figure 20As further shown, apparatus 2002 may include code (code 2022) stored in computer-readable medium 2010 for determining a client-specific encoder parameter set using an adjustment network based at least in part on observed environment vectors. For example, apparatus 2002 may include code 2022, which, when executed by processor 2008, enables processor 2008 to determine a client-specific encoder parameter set using an adjustment network based at least in part on observed environment vectors.

[0283] like Figure 20 As further shown, apparatus 2002 may include code (code 2024) stored in computer-readable medium 2010 for determining latent vectors using an autoencoder and based at least in part on a client-specific set of encoder parameters. For example, apparatus 2002 may include code 2024, which, when executed by processor 2008, enables processor 2008 to determine latent vectors using an autoencoder and based at least in part on a client-specific set of encoder parameters.

[0284] like Figure 20 As further shown, the apparatus 2002 may include code (code 2026) stored in the computer-readable medium 2010 for transmitting the observed environment vector and the latent vector. For example, the apparatus 2002 may include code 2026, which, when executed by the processor 2008, causes the transceiver 2012 to transmit the observed environment vector and the latent vector.

[0285] Figure 20 This is provided as an example. Other examples may differ from this combination. Figure 20 The example described.

[0286] Figure 21 This is a diagram illustrating an example 2100 of the code and circuitry implementation of a device 2102 for wireless communication. Device 2102 may be, similar to, include, or be included in... Figure 19 The equipment shown is from 1902. Figure 18 As shown in equipment 1800, etc. In some respects, equipment 2102 may be, similar to, include, or be included in the server (e.g., Figure 7 The server 705 shown Figure 6 The server 610 shown Figure 3 (e.g., server 304 shown). Equipment 2102 may include a processing system 2104, which may include a bus 2112 coupling one or more components (e.g., processor 2106, computer-readable medium / memory 2108, transceiver 2110, etc.). As shown, transceiver 2112 may be coupled to one or more antennas 2114.

[0287] like Figure 21 As further shown, apparatus 2102 may include a circuit system (circuit system 2116) for receiving from a client an observed environment vector associated with one or more features, the one or more features being associated with the client's environment. For example, apparatus 2102 may include circuit system 2116 for enabling apparatus 2102 to receive from a client an observed environment vector associated with one or more features, the one or more features being associated with the client's environment.

[0288] like Figure 21 As further shown, apparatus 2102 may include a circuit system (circuit system 2118) for receiving from the client a latent vector that is at least partially based on a client-specific set of encoder parameters. For example, apparatus 2102 may include a circuit system for enabling apparatus 2102 to receive from the client a latent vector that is at least partially based on a client-specific set of encoder parameters.

[0289] like Figure 21 As further shown, the apparatus 2102 may include a circuit system (circuit system 2120) for determining the observed wireless communication vector based at least in part on the latent vector and the observed environment vector. For example, the apparatus 2102 may include a circuit system 2120 for enabling the apparatus 2102 to determine the observed wireless communication vector based at least in part on the latent vector and the observed environment vector.

[0290] like Figure 21 As further shown, device 2102 may include a circuit system (circuit system 2122) for performing wireless communication actions at least in part based on a determined observed wireless communication vector. For example, device 2102 may include a circuit system 2120 for causing device 2102 to perform wireless communication actions at least in part based on a determined observed wireless communication vector.

[0291] like Figure 21 As further shown, apparatus 2102 may include code (code 2124) stored in computer-readable medium 2110 for receiving from a client an observed environment vector associated with one or more features, which are associated with the client's environment. For example, apparatus 2102 may include code 2124, which, when executed by processor 2108, causes transceiver 2112 to receive from a client an observed environment vector associated with one or more features, which are associated with the client's environment.

[0292] like Figure 21As further shown, apparatus 2102 may include code (code 2126) stored in computer-readable medium 2110 for receiving, from the client, a latent vector based at least in part on a client-specific set of encoder parameters. For example, apparatus 2102 may include code 2126, which, when executed by processor 2108, causes transceiver 2112 to receive, from the client, a latent vector based at least in part on a client-specific set of encoder parameters.

[0293] like Figure 21 As further shown, apparatus 2102 may include code (code 2128) stored in computer-readable medium 2110 for determining an observed wireless communication vector based at least in part on the latent vector and the observed environment vector. For example, apparatus 2102 may include code 2128, which, when executed by processor 2108, enables processor 2108 to determine the observed wireless communication vector based at least in part on the latent vector and the observed environment vector.

[0294] like Figure 21 As further shown, the apparatus 2102 may include code (code 2130) stored in the computer-readable medium 2110 for performing wireless communication actions at least in part based on a determined observed wireless communication vector. For example, the apparatus 2102 may include code 2130, which, when executed by the processor 2108, enables the processor 2108 to perform wireless communication actions at least in part based on a determined observed wireless communication vector.

[0295] Figure 21 This is provided as an example. Other examples may differ from this combination. Figure 21 The example described.

[0296] The following provides an overview of some aspects of this disclosure:

[0297] Aspect 1: A wireless communication method performed by a client, comprising: determining a client-specific set of encoder parameters using an conditioning network based at least in part on an observed environment vector; determining a latent vector using an autoencoder based at least in part on the client-specific set of encoder parameters; and transmitting the observed environment vector and the latent vector.

[0298] Aspect 2: The method of aspect 1, wherein the conditioning network is configured to receive an observed environment vector as input and provide a conditioning vector as output, wherein the conditioning vector includes: a client-specific set of encoder parameters and a client-specific set of decoder parameters.

[0299] Aspect 3: The method of aspect 2, wherein the adjustment network includes: a first subnetwork configured to determine a client-specific set of encoder parameters; and a second subnetwork configured to determine a client-specific set of decoder parameters.

[0300] Aspect 4: A method of any of Aspects 1-3, wherein the autoencoder comprises: an encoder configured to receive an observed wireless communication vector as input and provide the latent vector as output; and a decoder configured to receive the latent vector as input and provide the observed wireless communication vector as output.

[0301] Aspect 5: The method of aspect 4, wherein the encoder includes at least one encoder layer configured to use a client-specific set of encoder parameters, and wherein determining the latent vector includes: using the at least one encoder layer to map the observed wireless communication vector to the latent vector.

[0302] Aspect 6: The method of aspect 5, wherein mapping the observed wireless communication vector to the latent vector includes: using an encoder parameter set to map the observed wireless communication vector to the latent vector, wherein the encoder parameter set includes: a client-specific encoder parameter set and a shared encoder parameter set.

[0303] Aspect 7: The method of aspect 6, wherein the at least one encoder layer comprises: a first layer set which uses a first subset of a client-specific encoder parameter set; a second layer set which uses a first subset of a shared encoder parameter set; and a third layer set which uses a second subset of both the client-specific encoder parameter set and the shared encoder parameter set.

[0304] Aspect 8: A method as in any of Aspects 4-1, wherein the decoder includes at least one decoder layer configured to use a client-specific set of decoder parameters, wherein the decoder is configured to determine the observed wireless communication vector by mapping the latent vector to the observed wireless communication vector using the at least one decoder layer.

[0305] Aspect 9: The method of aspect 8, wherein mapping the latent vector to the observed wireless communication vector includes: using a decoder parameter set to map the latent vector to the observed wireless communication vector, wherein the decoder parameter set includes: a client-specific decoder parameter set and a shared decoder parameter set.

[0306] Aspect 10: The method of aspect 9, wherein the at least one decoder layer comprises: a first set of layers that uses a first subset of a client-specific decoder parameter set; a second set of layers that uses a first subset of a shared decoder parameter set; and a third set of layers that uses a second subset of both the client-specific decoder parameter set and the shared decoder parameter set.

[0307] Aspect 11: The method of any of Aspects 1-10, wherein the observed environment vector includes one or more feature components, wherein the one or more feature components indicate: client vendor identifier, client antenna configuration, large-scale channel characteristics, channel state information reference signal configuration, image obtained by an imaging device, a portion of the estimated propagation channel, or a combination thereof.

[0308] Aspect 12: The method of aspect 11, wherein the large-scale channel characteristics indicate: channel-associated delay spread, channel-associated power delay profile, channel-associated Doppler measurement, channel-associated Doppler spectrum, channel-associated signal-to-noise ratio, channel-associated signal-to-noise plus-interference ratio, reference signal received power, received signal strength indicator, or a combination thereof.

[0309] Aspect 13: The method of any of Aspects 1-12, wherein the implicit vector is associated with a wireless communication task.

[0310] Aspect 14: The method of aspect 13, wherein the wireless communication task includes: determining channel state feedback (CSF), determining location information associated with the client, determining modulation associated with wireless communication, determining waveform associated with wireless communication, or a combination thereof.

[0311] Aspect 15: The method of aspect 14, wherein the wireless communication task includes determining the CSF, and wherein the method further includes: receiving a channel state information (CSI) reference signal (CSI-RS); determining the CSI based at least in part on the CSI-RS; and providing the CSI as input to the autoencoder.

[0312] Aspect 16: The method of aspect 15, wherein the implicit vector includes compressed channel state feedback.

[0313] Aspect 17: A method of any of Aspects 1-16, wherein transmitting the observed environment vector and the implicit vector comprises: using a physical uplink control channel, a physical uplink shared channel, or a combination thereof to transmit the observed environment vector and the implicit vector.

[0314] Aspect 18: The method of any of Aspects 1-17, wherein the autoencoder includes a variational autoencoder.

[0315] Aspect 19: The method of any of Aspects 1-18 further includes: training at least one of the conditioning network or the autoencoder.

[0316] Aspect 20: The method of aspect 19, wherein training at least one of the conditioning network or the autoencoder includes: using an unsupervised learning procedure.

[0317] Aspect 21: The method of any of Aspects 1-20 further includes: using a joint learning procedure to collaboratively train the regulation network and the autoencoder.

[0318] Aspect 22: The method of aspect 21, wherein cooperatively training the conditioning network and the autoencoder includes: determining a shared set of neural network parameters that maximizes the variational lower bound function corresponding to the conditioning network and the autoencoder.

[0319] Aspect 23: The method of aspect 22, where the negative variational lower bound function corresponds to the loss function.

[0320] Aspect 24: The method of aspect 23, wherein the loss function includes the reconstruction loss of the autoencoder.

[0321] Aspect 25: The method of either Aspect 23 or 24, wherein the loss function includes a regularization term for the autoencoder.

[0322] Aspect 26: The method of any of Aspects 23-25, wherein the autoencoder is a regular autoencoder and wherein the negative variational lower bound function does not include a regularization term.

[0323] Aspect 27: The method of any of Aspects 21-26, wherein cooperating in training the modulating network and the autoencoder comprises: jointly training the modulating network and the autoencoder.

[0324] Aspect 28: The method of any of Aspects 21-26, wherein training the modulator network and the autoencoder collaboratively includes alternating between training the modulator network and training the autoencoder.

[0325] Aspect 29: The method of aspect 28, wherein alternating between training the conditioning network and training the autoencoder includes: performing a first plurality of training iterations associated with the conditioning network according to a first training frequency; and performing a second plurality of training iterations associated with the autoencoder according to a second training frequency, the second training frequency being different from the first training frequency.

[0326] Aspect 30: A method of any of Aspects 21-29, wherein collaboratively training the conditioning network and the autoencoder comprises: receiving from a server a shared set of neural network parameters corresponding to the conditioning network and the autoencoder; obtaining observed environment training vectors; inputting the observed environment training vectors into the conditioning network to determine conditioning training vectors, which include client-specific encoder training parameter sets and client-specific decoder training parameter sets; loading at least one layer of the encoder of the autoencoder using the client-specific encoder training parameter sets; loading at least one layer of the decoder of the autoencoder using the client-specific decoder training parameter sets; obtaining observed wireless communication training vectors; inputting the observed wireless communication training vectors into the encoder to determine training latent vectors; inputting the training latent vectors into the decoder to determine training output of the autoencoder, wherein the training output includes a reconstruction of the observed wireless communication training vectors; and determining a loss associated with the autoencoder based at least in part on the training output.

[0327] Aspect 31: The method of aspect 30 further includes: determining a regularization term corresponding to the autoencoder, wherein determining the loss includes: determining the loss based at least in part on the regularization term.

[0328] Aspect 32: The method of any of Aspects 30 or 31 further includes: determining a plurality of gradients of the loss relative to a shared neural network parameter set; and updating the shared neural network parameter set at least in part based on the plurality of gradients.

[0329] Aspect 33: The method of aspect 32 further includes: updating the shared neural network parameter set a specified number of times to determine the locally updated shared neural network parameter set.

[0330] Aspect 34: The method of aspect 33 further includes: transmitting a locally updated set of shared neural network parameters to the server.

[0331] Aspect 35: The method of any of Aspects 30-34 further includes: determining a plurality of gradients of the loss relative to a subset of a shared neural network parameter set, wherein the subset includes a shared regulation network parameter set; and updating the shared regulation network parameter set at least in part based on the plurality of gradients.

[0332] Aspect 36: The method of aspect 35 further includes: updating the shared regulation network parameter set a specified number of times to determine a locally updated shared regulation network parameter set.

[0333] Aspect 37: The method of aspect 36 further includes: transmitting a locally updated set of shared adjustment network parameters to the server.

[0334] Aspect 38: The method of any of Aspects 30-37 further includes: determining a plurality of gradients of the loss relative to a subset of a shared neural network parameter set, wherein the subset includes a shared autoencoder parameter set; and updating the shared autoencoder parameter set at least in part based on the plurality of gradients.

[0335] Aspect 39: The method of aspect 38 further includes: updating the shared autoencoder parameter set a specified number of times to determine the locally updated shared autoencoder parameter set.

[0336] Aspect 40: The method of aspect 39 further includes: transmitting a locally updated set of shared autoencoder parameters to the server.

[0337] Aspect 41: The method of any of Aspects 1-40, wherein the regulation network corresponds to a server regulation network implemented at the server.

[0338] Aspect 42: The method of aspect 41, wherein the regulating network is a copy of the client implementation of the server regulating network.

[0339] Aspect 43: The method of any of Aspects 1-42, wherein the autoencoder corresponds to a server autoencoder implemented at the server.

[0340] Aspect 44: The method of aspect 43, wherein the autoencoder is a copy of the client implementation of the server autoencoder.

[0341] Aspect 45: A wireless communication method performed by a server, comprising: receiving from a client an observed environment vector associated with one or more features, the one or more features being associated with an environment of the client; receiving from the client a latent vector based at least in part on a client-specific set of encoder parameters; determining an observed wireless communication vector based at least in part on the latent vector and the observed environment vector; and performing a wireless communication action based at least in part on the determined observed wireless communication vector.

[0342] Aspect 46: The method of aspect 45, wherein the observed environment vector includes one or more feature components, wherein the one or more feature components indicate: client vendor identifier, client antenna configuration, large-scale channel characteristics, channel state information reference signal configuration, image obtained by an imaging device, a portion of the estimated propagation channel, or a combination thereof.

[0343] Aspect 47: The method of aspect 46, wherein the large-scale channel characteristics indicate: channel-associated delay spread, channel-associated power delay profile, channel-associated Doppler measurement, channel-associated Doppler spectrum, channel-associated signal-to-noise ratio, channel-associated signal-to-noise plus-interference ratio, reference signal received power, received signal strength indicator, or a combination thereof.

[0344] Aspect 48: The method of any of Aspects 45-47, wherein the implicit vector is associated with the wireless communication task.

[0345] Aspect 49: The method of aspect 48, wherein the wireless communication task includes: receiving channel state feedback (CSF), receiving location information associated with the client, receiving modulation associated with wireless communication, receiving waveforms associated with wireless communication, or a combination thereof.

[0346] Aspect 50: The method of aspect 49, wherein the wireless communication task includes: receiving the CSF, and wherein the method further includes: transmitting a channel state information reference signal (CSI-RS), wherein the CSF corresponds to the CSI-RS.

[0347] Aspect 51: The method of any of Aspects 45-50, wherein the latent vector comprises a compressed CSF.

[0348] Aspect 52: The method of any of Aspects 45-51, wherein the observed environment vector and the implicit vector are carried in a physical uplink control channel, a physical uplink shared channel, or a combination thereof.

[0349] Aspect 53: A method of any of Aspects 45-52, wherein determining the observed wireless communication vector comprises: inputting the observed environment vector into a server conditioning network to determine a conditioning vector comprising a client-specific set of decoder parameters; loading at least one layer of a decoder of a server autoencoder using the client-specific set of decoder parameters; and inputting the implicit vector into the decoder to determine the observed wireless communication vector.

[0350] Aspect 54: The method of aspect 53, wherein the server autoencoder includes a regular autoencoder or a variational autoencoder.

[0351] Aspect 55: A method as in any of Aspects 53 or 54, wherein the server conditioning network is configured to receive an observed environment vector as input and provide a conditioning vector as output, wherein the conditioning vector includes: a client-specific set of decoder parameters and a client-specific set of encoder parameters.

[0352] Aspect 56: The method of aspect 55, wherein the server conditioning network includes: a first subnetwork configured to determine a client-specific set of encoder parameters; and a second subnetwork configured to determine a client-specific set of decoder parameters.

[0353] Aspect 57: A method of any of Aspects 53-56, wherein the server autoencoder comprises: an encoder configured to receive an observed wireless communication vector as input and provide the implicit vector as output; and a decoder configured to receive the implicit vector as input and provide the observed wireless communication vector as output.

[0354] Aspect 58: The method of aspect 57, wherein the encoder includes at least one encoder layer configured to use a set of encoder parameters that varies from client to client.

[0355] Aspect 59: A method as in any of Aspects 57 or 58, wherein the decoder includes at least one decoder layer configured to use a client-specific set of decoder parameters, wherein the decoder is configured to determine the observed wireless communication vector by mapping the latent vector to the observed wireless communication vector using the at least one decoder layer.

[0356] Aspect 60: The method of aspect 59, wherein mapping the latent vector to the observed wireless communication vector includes: using a decoder parameter set to map the latent vector to the observed wireless communication vector, wherein the decoder parameter set includes: a client-specific decoder parameter set and a shared decoder parameter set.

[0357] Aspect 61: The method of aspect 60, wherein the at least one decoder layer comprises: a first layer set which uses a first subset of a client-specific decoder parameter set; a second layer set which uses a first subset of a shared decoder parameter set; and a third layer set which uses a second subset of both the client-specific decoder parameter set and the shared decoder parameter set.

[0358] Aspect 62: The method of any of Aspects 53-61, wherein the server regulation network corresponds to the regulation network implemented at the client.

[0359] Aspect 63: The method of aspect 62, wherein the server regulating network is a copy of the server implementation of the regulating network.

[0360] Aspect 64: The method of any of Aspects 53-63, wherein the server autoencoder corresponds to the autoencoder implemented at the client.

[0361] Aspect 65: The method of aspect 64, wherein the server autoencoder is a copy of the server implementation of the autoencoder.

[0362] Aspect 66: The method of any of Aspects 53-65, wherein the server autoencoder includes a variational autoencoder.

[0363] Aspect 67: The method of any of Aspects 53-66 further includes: updating at least one of the server conditioning network or the server autoencoder.

[0364] Aspect 68: The method of any of Aspects 53-67 further includes: using a joint learning procedure to collaboratively update the server conditioning network and the server autoencoder.

[0365] Aspect 69: The method of aspect 68, wherein collaboratively updating the server conditioning network and the server autoencoder comprises: jointly training the server conditioning network and the server autoencoder.

[0366] Aspect 70: The method of any of Aspects 68 or 69, wherein cooperatingly updating the server conditioning network and the server autoencoder comprises alternating between updating the server conditioning network and updating the server autoencoder.

[0367] Aspect 71: The method of aspect 70, wherein alternating between updating the server conditioning network and updating the server autoencoder includes: performing a first plurality of update iterations associated with the server conditioning network according to a first update frequency; and performing a second plurality of update iterations associated with the server autoencoder according to a second update frequency, the second update frequency being different from the first update frequency.

[0368] Aspect 72: A method of any of Aspects 68-71, wherein collaboratively updating the server conditioning network and the server autoencoder comprises: selecting a set of clients from which to receive updates associated with the server conditioning network and the server autoencoder, wherein the set of clients includes the client and the at least one additional client; transmitting to the set of clients a shared neural network parameter set corresponding to the server conditioning network and the server autoencoder; receiving a plurality of locally updated shared neural network parameter sets from the set of clients; and determining a server-updated shared neural network parameter set based at least in part on the plurality of locally updated shared neural network parameter sets.

[0369] Aspect 73: The method of aspect 72 further includes: transmitting the server-updated shared neural network parameter set to the client set.

[0370] Aspect 74: The method of any of Aspects 72 or 73, wherein determining the shared neural network parameter set updated by the server includes: averaging multiple locally updated shared neural network parameter sets.

[0371] Aspect 75: A method of any of Aspects 68-74, wherein collaboratively updating the server conditioning network and the server autoencoder comprises: selecting a set of clients from which to receive updates associated with the server conditioning network, wherein the set of clients includes the client and the at least one additional client; transmitting to the set of clients a shared neural network parameter set corresponding to the server conditioning network and the server autoencoder; receiving a plurality of locally updated shared conditioning network parameter sets from the set of clients; and determining a server-updated shared conditioning network parameter set based at least in part on the plurality of locally updated shared conditioning network parameter sets.

[0372] Aspect 76: The method of aspect 75 further includes: transmitting a server-updated set of shared adjustment network parameters to the client set.

[0373] Aspect 77: The method of any of Aspects 75 or 76, wherein determining the shared conditioning network parameter set for server updates includes averaging the shared conditioning network parameter sets for multiple local updates.

[0374] Aspect 78: A method of any of Aspects 68-77, wherein collaboratively updating the server conditioning network and the server autoencoder comprises: selecting a set of clients from which to receive updates associated with the server autoencoder, wherein the set of clients includes the client and the at least one additional client; transmitting to the set of clients a shared neural network parameter set corresponding to the server conditioning network and the server autoencoder; receiving a plurality of locally updated shared autoencoder parameter sets from the set of clients; and determining a server-updated shared autoencoder parameter set based at least in part on the plurality of locally updated shared autoencoder parameter sets.

[0375] Aspect 79: The method of aspect 78 further includes: transmitting a server-updated set of shared autoencoder parameters to the client set.

[0376] Aspect 80: The method of any of Aspects 78 or 79, wherein determining the shared autoencoder parameter set of server updates includes: averaging the shared autoencoder parameter sets of multiple local updates.

[0377] Aspect 81: An apparatus for wireless communication at a device, comprising: a processor; a memory coupled to the processor; and instructions stored in the memory and executable by the processor to cause the apparatus to perform one or more of the methods of aspects 1-44.

[0378] Aspect 82: An apparatus for wireless communication, comprising a memory and one or more processors coupled to the memory, the one or more processors being configured to perform one or more of the methods of aspects 1-44.

[0379] Aspect 83: An apparatus for wireless communication, comprising at least one means for performing one or more of the methods of aspects 1-44.

[0380] Aspect 84: A non-transient computer-readable medium storing code for wireless communication, the code including instructions executable by a processor to perform one or more of the methods of aspects 1-44.

[0381] Aspect 85: A non-transient computer-readable medium storing a set of instructions for wireless communication, the set of instructions including one or more instructions which, when executed by one or more processors of a device, cause the device to perform one or more methods as described in aspects 1-44.

[0382] Aspect 86: An apparatus for wireless communication at a device, comprising: a processor; a memory coupled to the processor; and instructions stored in the memory and executable by the processor to cause the apparatus to perform one or more of the methods of aspects 45-80.

[0383] Aspect 87: An apparatus for wireless communication, comprising a memory and one or more processors coupled to the memory, the one or more processors being configured to perform one or more of the methods of aspects 45-80.

[0384] Aspect 88: An apparatus for wireless communication, comprising at least one means for performing one or more methods as described in aspects 45-80.

[0385] Aspect 89: A non-transient computer-readable medium storing code for wireless communication, the code including instructions executable by a processor to perform one or more methods as described in aspects 45-80.

[0386] Aspect 90: A non-transient computer-readable medium storing a set of instructions for wireless communication, the set of instructions including one or more instructions which, when executed by one or more processors of a device, cause the device to perform one or more methods as described in aspects 45-80.

[0387] The foregoing disclosure provides explanations and descriptions, but is not intended to be exhaustive or to limit the aspects to the precise forms disclosed. Modifications and variations may be made in light of the foregoing disclosure or may be obtained through practice.

[0388] As used herein, the term "component" is intended to be interpreted broadly as hardware, firmware, and / or a combination of hardware and software. As used herein, a processor is implemented using hardware, firmware, and / or a combination of hardware and software. It will be apparent that the systems and / or methods described herein can be implemented in various forms as hardware, firmware, and / or combinations of hardware and software. The actual dedicated control hardware or software code used to implement these systems and / or methods is not limiting in any way. Therefore, the operation and behavior of these systems and / or methods are described herein without reference to any specific software code—it is understood that software and hardware can be designed to implement these systems and / or methods, at least in part, based on the descriptions herein.

[0389] As used in this article, depending on the context, a threshold can refer to a value greater than the threshold, greater than or equal to the threshold, less than the threshold, less than or equal to the threshold, equal to the threshold, not equal to the threshold, etc.

[0390] Although specific combinations of features are described in the claims and / or disclosed in the specification, these combinations are not intended to limit the disclosure of aspects. In fact, many of these features can be combined in ways not specifically described in the claims and / or not disclosed in the specification. Although each dependent claim listed below may be directly subordinated to only one claim, the disclosure of aspects includes each dependent claim being combined with each other claim in this set of claims. As used herein, the phrase “at least one of” refers to any combination of these items, including single members. As an 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 having multiple identical elements (e.g., aa, aaa, aab, aac, abb, acc, bb, bbb, bbc, cc, and ccc, or any other ordering of a, b, and c).

[0391] The elements, actions, or instructions used herein should not be construed as critical or necessary unless explicitly stated otherwise. Furthermore, as used herein, the articles “a” and “a certain” are intended to include one or more items and may be used interchangeably with “one or more.” Additionally, as used herein, the article “the” is intended to include one or more items referenced in conjunction with the article “the” and may be used interchangeably with “one or more.” Furthermore, as used herein, the terms “set” and “group” are intended to include one or more items (e.g., related items, unrelated items, or a combination of related and unrelated items) and may be used interchangeably with “one or more.” Where only one item is intended, the phrase “only one” or similar language is used. Moreover, as used herein, the terms “have,” “contain,” “include,” etc., are intended to be open-ended terms. Additionally, the phrase “based on” is intended to mean “at least partially based on” unless otherwise explicitly stated. Moreover, as used herein, the term “or” is intended to be inclusive when used in a sequence and may be used interchangeably with “and / or” unless otherwise explicitly stated (e.g., in combination with “either of” or “only one of”).

Claims

1. A wireless communication method performed by a client, the client including an autoencoder and an associated conditioning network, the method comprising: The conditioning network is used to determine a client-specific set of encoder parameters based at least in part on observed environment vectors, which include information about the client's environment. The latent vector is determined using the autoencoder and at least in part based on the client-specific encoder parameter set and observed wireless communication vectors associated with the wireless communication task; and Transmit the observed environment vector and the implicit vector.

2. The method of claim 1, wherein the conditioning network is configured to receive the observed environment vector as input and provide a conditioning vector as output, wherein the conditioning vector comprises: The encoder parameter set that varies from client to client, and Decoder parameter sets vary depending on the client.

3. The method of claim 2, wherein the regulating network comprises: The first sub-network is configured to determine the client-specific set of encoder parameters; as well as The second sub-network is configured to determine the client-specific set of decoder parameters.

4. The method of claim 1, wherein the self-encoder comprises: An encoder configured to receive the observed wireless communication vector as input and provide the implicit vector as output; as well as The decoder is configured to receive the latent vector as input and provide the observed wireless communication vector as output.

5. The method of claim 4, wherein the encoder comprises at least one encoder layer configured to use the client-specific encoder parameter set, and wherein determining the latent vector comprises: The at least one encoder layer is used to map the observed wireless communication vector to the latent vector.

6. The method of claim 5, wherein mapping the observed wireless communication vector to the latent vector comprises: The observed wireless communication vectors are mapped to the latent vectors using an encoder parameter set. The encoder parameter set includes: The encoder parameter set that varies from client to client, and Shared encoder parameter set.

7. The method of claim 6, wherein the at least one encoder layer comprises: The first layer set uses a first subset of the client-specific encoder parameter set; The second layer set uses a first subset of the shared encoder parameter set; as well as The third layer set uses a second subset of the client-specific encoder parameter set and a second subset of the shared encoder parameter set.

8. The method of claim 4, wherein the decoder comprises at least one decoder layer configured to use a client-specific set of decoder parameters. The decoder is configured to determine the observed wireless communication vector by mapping the latent vector to the observed wireless communication vector using the at least one decoder layer.

9. The method of claim 8, wherein mapping the latent vector to the observed wireless communication vector comprises: The decoder parameter set is used to map the latent vector to the observed wireless communication vector. The decoder parameter set includes: The client-specific decoder parameter set, and Shared decoder parameter set.

10. The method of claim 9, wherein the at least one decoder layer comprises: The first layer set uses a first subset of the client-specific decoder parameter set; The second layer set uses a first subset of the shared decoder parameter set; as well as The third layer set uses a second subset of the client-specific decoder parameter set and a second subset of the shared decoder parameter set.

11. The method of claim 1, wherein the observed environment vector comprises one or more feature components, wherein the one or more feature components indicate: Client vendor identifier, Client antenna configuration, Large-scale channel characteristics Channel state information reference signal configuration, Images obtained through imaging equipment A portion of the estimated propagation channel, or Its combination.

12. The method of claim 11, wherein the large-scale channel characteristics indicate: Delay spread associated with the channel, Channel-associated power delay profile Doppler measurements associated with the channel, The Doppler spectrum associated with the channel, Signal-to-noise ratio associated with the channel, The signal-to-noise-plus-interference ratio associated with the channel. The power received by the reference signal, Received signal strength indicator, or Its combination.

13. The method of claim 1, wherein the latent vector is associated with the wireless communication task.

14. The method of claim 13, wherein the wireless communication task includes: Determine the Channel State Feedback (CSF). Determine the location information associated with the client. Determine the modulation associated with wireless communication. Determine the waveform associated with wireless communication, or Its combination.

15. The method of claim 14, wherein the wireless communication task includes determining the CSF, and wherein the method further comprises: Receive Channel State Information (CSI) Reference Signal (CSI-RS); The CSI is determined at least in part based on the CSI-RS; as well as The CSI is provided as input to the autoencoder.

16. The method of claim 15, wherein the implicit vector includes compressed channel state feedback.

17. The method of claim 1, wherein transmitting the observed environment vector and the latent vector comprises: Use the following to transmit the observed environment vector and the latent vector: Physical uplink control channel, Physical uplink shared channel, or Its combination.

18. The method of claim 1, wherein the autoencoder comprises a variational autoencoder.

19. The method of claim 1, further comprising: Training at least one of the conditioning network or the autoencoder, wherein training at least one of the conditioning network or the autoencoder includes: using an unsupervised learning procedure.

20. The method of claim 1, further comprising: The regulation network and the autoencoder are trained collaboratively using a joint learning procedure.

21. The method of claim 20, wherein cooperatively training the conditioning network and the autoencoder comprises: Determine a shared neural network parameter set that maximizes the variational lower bound function corresponding to the conditioning network and the autoencoder.

22. The method of claim 21, wherein the negative variational lower bound function corresponds to the loss function.

23. The method of claim 20, wherein cooperatively training the regulatory network and the autoencoder comprises: The conditioning network and the autoencoder are trained jointly.

24. The method of claim 20, wherein cooperatively training the conditioning network and the autoencoder comprises: The training alternates between training the conditioning network and training the autoencoder.

25. The method of claim 24, wherein alternating between training the conditioning network and training the autoencoder comprises: A first set of multiple training iterations associated with the conditioning network are performed according to a first training frequency; as well as A second set of multiple training iterations associated with the autoencoder is performed based on a second training frequency, which is different from the first training frequency.

26. The method of claim 1, wherein the regulation network corresponds to a server regulation network implemented at the server, and wherein the regulation network is a client-side copy of the server regulation network.

27. The method of claim 1, wherein the autoencoder corresponds to a server autoencoder implemented at the server, and wherein the autoencoder is a client-side copy of the server autoencoder.

28. A wireless communication method executed by a server, comprising: Receive from the client an observed environment vector associated with one or more features, the features being associated with the client's environment, the observed environment vector including information about the client's environment; Receive implicit vectors from the client, which are at least partially based on a client-specific set of encoder parameters; The observed wireless communication vector is determined at least in part based on the latent vector and the observed environment vector, the observed wireless communication vector being associated with the wireless communication task; and Wireless communication actions are performed, at least in part, based on determining the observed wireless communication vector.

29. An apparatus for wireless communication at a client, the client including a self-encoder and an associated conditioning network, the apparatus comprising: Memory; as well as One or more processors coupled to the memory, the one or more processors being configured to: The conditioning network is used to determine a client-specific set of encoder parameters based at least in part on observed environment vectors associated with one or more features of the environment related to the client. The latent vectors are determined using the autoencoder and at least in part based on the client-specific encoder parameter set; and Transmit the observed environment vector and the implicit vector.

30. An apparatus for wireless communication at a client, the client including a self-encoder and an associated conditioning network, the apparatus comprising: Memory; as well as One or more processors coupled to the memory, the one or more processors being configured to perform the method as described in any one of claims 2-27.

31. An apparatus for wireless communication at a server, comprising: Memory; as well as One or more processors coupled to the memory, the one or more processors being configured to: Receive from the client an observed environment vector associated with one or more features, the features being associated with the client's environment, the observed environment vector including information about the client's environment; Receive implicit vectors from the client, which are at least partially based on a client-specific set of encoder parameters; The observed wireless communication vector is determined at least in part based on the latent vector and the observed environment vector, and the observed wireless communication vector is associated with the wireless communication task. as well as Wireless communication actions are performed, at least in part, based on determining the observed wireless communication vector.

Images