Vector quantization method for UE-driven multi-supplier sequential training

Abstract

An entity associated with a UE may train an encoder to encode uplink control information. An entity associated with the UE may determine a quantization codebook to be applied to the encoded uplink control information. The entity associated with the UE may share a sequential training data set with the entity associated with the base station, the sequential training data set comprising one of a set of input vectors or a set of output vectors; and one of a set of encoded and unquantized intermediate vectors or a set of encoded and quantized intermediate vectors. The base station associated entity may train a decoder based on a quantized codebook from the UE associated entity and at least the sequential training data set. When the base station associated entity receives a plurality of sequential training data sets for different vendors, the base station associated entity may train a multi-vendor decoder based on the plurality of sequential training data sets.

Description

Background Art Technical Field

[0001] The present disclosure generally relates to communication systems and, more particularly, to a vector quantization method for UE-driven multi-vendor sequential training.

[0002] Introduction

[0003] Wireless communication systems are widely deployed to provide various telecommunication services such as telephony, video, data, messaging, and broadcasting. A typical wireless communication system may employ a multiple access technology capable of supporting communication with multiple users by sharing available system resources. Examples of such multiple access technologies include code division multiple access (CDMA) systems, time division multiple access (TDMA) systems, frequency division multiple access (FDMA) systems, orthogonal frequency division multiple access (OFDMA) systems, single carrier frequency division multiple access (SC-FDMA) systems, and time division synchronous code division multiple access (TD-SCDMA) systems.

[0004] These multiple access technologies have been adopted in various telecommunication standards to provide a common protocol that enables different wireless devices to communicate at the urban, national, regional, and even global levels. An example telecommunication standard is 5G New Radio (NR). 5G NR is part of the continuous mobile broadband evolution promulgated by the 3rd Generation Partnership Project (3GPP) to meet new requirements associated with latency, reliability, security, scalability (e.g., with the Internet of Things (IoT)), and other requirements. 5G NR includes services associated with enhanced mobile broadband (eMBB), massive machine type communication (mMTC), and ultra-reliable low latency communication (URLLC). Certain aspects of 5G NR may be based on the 4G Long Term Evolution (LTE) standard. Further improvements to 5G NR technology are needed. Additionally, these improvements may also be applicable to other multiple access technologies and telecommunication standards that employ these technologies. Summary of the Invention

[0005] A simplified summary of one or more aspects is presented below to provide a basic understanding of these aspects. This summary of the invention is not an extensive overview of all contemplated aspects, and is neither intended to identify key or critical elements of all aspects nor to describe the scope of any or all aspects. Its sole purpose is to present some concepts of one or more aspects in a simplified form as a prelude to the more detailed description that is presented later.

[0006] In some aspects, the techniques described herein relate to a method for wireless communication of an entity associated with a user equipment (UE), including: training an encoder to encode uplink control information; determining a quantization codebook to be applied to the encoded uplink control information; and sharing an ordered training dataset with an entity associated with a base station, the ordered training dataset including: one of an input vector set or an output vector set; and one of an encoded and unquantized intermediate vector set or an encoded and quantized intermediate vector set.

[0007] The present disclosure also provides an apparatus (e.g., an entity associated with a UE, such as a UE or a server) including a memory storing computer-executable instructions and at least one processor configured to execute these computer-executable instructions to perform the above method, an apparatus including components for performing the above method, and a non-transitory computer-readable medium storing computer-executable instructions for performing the above method.

[0008] An innovative aspect of the subject matter described in the present disclosure may be implemented in a method for wireless communication at an entity associated with a base station (BS), the method including: receiving an ordered training dataset from at least a first entity associated with a user equipment (UE), the ordered training dataset including: one of an input vector set or an output vector set; and one of an encoded and unquantized intermediate vector set or an encoded and quantized intermediate vector set; determining a quantization codebook to be applied to the encoded and quantized uplink control information; and training a decoder to decode the encoded and quantized uplink control information based on the ordered training dataset and the quantization codebook.

[0009] The present disclosure also provides an apparatus (e.g., an entity associated with a BS, such as a BS or a server) including a memory storing computer-executable instructions and at least one processor configured to execute these computer-executable instructions to perform the above method, an apparatus including components for performing the above method, and a non-transitory computer-readable medium storing computer-executable instructions for performing the above method.

[0010] To achieve the foregoing and related purposes, one or more aspects include the features described comprehensively below and particularly pointed out in the claims. The following description and the drawings set forth in detail some illustrative features of one or more aspects. However, these features are only indicative of some of the various ways in which the principles of the various aspects may be employed, and this specification is intended to include all such aspects and their equivalents. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 is a diagram illustrating an example of a wireless communication system including an access network in accordance with certain aspects of the present specification.

[0012] Figure 2A Is a diagram illustrating an example of a first frame according to certain aspects of the present specification.

[0013] Figure 2B Is a diagram illustrating an example of a downlink (DL) channel within a subframe according to certain aspects of the present specification.

[0014] Figure 2C Is a diagram illustrating an example of a second frame according to certain aspects of the present specification.

[0015] Figure 2D Is a diagram illustrating an example of an uplink (UL) channel within a subframe according to certain aspects of the present specification.

[0016] Figure 3 Is a diagram illustrating an example of a base station and a user equipment (UE) in an access network according to certain aspects of the present specification.

[0017] Figure 4 Is a diagram illustrating an example of a decomposed base station architecture.

[0018] Figure 5 Is a message diagram of an example multi-vendor training process.

[0019] Figure 6 Is a diagram of an example quantization and dequantization process.

[0020] Figure 7 Is a diagram of an example multi-vendor sequential training system for quantization training at entities associated with an encoder and a decoder and a UE.

[0021] Figure 8 Is a diagram of an example multi-vendor sequential training system for quantization training at entities associated with an encoder and a decoder and a base station.

[0022] Figure 9 Is a diagram of an example sequential training system for an encoder and a decoder and quantization.

[0023] Figure 10 Is a flowchart of an example method for channel state feedback reporting using a learned dictionary.

[0024] Figure 11 Is a flowchart of an example method for channel state feedback reporting using a learned dictionary. Detailed Description

[0025] The detailed description set forth below in connection with the accompanying drawings is intended as a description of various configurations and is not intended to represent the only configurations in which the concepts described herein may be practiced. To provide a thorough understanding of the various concepts, specific details are included in the detailed description. It will be apparent, however, to one of ordinary skill in the art that these concepts may be practiced without these specific details. In some instances, well-known structures and components are shown in block diagram form to avoid obscuring such concepts. While the following description may focus on 5G NR, the concepts described herein may be applicable to other similar domains, such as LTE, LTE-A, CDMA, GSM, and other wireless technologies.

[0026] In a wireless communication system, channel state feedback (CSF) can be used to determine transmission attributes. For example, a user equipment (UE) can send channel state information (CSI) to a base station. The base station can use the CSI to select downlink transmission attributes. The CSI can also be used to schedule the UE for uplink transmission.

[0027] Multiple-input multiple-output (MIMO) antenna technology can increase the dimensionality of the CSI. For example, the channel between each pair of antennas can vary. Thus, as the number of antennas used in MIMO increases, the overhead of reporting uplink control information such as CSF and / or CSI may also increase. Various techniques have been proposed to reduce CSI overhead, such as codebook-based reporting. However, predefined codebooks can reduce the granularity of CSI information. Another proposal for CSI feedback is to use machine learning algorithms to compress the CSI at the UE and decompress the CSI at the base station. It is expected that such proposals provide a gain in feedback accuracy relative to payload size.

[0028] Training of machine learning-based systems for CSF can pose several problems for real-world communication networks. For example, devices within a wireless network can be manufactured by different vendors, such that these devices operate in different ways, even while complying with rules and / or standards. For example, in the case of CSF, devices can have different antenna combinations or proprietary machine learning models. In one use case, a base station can communicate with UEs from multiple vendors, each vendor having an encoder based on a different machine learning model. Training and deploying models for each different UE vendor and base station vendor pair can be redundant, consuming additional resources and / or increasing complexity. Thus, it may be desirable to train a machine learning model (e.g., a decoder) that operates with encoders from multiple vendors. As another example, machine learning models can be considered proprietary, and vendors may be reluctant to share model details with other vendors. Thus, model training techniques such as joint training and model migration may not be available in a multi-vendor environment.

[0029] In one aspect, the present disclosure provides techniques for using sequential training for an encoder and a decoder with vector quantization. The disclosed techniques may be considered UE-driven because a UE-associated entity (e.g., a UE vendor server or the UE itself) may first train at least the encoder and then provide a training set that allows a base station-associated entity (e.g., a base station vendor server or the base station) to train the decoder. In various embodiments, training for vector quantization and / or inverse quantization may be performed at a UE-associated entity or a base station-associated entity. The content of the training set may be selected based on a level of agreement regarding quantization training between the UE-associated entity and the base station-associated entity.

[0030] Certain aspects of a telecommunications system will now be presented with reference to various apparatuses and methods. These apparatuses and methods will be described in detail below and illustrated in the drawings by various blocks, components, circuits, processes, algorithms, etc. (collectively referred to as "elements"). These elements may be implemented using electronic hardware, computer software, or any combination thereof. Whether such elements are implemented as hardware or software depends upon the particular application and design constraints imposed on the overall system.

[0031] As an example, an element, or any portion of an element, or any combination of elements may be implemented as a "processing system" that includes one or more processors. Examples of processors include a microprocessor, a microcontroller, a graphics processing unit (GPU), a central processing unit (CPU), an application processor, a digital signal processor (DSP), a reduced instruction set computing (RISC) processor, a system on a chip (SoC), a baseband processor, a field programmable gate array (FPGA), a programmable logic device (PLD), a state machine, gated logic, discrete hardware circuits, and other suitable hardware configured to perform the various functions described throughout the present disclosure. One or more processors in the processing system may execute software. Whether referred to as software, firmware, middleware, microcode, hardware description language, or other names, software should be broadly construed to mean instructions, instruction sets, code, code segments, program code, programs, subprograms, software components, applications, software applications, software packages, routines, subroutines, objects, executable files, executing threads, processes, functions, etc.

[0032] Thus, in one or more example embodiments, the described functionality may be implemented using hardware, software, or any combination thereof. If implemented in software, the functionality may be stored or encoded on a computer-readable medium as one or more instructions or code. Computer-readable media includes computer storage media. Storage media can be any available media that can be accessed by a computer. Specifically, non-transitory computer-readable media does not include transitory signals. By way of example and not limitation, such computer-readable media may include random access memory (RAM), read only memory (ROM), electrically erasable programmable ROM (EEPROM), optical disk storage, magnetic disk storage, other magnetic storage devices, combinations of the aforementioned types of computer-readable media, or any other medium capable of storing computer-executable code in the form of instructions or data structures that can be accessed by a computer.

[0033] Figure 1 is a diagram illustrating an example of a wireless communication system and access network 100. The wireless communication system (also referred to as a wireless wide area network (WWAN)) includes base stations 102, UEs 104, an evolved packet core (EPC) 160, and another core network (e.g., a 5G core (5GC) 190). The base stations 102 may include macro cells (high-power cellular base stations) and / or small cells (low-power cellular base stations). Macro cells include base stations. Small cells include femtocells, picocells, and microcells. In one aspect, one or more base stations 102 may communicate with a base station vendor server 106. For example, the base station vendor server 106 may be configured to provide firmware or software updates to the base stations 102. Similarly, one or more UEs 104 may communicate with a UE vendor server 108, which may be configured to provide firmware or software updates to the UEs 104. In some specific implementations, each UE 104 may communicate with a corresponding UE vendor server 108 of the vendor corresponding to the UE 104.

[0034] One or more of UE 104 or UE vendor server 108 may include a UE training component 140 that performs machine learning training for a model and / or codebook of UE 104. UE 104 or UE vendor server 108 may be referred to as a UE-associated entity. UE training component 140 may include an encoder training component 142 configured to train an encoder at the UE-associated entity to encode uplink control information. UE training component 140 may include a quantizer training component 144 configured to determine a quantization codebook to be applied to the encoded uplink control information. UE training component 140 may include a sharing component 146 configured to share an ordered training data set with an entity associated with the base station. The ordered training data set may include: a set of input vectors (V in ) or a set of output vectors (V out ) and either a set of encoded and unquantized intermediate vectors (z e ) or a set of encoded and quantized intermediate vectors (z q ).

[0035] In one aspect, one or more of base stations 102 may include a BS training component 120 that performs ordered machine learning training for a model and / or codebook of base station 102. For example, BS training component 120 may include a data set receiving component 122 configured to receive an ordered training data set from at least a first UE-associated entity. The ordered training data set may include: a set of input vectors (V in ) or a set of output vectors (V out ) and either a set of encoded and unquantized intermediate vectors (z e ) or a set of encoded and quantized intermediate vectors (z q ). BS training component 120 may include a quantizer training component 124 configured to determine a quantization codebook to be applied to the encoded and quantized uplink control information. BS training component 120 may include a decoder training component 126 configured to train a decoder at the base station-associated entity to decode the encoded and quantized uplink control information based on the ordered training data set and the quantization codebook.

[0036] The base station 102 configured for 4G LTE (collectively referred to as the evolved universal mobile telecommunication system (UMTS) terrestrial radio access network (E-UTRAN)) can interface with the EPC 160 via a backhaul link 132 (e.g., the S1 interface). The backhaul link 132 can be wired or wireless. The base station 102 configured for 5G NR (collectively referred to as the next-generation RAN (NG-RAN)) can interface with the 5GC 190 via a backhaul link 184. The backhaul link 184 can be wired or wireless. Among other functions, the base station 102 can perform one or more of the following functions: transfer of user data, radio channel encryption and decryption, integrity protection, header compression, mobility control functions (e.g., handover, dual connectivity), inter-cell interference coordination, connection establishment and release, load balancing, distribution of non-access stratum (NAS) messages, NAS node selection, synchronization, radio access network (RAN) sharing, multimedia broadcast multicast service (MBMS), subscriber and equipment tracking, radio access network information management (RIM), paging, positioning, and delivery of warning messages. The base stations 102 can communicate directly or indirectly with each other (e.g., via the EPC 160 or 5GC 190) on a backhaul link 134 (e.g., the X2 interface). The backhaul link 134 can be wired or wireless.

[0037] The base station 102 can communicate wirelessly with the UE 104. Each base station in the base station 102 can provide communication coverage for a corresponding geographical coverage area 110. There may be overlapping geographical coverage areas 110. For example, the small cell 102' may have a coverage area 110' that overlaps with the coverage areas 110 of one or more macro base stations 102. A network including both small cells and macro cells can be referred to as a heterogeneous network. The heterogeneous network may also include a Home evolved Node B (HeNB), which can provide services to a restricted group called a Closed Subscriber Group (CSG). The communication link 112 between the base station 102 and the UE 104 can include an uplink (UL) (also referred to as a reverse link) transmission from the UE 104 to the base station 102 and / or a downlink (DL) (also referred to as a forward link) transmission from the base station 102 to the UE 104. The communication link 112 can use multiple-input multiple-output (MIMO) antenna technology, including spatial multiplexing, beamforming, and / or transmit diversity. The communication link can pass through one or more carriers. For each carrier allocated in carrier aggregation with a total of up to Yx MHz (x component carriers) for transmission in each direction, the base station 102 / UE 104 can use a spectrum with a bandwidth of up to Y MHz (e.g., 5 MHz, 10 MHz, 15 MHz, 20 MHz, 100 MHz, 400 MHz, etc.). These carriers may or may not be adjacent to each other. The allocation of carriers can be asymmetric with respect to the DL and UL (e.g., more or fewer carriers may be allocated for the DL compared to the UL). The component carriers can include a primary component carrier and one or more secondary component carriers. The primary component carrier can be referred to as the Primary Cell (PCell) and the secondary component carriers can be referred to as the Secondary Cells (SCells).

[0038] Some UEs 104 can communicate with each other using device-to-device (D2D) communication links 158. The D2D communication links 158 can use the DL / UL WWAN spectrum. The D2D communication links 158 can use one or more sidelink channels, such as the Physical Sidelink Broadcast Channel (PSBCH), Physical Sidelink Discovery Channel (PSDCH), Physical Sidelink Shared Channel (PSSCH), Physical Sidelink Control Channel (PSCCH), and Physical Sidelink Feedback Channel (PSFCH). D2D communication can be through various wireless D2D communication systems, such as, for example, FlashLinQ, WiMedia, Bluetooth, ZigBee, Wi-Fi based on the IEEE 802.11 standard, LTE, or NR.

[0039] The wireless communication system may further include a Wi-Fi access point (AP) 150 that communicates with a Wi-Fi station (STA) 152 via a communication link 154 in the 5 GHz unlicensed spectrum. When communicating in the unlicensed spectrum, the STA 152 / AP 150 may perform a Clear Channel Assessment (CCA) before communication to determine whether the channel is available.

[0040] The small cell 102' may operate in licensed and / or unlicensed spectrum. When operating in the unlicensed spectrum, the small cell 102' may employ NR and use the same 5 GHz unlicensed spectrum as that used by the Wi-Fi AP 150. The small cell 102' adopting NR in the unlicensed spectrum may improve the coverage of the access network and / or increase the capacity of the access network.

[0041] The base station 102 (whether it is a small cell 102' or a large cell (e.g., a macro base station)) may include an eNB, a gNodeB (gNB), or other types of base stations. Some base stations (such as the gNB 180) may operate in one or more frequency bands within the electromagnetic spectrum.

[0042] The electromagnetic spectrum is generally subdivided into various categories, frequency bands, channels, etc. based on frequency / wavelength. In 5G NR, two initial operating frequency bands have been identified as Frequency Range Designation FR1 (410 MHz – 7.125 GHz) and FR2 (24.25 GHz – 52.6 GHz). The frequencies between FR1 and FR2 are generally referred to as mid-band frequencies. Although a part of FR1 is greater than 6 GHz, in various documents and articles, FR1 is generally (interchangeably) referred to as the "sub-6 GHz" band. Similar naming issues sometimes occur with FR2. Although it is different from the extremely high frequency (EHF) band (30 GHz to 300 GHz) determined by the International Telecommunication Union (ITU) as the "millimeter wave" (mmW) band, it is generally (interchangeably) referred to as the "millimeter wave" band in various documents and articles. Communications using the mmW radio frequency band have extremely high path loss and short range. The mmW base station 180 may utilize beamforming 182 with the UE 104 to compensate for this path loss and short range.

[0043] Taking the above aspects into consideration, unless otherwise specifically stated, it should be understood that if used in this article, terms such as "sub-6 GHz" may generally represent frequencies that may be less than 6 GHz, may be within FR1, or may include mid-band frequencies. In addition, unless otherwise specifically stated, it should be understood that if the term "millimeter wave" etc. is used in this article, it may generally represent frequencies that may include mid-band frequencies, may be within FR2, or may be within the EHF band. Communications using the mmW radio frequency band have extremely high path loss and short range. The mmW base station 180 may utilize beamforming 182 with the UE 104 to compensate for this path loss and short range.

[0044] Base station 180 may transmit beamformed signals to UE 104 on one or more transmit beams 182'. UE 104 may receive beamformed signals from base station 180 on one or more receive beams 182". UE 104 may also transmit beamformed signals to base station 180 in one or more transmission directions. Base station 180 may receive beamformed signals from UE 104 in one or more reception directions. Base station 180 / UE 104 may perform beam training to determine the optimal reception and transmission directions for each of base station 180 / UE 104. The transmission and reception directions of base station 180 may be the same or may not be the same. The transmission and reception directions of UE 104 may be the same or may not be the same. In the case of a synchronized network, the cells from base station 180 may typically be aligned. Different receive beams 182" may provide optimal performance for each cell. The UE may perform neighbor cell search and beam measurement to identify the optimal receive beam 182" for each cell.

[0045] EPC 160 may include a Mobility Management Entity (MME) 162, other MMEs 164, Serving Gateway 166, Multimedia Broadcast Multicast Service (MBMS) Gateway 168, Broadcast Multicast Service Center (BM-SC) 170, and Packet Data Network (PDN) Gateway 172. MME 162 may communicate with a Home Subscriber Server (HSS) 174. MME 162 is a control node that processes signaling between UE 104 and EPC 160. Generally speaking, MME 162 provides bearer and connection management. All user Internet Protocol (IP) packets are passed through Serving Gateway 166, which itself is connected to PDN Gateway 172. PDN Gateway 172 provides UE IP address allocation and other functions. PDN Gateway 172 and BM-SC 170 are connected to IP services 176. IP services 176 may include the Internet, intranet, IP Multimedia Subsystem (IMS), PS streaming services, and / or other IP services. BM-SC 170 may provide functions for MBMS user service configuration and delivery. BM-SC 170 may serve as an entry point for content provider MBMS transmissions, may be used to authorize and initiate MBMS bearer services in a Public Land Mobile Network (PLMN), and may be used to schedule MBMS transmissions. MBMS Gateway 168 may be used to distribute MBMS traffic to base stations 102 belonging to a Multicast Broadcast Single Frequency Network (MBSFN) area for a particular broadcast service, and may be responsible for session management (start / stop) and for collecting eMBMS-related charging information.

[0046] The 5GC 190 may include an Access and Mobility Management Function (AMF) 192, other AMFs 193, a Session Management Function (SMF) 194, and a User Plane Function (UPF) 195. The AMF 192 may communicate with a Unified Data Management (UDM) 196. The AMF 192 is a control node that processes signaling between the UE 104 and the 5GC 190. Generally speaking, the AMF 192 provides QoS flow and session management. All user Internet Protocol (IP) packets are passed through the UPF 195. The UPF 195 provides UE IP address allocation and other functions. The UPF 195 is connected to an IP service 197. The IP service 197 may include the Internet, an intranet, an IP Multimedia Subsystem (IMS), a PS streaming service, and / or other IP services.

[0047] A base station may also be referred to as a gNB, Node B, evolved Node B (eNB), access point, base station transceiver, radio base station, radio transceiver, transceiver function, Basic Service Set (BSS), Extended Service Set (ESS), Transmission and Reception Point (TRP), or some other suitable term. The base station 102 provides an access point for the UE 104 to the EPC 160 or the 5GC 190. Examples of the UE 104 include cellular phones, smart phones, Session Initiation Protocol (SIP) phones, laptop computers, personal digital assistants (PDAs), satellite radios, global positioning systems, multimedia devices, video devices, digital audio players (e.g., MP3 players), cameras, game consoles, tablets, smart devices, wearable devices, vehicles, electricity meters, gas pumps, large or small kitchen appliances, healthcare devices, implants, sensors / actuators, displays, or any other similar functional devices. Some of the UEs in the UE 104 may be referred to as IoT devices (e.g., parking meters, gas pumps, toasters, vehicles, heart monitors, etc.). The UE 104 may also be referred to as a station, mobile station, subscriber station, mobile unit, subscriber unit, wireless unit, remote unit, mobile device, wireless device, wireless communication device, remote device, mobile subscriber station, access terminal, mobile terminal, wireless terminal, remote terminal, cell phone, user agent, mobile client, client, or some other suitable term.

[0048] Figures 2A to 2D is a resource graph illustrating an example frame structure and channels that can be used for uplink transmission, downlink transmission, and side link transmission to the UE 104 including the UE training component 140. Figure 2A is a diagram 200 illustrating an example of a first subframe within the 5G NR frame structure. Figure 2B is a diagram 230 illustrating an example of a DL channel within a 5GNR subframe. Figure 2C is a diagram 250 illustrating an example of a second subframe within the 5GNR frame structure.Figure 2D FIG. 280 is an illustration showing examples of UL channels within a 5G NR subframe. The 5G NR frame structure can be FDD, where for a particular set of subcarriers (carrier system bandwidth), the subframes within that set of subcarriers are dedicated to DL or UL, or can be TDD, where for a particular set of subcarriers (carrier system bandwidth), the subframes in the set of subcarriers are dedicated to both DL and UL. In Figure 2A 、 Figure 2C the example provided, the 5G NR frame structure is assumed to be TDD, where subframe 4 is configured with slot format 28 (where most are DL), where D is DL, U is UL, and X can be flexibly used between DL / UL, and subframe 3 is configured with slot format 34 (where most are UL). Although subframes 3 and 4 are shown as having slot formats 34 and 28 respectively, any particular subframe can be configured with any of the various available slot formats 0 - 61. Slot formats 0 and 1 are all-DL and all-UL respectively. The other slot formats 2 - 61 include a mixture of DL, UL, and flexible symbols. The UE is configured with the slot format by receiving a slot format indicator (SFI) (configured dynamically via DL control information (DCI) or semi-statically / statically via radio resource control (RRC) signaling). Note that the following description also applies to the 5G NR frame structure as TDD.

[0049] Other wireless communication technologies may have different frame structures and / or different channels. One frame (10 ms) can be divided into 10 equally sized subframes (1 ms). Each subframe can include one or more slots. A subframe can also include mini-slots, which can include 7, 4, or 2 symbols. Each slot may include 7 or 14 symbols, depending on the slot configuration. For slot configuration 0, each slot can include 14 symbols, and for slot configuration 1, each slot can include 7 symbols. The symbols on the DL can be cyclic prefix (CP) OFDM (CP-OFDM) symbols. The symbols on the UL can be CP-OFDM symbols (for high throughput scenarios) or discrete Fourier transform (DFT) spread OFDM (DFT-s-OFDM) symbols (also known as single carrier frequency division multiple access (SC-FDMA) symbols) (for power-constrained scenarios; limited to single-stream transmission). The number of slots within a subframe is based on the slot configuration and numerology. For slot configuration 0, the different numerologies μ0 to 5 allow each subframe to have 1, 2, 4, 8, 16, and 32 slots respectively. For slot configuration 1, the different numerologies 0 to 2 allow each subframe to have 2, 4, and 8 slots respectively. Thus, for slot configuration 0 and numerology μ, there are 14 symbols per slot and 2 μtime slots. The subcarrier spacing and symbol length / duration are functions of the parameter set. The subcarrier spacing can be equal to 2 μ * 15 kHz, where μ is the parameter set from 0 to 5. Thus, the subcarrier spacing for parameter set μ = 0 is 15 kHz, and the subcarrier spacing for parameter set μ = 5 is 480 kHz. The symbol length / duration is negatively correlated with the subcarrier spacing. Figures 2A to 2D An example of time slot configuration 0 with 14 symbols per time slot and parameter set μ = 0 with 1 time slot per subframe is provided. The subcarrier spacing is 15 kHz and the symbol duration is approximately 66.7 μs.

[0050] A resource grid can be used to represent the frame structure. Each time slot includes a resource block (RB) (also referred to as a physical RB (PRB)) that extends 12 consecutive subcarriers. The resource grid is divided into multiple resource elements (REs). The number of bits carried by each RE depends on the modulation scheme.

[0051] As Figure 2A illustrated, some of the REs carry reference (pilot) signals (RSs) for the UE. The RSs can include demodulation RSs (DMRS) 202 for channel estimation at the UE (indicated as Rx for one particular configuration, where 100x is the port number, but other DMRS configurations are possible) and channel state information reference signals (CSI-RS). The RSs can also include beam measurement RSs (BRS), beam refinement RSs (BRRS), and phase tracking RSs (PT-RS).

[0052] Figure 2BExamples of various DL channels within a subframe of a frame are illustrated. The Physical Downlink Control Channel (PDCCH) carries DCI within one or more Control Channel Elements (CCEs), each CCE including nine Resource Element Groups (REGs), each REG including four consecutive Resource Elements (REs) in an OFDM symbol. The Primary Synchronization Signal (PSS) may be within symbol 2 of a specific subframe of a frame (e.g., PSS symbol 242). The PSS is used by UE 104 to determine subframe / symbol timing and the physical layer identity. The Secondary Synchronization Signal (SSS) may be within symbol 4 of a specific subframe of a frame (e.g., SSS symbol 246). The SSS is used by the UE to determine the physical layer cell identity group number and radio frame timing. Based on the physical layer identity and the physical layer cell identity group number, the UE can determine the Physical Cell Identifier (PCI). Based on the PCI, the UE can determine the location of the aforementioned DMRS 202. The Physical Broadcast Channel (PBCH) carrying the Master Information Block (MIB) may be logically grouped with the PSS and SSS to form a Synchronization Signal (SS) / PBCH block, also referred to as SSB 232. The PBCH may be transmitted on symbols 3 to 5 of the subframe, where symbols 3 and 5 are, for example, referred to as PBCH symbols 244, 248 as these symbols mainly include the RBs for the PBCH. The DMRS 202 may be interleaved with the RBs for the PBCH (e.g., every four RBs) to allow decoding of the PBCH. The MIB provides the System Frame Number (SFN) and the number of RBs in the system bandwidth. The Physical Downlink Shared Channel (PDSCH) carries user data, broadcast system information not transmitted via the PBCH (such as System Information Blocks (SIBs)) and paging messages.

[0053] As illustrated in Figure 2C Some REs carry DMRS for channel estimation at the base station (indicated as R for one particular configuration, but other DMRS configurations are possible). The UE may transmit DMRS for the Physical Uplink Control Channel (PUCCH) and DMRS for the Physical Uplink Shared Channel (PUSCH). The PUSCH DMRS may be transmitted in the previous one or two symbols of the PUSCH. The PUCCH DMRS may be transmitted in different configurations depending on whether a short PUCCH or a long PUCCH is transmitted and depending on the specific PUCCH format used. Although not shown, the UE may transmit a Sounding Reference Signal (SRS). The SRS may be used by the base station for channel quality estimation to enable frequency - dependent scheduling of the UL.

[0054] Figure 2DIllustrates examples of various UL channels within a subframe of a frame. The PUCCH can be located at the position indicated in one configuration. The PUCCH carries uplink control information (UCI), such as a scheduling request, a channel quality indicator (CQI), a precoding matrix indicator (PMI), a rank indicator (RI), and HARQ ACK / NACK feedback. The PUSCH carries data and can additionally be used to carry a buffer status report (BSR), a power headroom report (PHR), and / or UCI.

[0055] Figure 3 Is a block diagram of the communication between the base station 310 in the access network and the UE 350. In the DL, IP packets from the EPC 160 can be provided to the controller / processor 375. The controller / processor 375 implements layer 3 and layer 2 functionality. Layer 3 includes the radio resource control (RRC) layer, and layer 2 includes the service data adaptation protocol (SDAP) layer, the packet data convergence protocol (PDCP) layer, the radio link control (RLC) layer, and the media access control (MAC) layer. The controller / processor 375 provides RRC layer functionality associated with the broadcast of system information (e.g., MIB, SIB), RRC connection control (e.g., RRC connection paging, RRC connection establishment, RRC connection modification, and RRC connection release), inter-radio access technology (RAT) mobility, and measurement configuration for UE measurement reporting; PDCP layer functionality associated with header compression / decompression, security (encryption, decryption, integrity protection, integrity verification), and handover support functions; RLC layer functionality associated with the transfer of upper layer packet data units (PDUs), error correction via ARQ, concatenation, segmentation, and reassembly of RLC service data units (SDUs), re-segmentation of RLC data PDUs, and re-ordering of RLC data PDUs; and MAC layer functionality associated with the mapping between logical channels and transport channels, multiplexing of MAC SDUs onto transport blocks (TBs), demultiplexing of MAC SDUs from TBs, scheduling information reporting, error correction via HARQ, priority handling, and logical channel prioritization.

[0056] The transmit (Tx) processor 316 and the receive (Rx) processor 370 implement layer 1 functionality associated with various signal processing functions. Layer 1, which includes the physical (PHY) layer, may include error detection on the transport channel, forward error correction (FEC) encoding / decoding of the transport channel, interleaving, rate matching, mapping to the physical channel, modulation / demodulation of the physical channel, and MIMO antenna processing. The Tx processor 316 handles the mapping to the signal constellation based on various modulation schemes (e.g., binary phase shift keying (BPSK), quadrature phase shift keying (QPSK), M-phase shift keying (M-PSK), M-quadrature amplitude modulation (M-QAM)). The encoded and modulated symbols can then be divided into parallel streams. Subsequently, each stream can be mapped to OFDM subcarriers, multiplexed with reference signals (e.g., pilots) in the time domain and / or frequency domain, and then combined together using an inverse fast Fourier transform (IFFT) to generate a physical channel carrying a time-domain OFDM symbol stream. The OFDM stream undergoes spatial precoding to generate multiple spatial streams. Channel estimates from the channel estimator 374 can be used to determine the encoding and modulation schemes, as well as for spatial processing. The channel estimates can be derived from reference signals transmitted by the UE 350 and / or channel state feedback. Then, each spatial stream can be provided to a different antenna 320 via a separate transmitter 318Tx. Each transmitter 318Tx modulates an RF carrier with the corresponding spatial stream for transmission.

[0057] At the UE 350, each receiver 354Rx receives signals via its corresponding antenna 352. Each receiver 354Rx recovers the information modulated onto the RF carrier and provides the information to the receive (Rx) processor 356. The Tx processor 368 and the Rx processor 356 implement layer 1 functionality associated with various signal processing functions. The Rx processor 356 can perform spatial processing on the information to recover any spatial streams destined for the UE 350. If multiple spatial streams are destined for the UE 350, they can be combined by the Rx processor 356 into a single OFDM symbol stream. Then, the Rx processor 356 uses a fast Fourier transform (FFT) to convert the OFDM symbol stream from the time domain to the frequency domain. The frequency-domain signal includes a separate OFDM symbol stream for each subcarrier of the OFDM signal. The symbols and reference signals on each subcarrier are recovered and demodulated by determining the most likely signal constellation points transmitted by the base station 310. These soft decisions can be based on the channel estimates calculated by the channel estimator 358. Then, the soft decisions are decoded and deinterleaved to recover the data and control signals originally transmitted by the base station 310 on the physical channel. The data and control signals are then provided to the controller / processor 359, which implements layer 3 and layer 2 functionality.

[0058] The controller / processor 359 may be associated with a memory 360 that stores program code and data. The memory 360 may be referred to as a computer-readable medium. In the UL, the controller / processor 359 provides demultiplexing between the transport channel and the logical channel, packet reassembly, decryption, header decompression, and control signal processing to recover IP packets from the EPC 160 or the 5GC 190. The controller / processor 359 is also responsible for error detection using the ACK and / or NACK protocols to support HARQ operations.

[0059] Similar to the functionality described in connection with DL transmission by the base station 310, the controller / processor 359 provides RRC layer functionality associated with system information (e.g., MIB, SIB) acquisition, RRC connection, and measurement reporting; PDCP layer functionality associated with header compression / decompression and security (encryption, decryption, integrity protection, integrity verification); RLC layer functionality associated with the transfer of upper layer PDUs, error correction via ARQ, concatenation, segmentation, and reassembly of RLC SDUs, re-segmentation of RLC data PDUs, and re-ordering of RLC data PDUs; and MAC layer functionality associated with the mapping between the logical channel and the transport channel, multiplexing of MAC SDUs onto TBs, demultiplexing of MAC SDUs from TBs, scheduling information reporting, error correction via HARQ, priority handling, and logical channel prioritization.

[0060] Channel estimates derived by the channel estimator 358 from reference signals or feedback transmitted by the base station 310 may be used by the Tx processor 368 to select appropriate decoding and modulation schemes, and to facilitate spatial processing. The spatial streams generated by the Tx processor 368 may be provided to different antennas 352 via a separate transmitter 354Tx. Each transmitter 354Tx modulates an RF carrier with the corresponding spatial stream for transmission.

[0061] UL transmission is processed at the base station 310 in a manner similar to that described in connection with the receiver functionality at the UE 350. Each receiver 318Rx receives signals via its corresponding antenna 320. Each receiver 318Rx recovers the information modulated onto the RF carrier and provides the information to the Rx processor 370.

[0062] The controller / processor 375 may be associated with a memory 376 that stores program code and data. The memory 376 may be referred to as a computer-readable medium. In the UL, the controller / processor 375 provides demultiplexing between the transport channel and the logical channel, packet reassembly, decryption, header decompression, and control signal processing to recover IP packets from the UE 350. The IP packets from the controller / processor 375 may be provided to the EPC 160. The controller / processor 375 is also responsible for error detection using the ACK and / or NACK protocols to support HARQ operations.

[0063] At least one of the Tx processor 368, the Rx processor 356, and the controller / processor 359 may be configured to perform aspects related to Figure 1 the UE training component 140. For example, the memory 360 may include executable instructions that define the UE training component 140. The Tx processor 368, the Rx processor 356, and / or the controller / processor 359 may be configured to execute the UE training component 140.

[0064] At least one of the Tx processor 316, the Rx processor 370, and the controller / processor 375 may be configured to perform aspects related to Figure 1 the BS training component 120. For example, the memory 376 may include executable instructions that define the BS training component 120. The Tx processor 316, the Rx processor 370, and / or the controller / processor 375 may be configured to execute the BS training component 120.

[0065] Figure 4 A diagram illustrating an exemplary disaggregated base station 400 architecture. The disaggregated base station 400 architecture may include one or more central units (CUs) 410 that may communicate directly with the core network 420 via a backhaul link, or indirectly with the core network 420 through one or more disaggregated base station units (such as a near real-time (near RT) RAN intelligent controller (RIC) 425 via an E2 link, or a non-real-time (non RT) RIC 415 associated with the service management and orchestration (SMO) framework 405, or both). The CU 410 may communicate with one or more distributed units (DUs) 430 via a respective midhaul link (such as an F1 interface). The DU 430 may communicate with one or more radio units (RUs) 440 via a respective fronthaul link. The RU 440 may communicate with a respective UE 104 via one or more radio frequency (RF) access links. In some embodiments, the UE 104 may be served simultaneously by multiple RUs 440.

[0066] Each of the units (i.e., CU 410, DU 430, RU 440, and the near RT RIC 425, non-RT RIC 415, and SMO framework 405) may include one or more interfaces or be coupled to one or more interfaces that are configured to receive or transmit signals, data, or information (collectively referred to as signals) via a wired or wireless transmission medium. Each of the units or the associated processor or controller that provides instructions to the communication interfaces of these units may be configured to communicate with one or more of the other units via the transmission medium. For example, the units may include a wired interface that is configured to receive or transmit signals to one or more of the other units via the wired transmission medium. Additionally, the unit may include a wireless interface that may include a receiver, transmitter, or transceiver (such as a radio frequency (RF) transceiver) that is configured to receive or transmit signals, or both, to one or more of the other units via the wireless transmission medium.

[0067] In some aspects, the CU 410 may host one or more higher layer control functions. Such control functions may include radio resource control (RRC), packet data convergence protocol (PDCP), service data adaptation protocol (SDAP), etc. Each control function may utilize an interface that is configured to communicate signals with other control functions hosted by the CU 410. The CU 410 may be configured to handle user plane functionality (i.e., Central Unit - User Plane (CU-UP)), control plane functionality (i.e., Central Unit - Control Plane (CU-CP)), or a combination thereof. In some embodiments, the CU 410 may be logically split into one or more CU-UP units and one or more CU-CP units. When implemented in an O-RAN configuration, the CU-UP units may communicate bi-directionally with the CU-CP units via an interface (such as an E1 interface). As needed, the CU 410 may be implemented to communicate with the DU 430 for network control and signaling.

[0068] The DU 430 may correspond to a logical unit that includes one or more base station functions for controlling the operation of one or more RUs 440. In some aspects, the DU 430 may host one or more of the radio link control (RLC) layer, the media access control (MAC) layer, and one or more high physical (PHY) layers (such as modules for forward error correction (FEC) encoding and decoding, scrambling, modulation and demodulation, etc.) at least partially according to a functional split (such as those defined by the 3rd Generation Partnership Project (3GPP)). In some aspects, the DU 430 may also host one or more low PHY layers. Each layer (or module) may be implemented using an interface configured to communicate signals with other layers (and modules) hosted by the DU 430 or with control functions hosted by the CU 410.

[0069] Lower layer functionality may be implemented by one or more RUs 440. In some deployments, the RUs 440 controlled by the DU 430 may correspond to logical nodes that host RF processing functions or low PHY layer functions (such as performing fast Fourier transform (FFT), inverse FFT (iFFT), digital beamforming, physical random access channel (PRACH) extraction and filtering, etc.) or both at least partially based on a functional split (such as a lower layer functional split). In such an architecture, the RUs 440 may be implemented to handle over-the-air (OTA) communication with one or more UEs 104. In some embodiments, the real-time and non-real-time aspects of the control plane and user plane communication with the RUs 440 may be controlled by the corresponding DU 430. In some scenarios, this configuration may enable the implementation of the DU 430 and the CU 410 in a cloud-based RAN architecture (such as a vRAN architecture).

[0070] The SMO framework 405 can be configured to support the RAN deployment and orchestration of non-virtualized network elements and virtualized network elements. For non-virtualized network elements, the SMO framework 405 can be configured to support the deployment of dedicated physical resources for RAN coverage requirements, and these dedicated physical resources can be managed via an operation and maintenance interface (such as the O1 interface). For virtualized network elements, the SMO framework 405 can be configured to interact with a cloud computing platform (such as the Open Cloud (O-Cloud) 490) to perform network element lifecycle management (such as instantiating virtualized network elements) via a cloud computing platform interface (such as the O2 interface). Such virtualized network elements can include, but are not limited to, the CU 410, DU 430, RU 440, and the Near RT RIC 425. In some specific implementations, the SMO framework 405 can communicate with the hardware aspect of the 4G RAN (such as the Open eNB (O-eNB) 411) via the O1 interface. Additionally, in some specific implementations, the SMO framework 405 can communicate directly with one or more RUs 440 via the O1 interface. The SMO framework 405 can also include a Non-RT RIC 415 configured to support the functionality of the SMO framework 405.

[0071] The Non-RT RIC 415 can be configured to include logical functions that can implement non-real-time control and optimization of RAN elements and resources, artificial intelligence / machine learning (AI / ML) workflows including model training and update, or policy-based guidance of applications / features in the Near RT RIC 425. The Non-RT RIC 415 can be coupled to or communicate with the Near RT RIC 425 (such as via the A1 interface). The Near RT RIC 425 can be configured to include logical functions that can achieve near-real-time control and optimization of RAN elements and resources through an interface (such as via the E2 interface) via data collection and actions, and this interface connects one or more CUs 410, one or more DUs 430, or both, and the O-eNB to the Near RT RIC 425.

[0072] In some specific implementations, to generate the AI / ML models to be deployed in the near RT RIC 425, the non-RT RIC 415 may receive parameters or external enrichment information from an external server. Such information can be utilized by the near RT RIC 425 and can be received from non-network data sources or from network functions at the SMO framework 405 or the non-RT RIC 415. In some examples, the non-RT RIC 415 or the near RT RIC 425 may be configured to regulate RAN behavior or performance. For example, the non-RT RIC 415 may monitor long-term trends and patterns of performance and employ an AI / ML model to perform corrective actions via the SMO framework 405 (such as reconfiguration via O1) or via creating RAN management policies (such as A1 policies).

[0073] Figure 5 is a message diagram of an example multi-vendor training process 500. The process 500 may be executed between entities 508 associated with multiple UEs (e.g., UE vendor servers 508a and 508b) and entities 506 associated with base stations (e.g., base station servers). The UE-associated entities 508 may include UE training components 140, and the base station-associated entities 506 may include BS training components 120. The process 500 may be UE-driven because the UE-associated entities 508 perform initial training in sequential training. For example, the UE-associated entities 508 may each perform UE training 510 to train an encoder 514. In one aspect, the encoder 514 is a machine learning model, such as a neural network.

[0074] UE training 510 may start from an input vector set (V in )512. For example, V in 512 may be a vector representing uplink control information such as CSF or CSI reported by the UE 104 to the base station 102. V in 512 may be collected from one or more UEs 104, generated, engineered, and / or synthesized in a modeling or testing environment. In some specific implementations, V in 512 may be considered proprietary.

[0075] The UE training component 140 may provide V in 512 to the encoder 514 to generate an encoded intermediate vector set (z e )516, which may also be referred to as a latent vector or a compressed vector. The UE training component 140 may provide z e516 is provided to UE decoder 518. UE decoder 518 can be a machine learning model, such as a neural network for training purposes, and may not actually be deployed to UE 104 because UE 104 does not need to decode uplink control information. UE decoder 518 can generate a set of output vectors (V out ) 520, which ideally should be the same as V in 512. However, some error is expected. Loss function 522 can calculate the error between V in 512 and V out 520. Loss function 522 can be used, for example, to calculate gradients that can be used to update the weights within encoder 514 and UE decoder 518.

[0076] UE-associated entity 508 can send sequential training dataset 524 to base-station-associated entity 506. Sequential training dataset 524 can include z e 516 and V in 512 or V out 520. At block 530, base-station-associated entity 506 can aggregate sequential training dataset 524 to form a training set that includes intermediate vector (z) 532 and V in 512 or V out 520.

[0077] At block 540, base-station-associated entity 506 can train base-station decoder 542. Block 540 can include providing z 532 to base-station decoder 542. Base-station decoder 542 can be a machine learning model, such as a neural network trained to decode an encoded vector into uplink control information. Base-station decoder 542 can generate a set of output vectors (Vout,BS) 544, which ideally should be the same as Vin 512 or Vout 520. Depending on the content of sequential training dataset 524, loss function 546 can calculate the error of Vout,BS 544 based on Vin 512 or Vout 520. Loss function 546 can determine gradients to update the weights of base-station decoder 542. Different from UE decoder 518, base-station decoder 542 can be deployed to base station 102. Sequential training allows encoder 514 and base-station decoder 542 to be trained by corresponding entities without being shared.

[0078] In one aspect, a separate training framework may include a process for training one base station decoder to work with multiple UE encoders. In an actual system, the UE quantizes the latent vector (e.g., ze) before sending it to the base station so as to convey the latent vector using only a limited number of bits. Scalar or vector quantization can be applied to the latent vector. Quantization is achieved by using a codebook containing a limited number of scalars or vectors.

[0079] Figure 6 is a diagram of an example quantization and inverse quantization process 610 that can be used within the encoding / decoding process 600. The process 600 may start from providing V in 512 to the encoder 514 to generate z e 516. The quantization and inverse quantization process 610 can be performed on z e 516. For example, during operation, z e 516 can be quantized by vector quantization 612 to generate a set of encoded and quantized intermediate vectors (z q ) 618, and this set of intermediate vectors can be represented by bits 614 for transmission. The bits 614 can be sent by the UE 104 to the base station 102. The base station 102 can optionally perform vector inverse quantization 616 to generate a set of encoded and quantized intermediate vectors (z q ) 618. For example, the vector inverse quantization 616 can use the reconstruction codebook to convert the bits into quantized vectors. The base station 102 can provide z q 618 to the BS decoder 542 to generate V out,BS 544. In some embodiments, the BS decoder 542 can be trained to operate directly on the bits 614.

[0080] In one aspect, the quantization process 610 can use a quantization codebook 620 to further reduce the size of z e 516 for transmission. The quantization codebook 620 can map the vectors to a finite real value or bit stream. For example, the quantization codebook 620 can map sub-vectors of size 2 or 4, where each vector entry is represented by 2 bits. Each entry in the quantization codebook 620 is a vector of size d subset. z e 516 can be larger, and each entry in the codebook is a vector of size d subset. The vector quantization 612 can quantize z eThe vector 516 is partitioned into sub-vectors 626 (e.g., sub-vectors 626a, 626b, 626c, 626d) of size d (e.g., 2 or 4). Vector quantization 612 can use a quantization codebook 620 to map each sub-vector 626 to a finite real value (e.g., K) or a bit value. For example, in diagram 630, the quantization codebook 620 can define quantization values. Vector quantization 612 can map each sub-vector 626 to the closest quantization value. Each quantization value can be associated with a bitstream 614. At the base station, the bits 614 can be mapped to sub-vectors 632 (e.g., sub-vectors 632a, 632b, 632c, 632d) by using a reconstruction codebook (which can be the same as or the opposite of the quantization codebook 620) and the multiple sub-vectors are combined into z q 618 to recreate the encoded vector.

[0081] In one aspect, training the quantization codebook 620 can involve selecting quantization values based on a training set of input values. The quantization values can be values that minimize the average distance from the input values. For example, the quantization values can be selected by clustering to find quantization values close to many input values. Since quantization can depend on the output of the encoder 514 and / or affect the decoder 542, the training of quantization can be done jointly with the training of the encoder or decoder. That is, whenever the weights of the encoder or decoder are updated, the selected quantization values can be updated. The quantization codebook 620 can be shared between the UE-associated entity 508 and the base-station-associated entity 506.

[0082] Figure 7 is a diagram of an example sequential training system 700 for the encoder 514, the decoder 542, and quantization. For optimizing performance, in end-to-end learning, the quantization codebook should be learned together with the neural networks for the encoder and decoder. For example, the encoder 514 and the quantization codebook 620 can be trained at the UE-associated entity 508 and deployed to the UE 104, and the decoder 542 can be trained at the base-station-associated entity 506 and deployed to the base station 102. As another example, the UE-associated entity 508 can train the encoder 514 and deploy the encoder 514 to the UE 104, and the base-station-associated entity 506 can train the decoder 542 and the quantization codebook 620 for deployment to the base station 102. The trained quantization codebook 620 can be shared between the UE-associated entity 508 and the base-station-associated entity 506.

[0083] During training at the UE-associated entity 508, the UE-associated entity 508 can receive V in 512 (e.g., from one or more UEs 104 or a synthetic source). UE encoder training 710 can train the neural network for the encoder 514. In some embodiments, z e516 performs quantizer training 720. The quantizer training 720 can quantize z q 618 is output to decoder training 730. The decoder training 730 can train a UE decoder for training only. The decoder training 730 can output V out 520 is output to loss function 522. The loss function can calculate the gradient between V in 512 and V out 520, and use this gradient to adjust the weights in UE encoder training 710 and UE decoder training 730. The UE encoder training 710 can deploy the trained UE encoder to the encoder 514 at UE 104. The quantizer training 720 can deploy the trained quantizer to the quantizer 612 at UE 104. The quantizer training 720 can also share the quantization codebook 620 with the base station - associated entity 506. The UE - associated entity 508 can share the training dataset 732. The training dataset 732 can include z e 516 or z q 618. The training dataset 732 can include V in 512 or V out 520.

[0084] During training at the base - station - associated entity 506, the base - station - associated entity 506 can optionally perform quantizer training 722. For example, the quantizer training 722 can receive z e 516 and train the quantization codebook 620 to produce z q 618. The quantizer training 722 can share the quantization codebook 620 back with the UE - associated entity 508. Decoder training 740 can be performed on z q 618 received in the training dataset 732 or generated by the quantizer training 722. The decoder training 740 can train a neural network to generate V out,BS 544. The loss function 546 can compare the received V in 512 or V out 520 with V out,BS 544 to determine the gradient and update the weights of the neural network. The base - station - associated entity 506 can output the trained decoder to the base station 102 for use as decoder 542.

[0085] In operation (e.g., during the inference phase), UE 104 can obtain a channel estimate 702 (e.g., based on reference signal measurements). UE 104 can provide the channel estimate to the encoder 514, which can generate an encoded intermediate vector (z e)516. In some specific implementations, the UE 104 may also store the channel estimate 702 as a CSI log 704, and this CSI log can be used as training data for encoder training (e.g., V in ). The quantizer 612 may quantize z e 516 to generate a bitstream for transmission as uplink control information 706 (e.g., based on the quantization codebook 620). At the base station 102, the dequantizer 616 may convert the bit 614 back to an intermediate decoded and quantized vector (z q ). The decoder 542 may decode z q to obtain V out,BS , which can be interpreted as, for example, CSI 708.

[0086] Figure 8 is a diagram of an example multi-vendor sequential training system 800 for quantization training at entities associated with the encoder and decoder and the UE. Each UE-associated entity 508 may separately train the encoder 514, the quantization codebook 620, and the UE decoder 518. Each UE-associated entity 508 may generate a data set 732 for transmission to the base-station-associated entity 506.

[0087] In one aspect, the design of the multi-vendor sequential training system 800 may depend on how much information is shared between the UE-associated entity 508 and the base-station-associated entity 506. For example, based on standards or rules, some information such as the payload size (number of bits and z dimension) of z is known to both entities. In the first option, there may be no quantization-related consistency between the UE-associated entity 508 and the base-station-associated entity 506. The UE-associated entity 508 may use scalar or vector quantization to train its encoder-decoder pair. The UE-associated entity 508 may share with the base-station-associated entity 506 a data set 732 including either z q and V in or V out in it. By analyzing the data set 732, the base-station-associated entity 506 may calculate that the z space is quantized and no quantization is required when training the decoder. Alternatively, the base-station-associated entity 506 may train additional quantization as part of the decoder; however, two-stage quantization may not be desirable (e.g., due to complexity).

[0088] In the second option, the UE - associated entity 508 and the base - station - associated entity 506 agree to perform quantization on the UE side. However, the choice of quantization method depends on the UE - associated entity 508. The UE - associated entity 508 selects the quantization method (scalar versus vector quantization) and the quantization parameters. The base - station - associated entity 506 does not need to know the details of the quantization method. The UE - associated entity 508 may share with the base - station - associated entity 506 a data set 732 that includes either z q and V in or V out The base - station - associated entity 506 may train the BS decoder 542 without performing quantization.

[0089] In the third option, the UE - associated entity 508 and the base - station - associated entity 506 agree on the exact quantization method to be used in the training at the UE - associated entity 508. The UE - associated entity 508 shares with the base - station - associated entity 506 a data set 732 that includes either z q and V in or V out The base - station - associated entity 506 may train the BS decoder 542 without performing quantization.

[0090] In one aspect, the BS decoder 542 can be a multi - UE decoder that includes a first vendor - specific layer 810 (e.g., corresponding to a first UE vendor server 508a) and a second vendor - specific layer 820 (e.g., corresponding to a second UE vendor server 508b). The BS decoder 542 also includes a shared decoder layer 830. The BS decoder 542 can output V out,BS 544 to a loss function 546. The loss function 546 can compare V out,BS 544 with V in 512 for each UE - associated entity 508 to determine the gradient for adjusting the weights of the first vendor - specific layer 810, the second vendor - specific layer 820, and / or the shared decoder layer 830.

[0091] Figure 9 is a diagram of an example multi - vendor sequential training system 900 for quantization training of the encoder and decoder at the base - station - associated entity 506. Each UE - associated entity 508 can train the encoder 514 and the UE decoder 518 separately. Each UE - associated entity 508 can generate a data set 732 for transmission to the base - station - associated entity 506. The base - station - associated entity 506 can train the quantization codebook 620 and the BS decoder 542.

[0092] In one aspect, the design of the multi-vendor sequential training system 900 can depend on how much information is shared between the UE-associated entity 508 and the base station-associated entity 506. In a first option, the UE-associated entity 508 and the base station-associated entity 506 agree that the quantization codebook 620 will be trained at the base station-associated entity 506 together with the BS decoder 542. In some embodiments, the UE-associated entity 508 can train the encoder 514 and UE decoder 518 pair without VQ and share with the base station-associated entity 506 a data set 732 including either z e and V in or V out . The base station-associated entity 506 can include the quantizer training 720 as part of the BS decoder 542. Alternatively, the UE-associated entity 508 can train its encoder and decoder pair using scalar or vector quantization and share with the base station-associated entity 506 a data set 732 including either z e or z q and V in or V out . The base station-associated entity 506 can train a common quantization codebook 620 for both the UE 104 and the base station 102 based on z e . The base station-associated entity 506 can share the quantization codebook 620 back with the UE-associated entity 508 for use at the UE 104. In some embodiments, in the case where the base station-associated entity 506 receives only z q , the base station-associated entity 506 can train an additional quantization codebook 620 based on z q , which may be undesirable, for example due to complexity.

[0093] In a second option, the UE-associated entity 508 and the base station-associated entity 506 agree on the exact quantization method to be used in the training at the UE-associated entity 508. The UE-associated entity 508 can train its encoder 514 and decoder 518 pair without quantization and share with the base station-associated entity 506 either z e and V in or V out . Alternatively, the UE-associated entity 508 can train its encoder 514 and decoder 518 pair using the agreed quantization method and share with the base station-associated entity 506 either z e and V in or V out . In either case, the base station-associated entity 506 includes the quantizer training as part of the BS decoder 542.

[0094] In one aspect, the base station - associated entity 506 may train the vendor - specific quantization codebooks 620a and 620b along with the BS decoder 542. For example, the base station - associated entity 506 may receive the corresponding Z in the data set 732 e 516 and train the vendor - specific quantization codebook 620. The corresponding quantizer 612 may quantize the vendor - specific z q output to the BS decoder 542. The BS decoder 542 may be a multi - UE decoder including a first vendor - specific layer 910 (e.g., corresponding to the first UE vendor server 508a) and a second vendor - specific layer 920 (e.g., corresponding to the second UE vendor server 508b). The BS decoder 542 also includes a shared decoder layer 930. The BS decoder 542 may output V out,BS 544 to the loss function 546. The loss function 546 may compare V out,BS 544 with V in 512 for each UE - associated entity 508 to determine the gradients for adjusting the weights of the first vendor - specific layer 910, the second vendor - specific layer 920, and / or the shared decoder layer 930. Additionally, in some embodiments, the loss function 546 may adjust the vendor - specific quantization codebooks 620a and 620b.

[0095] Figure 10 is a flowchart of an example method 1000 for a UE - associated entity 508 to train an encoder and a quantizer. The method 1000 may be performed by a UE - associated entity 508 (such as a UE vendor server 108 or a UE 104, which may include a memory 360 and may be the entire UE 104 or a component of the UE 104 (such as a UE training component 140, a Tx processor 368, an Rx processor 356, or a controller / processor 359)). The method 1000 may be performed by a UE training component 140 in communication with a BS training component 120 of an entity 506 associated with one or more base stations. Optional blocks are shown in dashed lines.

[0096] At block 1010, the method 1000 optionally includes determining a level of consistency between the UE vendor and the base station equipment vendor. In some embodiments, for example, the UE 104, the Rx processor 356, or the controller / processor 359 may execute the UE training component 140 or the shared component 146 to determine the level of consistency between the UE vendor and the base station equipment vendor. Thus, the UE 104, the Rx processor 356, or the controller / processor 359 that executes the UE training component 140 or the shared component 146 may provide the components for determining the level of consistency between the UE vendor and the base station equipment vendor.

[0097] At block 1020, method 1000 includes training an encoder to encode uplink control information. In some embodiments, for example, UE 104, TX processor 368, or controller / processor 359 may execute UE training component 140 or encoder training component 142 to train encoder 514 to encode uplink control information. In some embodiments, the encoder is trained based on a loss function 522 between a set of input vectors (e.g., V in 512) and a set of output vectors (e.g., V out ) from UE decoder 518. In some embodiments, at sub-block 1022, block 1020 may optionally include training the encoder without quantization. In some embodiments, at sub-block 1024, block 1020 may include training the encoder with a first quantization codebook. For example, the first quantization codebook may be trained at an entity 508 associated with the UE. Thus, UE 104, TX processor 368, or controller / processor 359, which executes UE training component 140 or encoder training component 142, may provide components for training the encoder to encode uplink control information.

[0098] At block 1030, method 1000 may optionally include determining a quantization codebook to apply to the encoded uplink control information. In some embodiments, for example, UE 104, TX processor 368, or controller / processor 359 may execute UE training component 140 or quantizer training component 144 to determine quantization codebook 620 to apply to the encoded uplink control information. In some embodiments, for example, at sub-block 1032, block 1030 may optionally include training the quantization codebook based on a set of encoded and unquantized intermediate vectors. For example, at sub-block 1034, sub-block 1032 may optionally include selecting a quantization scheme (e.g., when there is no agreement between the UE vendor and the base station vendor). Example quantization schemes may include scalar quantization or vector quantization. Parameters for vector quantization may include subset size and bit size. Also, for example, at sub-block 1036, sub-block 1032 may optionally include training the quantization codebook using an agreed quantization method and quantization parameters.

[0099] In some specific implementations, at sub - block 1040, block 1030 may optionally include receiving one or more quantization codebooks from an entity 506 associated with the base station. For example, at sub - block 1042, sub - block 1040 may optionally include receiving one or more second quantization codebooks according to an agreed - upon quantization method and quantization parameters. The one or more second quantization codebooks may be trained by an entity 506 associated with the base station and replace the first quantization codebook trained at an entity associated with the UE. As another example, at sub - block 1044, sub - block 1040 may optionally include receiving a finally refined quantization codebook from an entity associated with the base - station vendor. The finally refined quantization codebook may be trained by an entity 506 associated with the base station based on the first quantization codebook trained at an entity associated with the UE (e.g., in sub - block 1032).

[0100] In some specific implementations, at sub - block 1046, block 1030 may optionally include generating a finally refined quantization codebook based on a second quantization codebook from an entity associated with the base station or a reconstruction codebook from an entity associated with the base station. For example, the second quantization codebook or the reconstruction codebook may be received in sub - block 1040. The quantizer training component 144 may further train the received codebook based on the encoder 514.

[0101] In view of the foregoing, the UE 104, the TX processor 368, or the controller / processor 359 that executes the UE training component 140 or the quantizer training component 144 may provide components for determining a quantization codebook to be applied to the encoded uplink control information.

[0102] At block 1050, method 1000 may optionally include sharing one or more quantization codebooks with an entity associated with the base station. In some specific implementations, for example, the UE 104, the Rx processor 368, or the controller / processor 359 may execute the UE training component 140 or the sharing component 146 to share one or more quantization codebooks 620 with an entity 506 associated with the base station. Thus, the UE 104, the Tx processor 368, or the controller / processor 359 that executes the UE training component 140 or the sharing component 146 may provide components for sharing one or more quantization codebooks with an entity associated with the base station.

[0103] At block 1060, method 1000 includes sharing an ordered training data set with an entity associated with the base station. In some specific implementations, for example, the UE 104, the Rx processor 368, or the controller / processor 359 may execute the UE training component 140 or the sharing component 146 to share the ordered training data set 732 with an entity 506 associated with the base station. The ordered training data set 732 includes one of an input vector set or an output vector set; and a set of encoded and unquantized intermediate vectors (e.g., z e 516) or a set of encoded and quantized intermediate vectors (e.g., zq One of 618). Thus, the UE 104, Tx processor 368, or controller / processor 359 that executes the UE training component 140 or the sharing component 146 can provide components for sharing the sequential training data set with entities associated with the base station.

[0104] At block 1070, method 1000 can optionally include deploying an encoder and a quantization codebook to one or more UEs for use with a base station of a base station equipment vendor. In some specific implementations, for example, the UE 104, Tx processor 368, or controller / processor 359 can execute the UE training component 140 or the deployment component 148 to deploy the encoder 514 and the quantization codebook 620 to one or more UEs 104 for use with the base station 102 of a base station equipment vendor. Thus, the UE 104, Tx processor 368, or controller / processor 359 that executes the UE training component 140 or the deployment component 148 can provide components for deploying an encoder and a quantization codebook to one or more UEs for use with a base station of a base station equipment vendor.

[0105] Figure 11 Is a flowchart of an example method 1100 for a decoder and a quantization codebook to be trained by an entity 506 associated with a base station. Method 1100 can be executed by an entity 506 associated with a base station (such as a base station server 106 or a base station 102, which can include a memory 376 and can be the entire base station 102 or a component of the base station 102 (such as a BS training component 120, Tx processor 316, Rx processor 370, or controller / processor 375)). Method 1100 can be executed by the BS training component 120 that communicates with the UE training component 140 of an entity 508 associated with one or more UEs. Optional blocks are shown with dashed lines.

[0106] At block 1110, method 1100 can optionally include determining a level of consistency between a UE vendor and a base station equipment vendor. In some specific implementations, for example, the base station 102, Tx processor 316, or controller / processor 375 can execute the BS training component 120 to determine the level of consistency between a UE vendor and a base station equipment vendor. Thus, the base station 102, Tx processor 316, or controller / processor 375 that executes the BS training component 120 can provide components for determining the level of consistency between a UE vendor and a base station equipment vendor.

[0107] At block 1120, method 1100 includes receiving an ordered training data set from an entity associated with at least a first UE. In some specific implementations, for example, base station 102, Rx processor 370, or controller / processor 375 may execute BS training component 120 or data set receiving component 122 to receive an ordered training data set 732 from an entity 508 associated with at least a first UE (e.g., UE vendor server 508a). The ordered training data set 732 includes one of a set of input vectors or a set of output vectors; and a set of encoded and unquantized intermediate vectors (e.g., z e 516) or one of a set of encoded and quantized intermediate vectors (e.g., z q 618). In some specific implementations, at block 1130, method 1100 may optionally include receiving an ordered training data set from a second UE-associated entity 508 (e.g., second UE vendor server 508b). Thus, base station 102, Rx processor 370, or controller / processor 375 that executes BS training component 120 or data set receiving component 122 may provide means for receiving an ordered training data set from an entity associated with at least a first UE.

[0108] At block 1140, method 1100 includes determining a quantization codebook to be applied to encoded uplink control information. In some specific implementations, for example, base station 102, Rx processor 370, or controller / processor 375 may execute BS training component 120 or quantizer training component 124 to determine a quantization codebook to be applied to encoded uplink control information.

[0109] For example, in some specific implementations, at sub-block 1142, block 1140 may optionally include analyzing a set of encoded and quantized intermediate vectors to determine that the set of encoded and quantized intermediate vectors is quantized. For example, the set of encoded and quantized intermediate vectors may include a finite number of values. Quantizer training component 124 may generate a codebook based on the finite number of values.

[0110] In some specific implementations, at sub-block 1144, block 1140 may optionally include receiving one or more quantization codebooks 620 or reconstruction codebooks from a UE-associated entity 508. For example, quantization codebook 620 may be trained at a UE-associated entity 508.

[0111] In some specific implementations, at sub-block 1150, block 1140 may optionally include training quantization codebook 620 with decoder 542. For example, at sub-block 1152, when the ordered training data set 732 includes a set of encoded and unquantized intermediate vectors (e.g., z eAt 514), the quantizer training component 124 may optionally train a single quantization codebook 620 for the first UE and the base station. For another example, at sub - box 1154, sub - box 1150 may optionally include: when the sequential training data set 732 includes an encoded and quantized set of intermediate vectors (e.g., z q 618), train a second quantization codebook 620 for the base station based on the UE quantization codebook. That is, the quantizer training component 124 may train second - level quantization.

[0112] In view of the foregoing, the base station 102, the Rx processor 370, or the controller / processor 375 that executes the BS training component 120 or the quantizer training component 124 may provide components for determining the quantization codebook to be applied to the encoded uplink control information.

[0113] At block 1160, the method 1100 may optionally include transmitting one or more quantization codebooks to an entity associated with at least the first UE according to an agreed - upon scheme and quantization parameters. In some specific implementations, for example, the base station 102, the Tx processor 316, or the controller / processor 375 may execute the BS training component 120 to transmit one or more quantization codebooks to an entity associated with at least the first UE according to an agreed - upon scheme and quantization parameters. For example, one or more quantization codebooks may be trained in sub - box 1150. In some specific implementations, at sub - box 1062, block 1060 may optionally include transmitting a final refined quantization codebook to an entity associated with at least the first UE. For example, the final refined quantization codebook may be refined based on the quantization codebook received in sub - box 1144. Thus, the base station 102, the Tx processor 316, the Rx processor 370, or the controller / processor 375 that executes the BS training component 120 may provide components for transmitting one or more quantization codebooks to an entity associated with at least the first UE according to an agreed - upon scheme and quantization parameters.

[0114] At block 1170, the method 1100 includes training a decoder to decode the encoded and quantized uplink control information based on the sequential training data set and the quantization codebook. In some specific implementations, for example, the base station 102, the Rx processor 370, or the controller / processor 375 may execute the BS training component 120 or the decoder training component 126 to train the decoder 542 to decode the encoded and quantized uplink control information based on the sequential training data set 732 and the quantization codebook 620. In some specific implementations, the decoder is trained based on an input vector set (e.g., V in 512) or an output vector set (e.g., V out ) and an applied set of encoded and unquantized intermediate vectors (e.g., z e 516) or an encoded and quantized set of intermediate vectors (e.g., zq output vector set (e.g., V of 618) out,BS A loss function 546 between (e.g., V of 544). Thus, the base station 102, Rx processor 370, or controller / processor 375 that executes the BS training component 120 or the decoder training component 126 can provide components for training the decoder to decode the encoded and quantized uplink control information based on the sequential training dataset and the quantization codebook.

[0115] At block 1180, the method 1100 may optionally include deploying the decoder and the quantization codebook to one or more base stations for use with UEs of at least a first UE vendor. In some specific implementations, for example, the base station 102, Tx processor 316, or controller / processor 375 may execute the BS training component 120 to deploy the decoder 542 and the quantization codebook 620 (e.g., for the dequantizer 616) to one or more base stations for use with UEs of at least a first UE vendor. Thus, the base station 102, Tx processor 316, or controller / processor 375 that executes the BS training component 120 can provide components for deploying the decoder and the quantization codebook to one or more base stations for use with UEs of at least a first UE vendor.

[0116] The following numbered clauses provide an overview of aspects of the present disclosure:

[0117] 1. A method performed at a UE-associated entity, comprising:

[0118] Training an encoder to encode uplink control information;

[0119] Determining a quantization codebook to be applied to the encoded uplink control information; and sharing a sequential training dataset with a base-station-associated entity, the sequential training dataset including:

[0120] One of an input vector set or an output vector set; and one of an encoded and unquantized intermediate vector set or an encoded and quantized intermediate vector set.

[0121] 2. The method according to clause 1, further comprising: determining a level of consistency between a UE vendor and a base-station equipment vendor, wherein the training of the encoder, the determining of the quantization codebook, or the content of the sequential training dataset is based on the level of consistency.

[0122] 3. The method according to clause 2, further comprising: deploying the encoder and the quantization codebook to one or more UEs for use with base stations of a base-station equipment vendor.

[0123] 4. The method according to any one of clauses 1 to 3, wherein training the encoder is based on a loss function between the set of input vectors and the set of output vectors from the UE decoder.

[0124] 5. The method according to any one of clauses 1 to 4, wherein determining the quantization codebook includes: training the quantization codebook at the entity associated with the UE based on the set of encoded and unquantized intermediate vectors, wherein the sequential training data set includes the set of encoded and quantized intermediate vectors.

[0125] 6. The method according to clause 5, further comprising: sharing one or more quantization codebooks with the entity associated with the base station.

[0126] 7. The method according to clause 5 or 6, wherein training the quantization codebook at the entity associated with the UE based on the set of encoded and unquantized intermediate vectors includes: selecting a quantization scheme.

[0127] 8. The method according to clause 5 or 6, wherein training the quantization codebook at the entity associated with the UE based on the set of encoded and unquantized intermediate vectors includes: training the quantization codebook using an agreed quantization method and quantization parameters.

[0128] 9. The method according to any one of clauses 1 to 4, wherein determining the quantization codebook includes: receiving one or more quantization codebooks from the entity associated with the base station, wherein training the encoder at the entity associated with the UE to encode uplink control information includes: training the encoder without quantization.

[0129] 10. The method according to any one of clauses 1 to 4, wherein training the encoder to encode uplink control information includes: training the encoder with a first quantization codebook, and wherein determining the quantization codebook includes: receiving a second quantization codebook from the entity associated with the base station.

[0130] 11. The method according to any one of clauses 1 to 4, wherein determining the quantization codebook to be applied to the encoded uplink control information includes: receiving one or more quantization codebooks according to an agreed quantization method, wherein training the encoder at the entity associated with the UE to encode uplink control information includes: training the encoder without quantization, wherein the content of the training data set includes the set of encoded and unquantized intermediate vectors.

[0131] 12. The method according to any one of clauses 1 to 4, wherein training the encoder to encode uplink control information includes: training the encoder with a first quantization codebook based on an agreed quantization method and quantization parameters, wherein the content of the training data set includes the set of encoded and unquantized intermediate vectors, and wherein determining the quantization codebook to be applied to the encoded uplink control information includes: receiving one or more second quantization codebooks according to the agreed quantization method and quantization parameters.

[0132] 13. The method according to any one of clauses 1 to 4, wherein determining the quantization codebook includes:

[0133] training the quantization codebook at the entity associated with the UE based on the set of encoded and unquantized intermediate vectors; and

[0134] receiving a finally refined quantization codebook from the entity associated with the base station.

[0135] 14. The method according to any one of clauses 1 to 4, wherein determining the quantization codebook to be applied to the encoded uplink control information includes:

[0136] training the quantization codebook at the entity associated with the UE based on the set of encoded and unquantized intermediate vectors; and

[0137] generating a finally refined quantization codebook based on a second quantization codebook from the entity associated with the base station or a reconstruction codebook from the entity associated with the base station.

[0138] 15. A method performed at an entity associated with a base station, including:

[0139] receiving a sequential training data set from at least a first entity associated with a UE, the sequential

[0140] training data set including:

[0141] either a set of input vectors or a set of output vectors; and

[0142] either a set of encoded and unquantized intermediate vectors or a set of encoded and quantized intermediate vectors;

[0143] determining a quantization codebook to be applied to encoded and quantized uplink control information; and

[0144] training a decoder to decode the encoded and quantized uplink control information based on the sequential training data set and the quantization codebook.

[0145] 16. The method according to clause 15 further includes: determining a level of consistency between a first UE vendor and a base station equipment vendor, wherein the training of the decoder, the determining of the quantization codebook, or the content of the sequential training dataset is based on the level of consistency.

[0146] 17. The method according to clause 16 further includes: deploying the decoder and the quantization codebook to one or more base stations for use with UEs of the first UE vendor.

[0147] 18. The method according to any one of clauses 15 to 17, wherein the training of the decoder is based on a loss function between the input vector set or the output vector set and an output vector set from the decoder applied to the unquantized intermediate vector set or the quantized intermediate vector set of the encoding.

[0148] 19. The method according to any one of clauses 15 to 18 further includes: receiving a sequential training dataset from an entity associated with a second UE, and wherein the decoder includes a first vendor-specific layer, a second vendor-specific layer, and a shared decoder layer.

[0149] 20. The method according to any one of clauses 15 to 19, wherein the quantizer is trained at the entity associated with the UE, and wherein the sequential training dataset includes the quantized intermediate vector set of the encoding.

[0150] 21. The method according to clause 20, wherein determining the quantization codebook includes: analyzing the quantized intermediate vector set of the encoding to determine that the quantized intermediate vector set of the encoding is quantized.

[0151] 22. The method according to clause 20, wherein the quantized intermediate vector set of the encoding is quantized based on an agreed quantization method and quantization parameters.

[0152] 23. The method according to clause 20 further includes: receiving one or more quantization codebooks or reconstruction codebooks from an entity associated with the UE.

[0153] 24. The method according to any one of clauses 15 to 19, wherein determining the quantization codebook to be applied to the quantized uplink control information of the encoding includes: training the quantization codebook with the decoder.

[0154] 25. The method according to clause 24, wherein training the quantizer with the decoder is based on an agreed quantization scheme and quantization parameters, and wherein the content of the training dataset includes the unquantized intermediate vector set of the encoding.

[0155] 26. The method according to clause 25 further includes: transmitting one or more quantization codebooks to at least an entity associated with the first UE according to the agreed-upon scheme and quantization parameters.

[0156] 27. The method according to any one of clauses 15 to 19, wherein determining a quantizer to be applied to the encoded and quantized uplink control information includes:

[0157] training the quantization codebook at an entity associated with the base station based on the set of encoded and unquantized intermediate vectors;

[0158] receiving one or more quantization codebooks or reconstruction codebooks from at least an entity associated with the first UE; and

[0159] transmitting a finally refined quantization codebook to at least an entity associated with the first UE.

[0160] 28. The method according to any one of clauses 15 to 19, wherein determining a quantizer to be applied to the encoded and quantized uplink control information includes:

[0161] training the quantizer at an entity associated with the base station based on the set of encoded and unquantized intermediate vectors;

[0162] transmitting one or more quantization codebooks or reconstruction codebooks to at least an entity associated with the first UE; and

[0163] receiving a finally refined quantization codebook from at least an entity associated with the first UE.

[0164] 29. An entity associated with a UE, comprising:

[0165] a memory storing computer-executable instructions; and

[0166] a processor configured to execute the instructions and cause the entity associated with the UE to perform the method according to any one of clauses 1 to 14.

[0167] 30. A base station, comprising:

[0168] a memory storing computer-executable instructions; and

[0169] a processor configured to execute the instructions and cause the base station to perform the method according to any one of clauses 15 to 28.

[0170] 31. A device for wireless communication, comprising: components for performing the method according to any one of Examples 1 to 14.

[0171] 32. An apparatus for wireless communication, comprising: components for performing the method according to any one of Examples 15 to 28.

[0172] 33. A non-transitory computer-readable medium, comprising: instructions that, when executed by a device, cause the device to perform the method according to any one of Examples 1 to 14.

[0173] 34. A non-transitory computer-readable medium, comprising: instructions that, when executed by a device, cause the device to perform the method according to any one of Examples 15 to 28.

[0174] The foregoing description is provided to enable any person skilled in the art to practice the various aspects described herein. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects. Thus, the claims are not intended to be limited to the aspects shown herein but are to be accorded the full scope consistent with the language of the claims, where elements are recited in the singular and are not intended to mean "one and only one" but rather "one or more" unless specifically stated otherwise. The word "exemplary" is used herein to mean "serving as an example, instance, or illustration." Any aspect described herein as "exemplary" is not necessarily to be construed as preferred or superior to other aspects. Unless specifically stated otherwise, the term "some" refers to one or more. Combinations such as "at least one of A, B, or C," "one or more of A, B, or C," "at least one of A, B, and C," "one or more of A, B, and C," and "any combination of A, B, C, or their combinations," including any combination of A, B, and / or C, may include multiple A's, multiple B's, or multiple C's. Specifically, combinations such as "at least one of A, B, or C," "one or more of A, B, or C," "at least one of A, B, and C," "one or more of A, B, and C," and "any combination of A, B, C, or their combinations" may be only A, only B, only C, A and B, A and C, B and C, or A and B and C, where any such combination may include one or more members of A, B, or C. All structural and functional equivalents of the elements of the various aspects described throughout this disclosure that are known or later will be known to those of ordinary skill in the art are expressly incorporated herein by reference and are intended to be covered by the claims. Furthermore, nothing disclosed herein is intended to be dedicated to the public, whether or not such disclosure is explicitly recited in the claims. The words "module," "mechanism," "element," "device," etc. do not substitute for the word "component." Thus, no claim element will be construed as a means-plus-function unless the element is expressly recited using the phrase "means for...".

Claims

1. A method performed at an entity associated with a UE, comprising: training an encoder to encode uplink control information; determining a quantization codebook to be applied to the encoded uplink control information; and sharing a sequential training dataset with an entity associated with a base station, the sequential training dataset comprising: either a set of input vectors or a set of output vectors; and either a set of encoded and unquantized intermediate vectors or a set of encoded and quantized intermediate vectors.

2. The method according to claim 1, further comprising: determining a level of consistency between a UE vendor and a base station equipment vendor, wherein the training of the encoder, the determining of the quantization codebook, or the content of the sequential training dataset is based on the level of consistency.

3. The method according to claim 2, further comprising: deploying the encoder and the quantization codebook to one or more UEs for use with a base station of a base station equipment vendor.

4. The method according to claim 1, wherein training the encoder is based on a loss function between the set of input vectors and a set of output vectors from a UE decoder.

5. The method according to claim 1, wherein determining the quantization codebook comprises: training the quantization codebook at the entity associated with the UE based on the set of encoded and unquantized intermediate vectors, wherein the sequential training dataset comprises the set of encoded and quantized intermediate vectors.

6. The method according to claim 5, further comprising: sharing one or more quantization codebooks with the entity associated with the base station.

7. The method according to claim 5, wherein training the quantization codebook at the entity associated with the UE based on the set of encoded and unquantized intermediate vectors comprises: selecting a quantization scheme.

8. The method according to claim 5, wherein training the quantization codebook at the entity associated with the UE based on the set of encoded and unquantized intermediate vectors comprises: training the quantization codebook using an agreed quantization method and quantization parameters.

9. The method according to claim 1, wherein determining the quantization codebook comprises: receiving one or more quantization codebooks from the entity associated with the base station, wherein training the encoder at the entity associated with the UE to encode uplink control information comprises: training the encoder without quantization.

10. The method according to claim 1, wherein training the encoder to encode uplink control information comprises: training the encoder with a first quantization codebook, and wherein determining the quantization codebook comprises: receiving a second quantization codebook from the entity associated with the base station.

11. The method according to claim 1, wherein determining the quantization codebook to be applied to the encoded uplink control information comprises: receiving one or more quantization codebooks according to an agreed quantization method, wherein training the encoder at the entity associated with the UE to encode uplink control information comprises: training the encoder without quantization, wherein the content of the training dataset comprises the set of encoded and unquantized intermediate vectors.

12. The method according to claim 1, wherein training the encoder to encode uplink control information comprises: training the encoder and a first quantization codebook based on an agreed quantization method and quantization parameters, wherein the content of the training dataset includes the set of encoded and unquantized intermediate vectors, and wherein determining the quantization codebook to be applied to the encoded uplink control information comprises: receiving one or more second quantization codebooks according to the agreed quantization method and quantization parameters.

13. The method according to claim 1, wherein determining the quantization codebook comprises: training the quantization codebook at the entity associated with the UE based on the set of encoded and unquantized intermediate vectors; and receiving a finally refined quantization codebook from the entity associated with the base station.

14. The method according to claim 1, wherein determining the quantization codebook to be applied to the encoded uplink control information comprises: training the quantization codebook at the entity associated with the UE based on the set of encoded and unquantized intermediate vectors; and generating a finally refined quantization codebook based on a second quantization codebook from the entity associated with the base station or a reconstruction codebook from the entity associated with the base station.

15. A method performed at an entity associated with a base station, comprises: receiving a sequential training dataset from at least an entity associated with a first UE, the sequential training dataset including: either a set of input vectors or a set of output vectors; and either a set of encoded and unquantized intermediate vectors or a set of encoded and quantized intermediate vectors; determining a quantization codebook to be applied to the encoded and quantized uplink control information; and training a decoder to decode the encoded and quantized uplink control information based on the sequential training dataset and the quantization codebook.

16. The method according to claim 15, further comprises: determining a level of consistency between a first UE vendor and a base station equipment vendor, wherein the training of the decoder, the determining of the quantization codebook, or the content of the sequential training dataset is based on the level of consistency.

17. The method according to claim 16, further comprises: deploying the decoder and the quantization codebook to one or more base stations for use with UEs of the first UE vendor.

18. The method according to claim 15, wherein training the decoder is based on a loss function between the set of input vectors or the set of output vectors and an output vector set from the decoder applied to the set of encoded and unquantized intermediate vectors or the set of encoded and quantized intermediate vectors.

19. The method according to claim 15, further comprises: receiving a sequential training dataset from an entity associated with a second UE, and wherein the decoder comprises a first vendor-specific layer, a second vendor-specific layer, and a shared decoder layer.

20. The method according to claim 15, wherein the quantizer is trained at the entity associated with the UE, and wherein the sequential training dataset includes the set of encoded and quantized intermediate vectors.

21. The method according to claim 20, wherein determining the quantization codebook comprises: analyzing the set of encoded and quantized intermediate vectors to determine that the set of encoded and quantized intermediate vectors is quantized.

22. The method according to claim 20, wherein the set of encoded and quantized intermediate vectors is quantized based on an agreed quantization method and quantization parameters.

23. The method according to claim 20, further comprises: receiving one or more quantization codebooks or reconstruction codebooks from an entity associated with the UE.

24. The method according to claim 15, wherein determining the quantization codebook to be applied to the encoded and quantized uplink control information comprises: training the quantization codebook with the decoder.

25. The method according to claim 24, wherein training the quantizer with the decoder is based on an agreed quantization scheme and quantization parameters, wherein the content of the training data set comprises the set of encoded and unquantized intermediate vectors.

26. The method according to claim 25, further comprises: transmitting one or more quantization codebooks to at least the entity associated with the first UE according to the agreed scheme and quantization parameters.

27. The method according to claim 15, wherein determining the quantizer to be applied to the encoded and quantized uplink control information comprises: training the quantization codebook at the entity associated with the base station based on the set of encoded and unquantized intermediate vectors; receiving one or more quantization codebooks or reconstruction codebooks from at least the entity associated with the first UE; and transmitting the finally refined quantization codebook to at least the entity associated with the first UE.

28. The method according to claim 15, wherein determining the quantizer to be applied to the encoded and quantized uplink control information comprises: training the quantizer at the entity associated with the base station based on the set of encoded and unquantized intermediate vectors; transmitting one or more quantization codebooks or reconstruction codebooks to at least the entity associated with the first UE; and receiving the finally refined quantization codebook from at least the entity associated with the first UE.

29. An entity associated with a UE, comprises: a memory storing computer-executable instructions; and a processor configured to execute the instructions and cause the entity associated with the UE to: train an encoder to encode uplink control information; determine a quantization codebook to be applied to the encoded uplink control information; and share an ordered training data set with an entity associated with a base station, the ordered training data set comprising: one of an input vector set or an output vector set; and one of an encoded and unquantized intermediate vector set or an encoded and quantized intermediate vector set.

30. A base station, comprises: a memory storing computer-executable instructions; and a processor configured to execute the instructions and cause the base station to: receive an ordered training data set from at least an entity associated with a first user equipment (UE), the ordered training data set comprising: one of an input vector set or an output vector set; and either the coded and unquantized set of intermediate vectors (ze) or the coded and quantized set of intermediate vectors (zq); determine a quantization codebook to be applied to the coded and quantized uplink control information; and train a decoder to decode the coded and quantized uplink control information based on the sequential training data set and the quantization codebook.

Images