Machine Learning to Address Transmit (Tx) Nonlinearities

Abstract

A method for wireless communication by a transmitting device transforms a transmit waveform by an encoder neural network to control power amplifier (PA) operation with respect to nonlinearity. The method also transmits the transformed transmit waveform over a propagation channel. A method for wireless communication by a receiving device receives the waveform transformed by the encoder neural network. The method also recovers encoder input symbols from the received waveform using a decoder neural network. The transmitting device for wireless communication calculates a distortion error based on the undistorted digital transmit waveform and the distorted digital transmit waveform. The transmitting device also compresses the distortion error using the encoder neural network of an autoencoder. The transmitting device transmits the compressed distortion error to the receiving device to compensate for the power amplifier (PA) nonlinearity.

Description

[Technical Field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This application claims priority to U.S. Provisional Patent Application No. 62 / 980,924, entitled "MACHINE LEARNING FOR ADDRESSING TRANSMIT (Tx) NON-LINEARITY," filed February 24, 2020, which claims priority to U.S. Provisional Patent Application No. 17 / 181,927, entitled "MACHINE LEARNING FOR ADDRESSING TRANSMIT (Tx) NON-LINEARITY," filed February 22, 2021, the disclosures of which are expressly incorporated herein by reference in their entireties.

[0002] Aspects of the present disclosure relate generally to wireless communications, and more particularly to machine learning to address transmit (Tx) nonlinearities. [Background technology]

[0003] Wireless communication systems are widely deployed to provide various telecommunication services such as telephony, video, data, messaging, and broadcasts. A typical wireless communication system may employ multiple-access technologies capable of supporting communication with multiple users by sharing available system resources (e.g., bandwidth, transmit power, etc.). Examples of such multiple-access technologies include code division multiple access (CDMA) systems, time division multiple access (TDMA) systems, frequency division multiple access (FDMA) systems, orthogonal frequency division multiple access (OFDMA) systems, single-carrier frequency division multiple access (SC-FDMA) systems, time division synchronous code division multiple access (TD-SCDMA) systems, and Long Term Evolution (LTE). LTE / LTE-Advanced is a set of extensions to the Universal Mobile Telecommunications System (UMTS) mobile standard promulgated by the 3rd Generation Partnership Project (3GPP®).

[0004] A wireless communication network may include several base stations (BSs) that can support communication for several user equipments (UEs). The user equipments (UEs) may communicate with the base stations (BSs) via downlinks and uplinks. The downlink (or forward link) refers to the communication link from the BSs to the UEs, and the uplink (or reverse link) refers to the communication link from the UEs to the BSs. As described in more detail herein, a BS may be referred to as a Node B, gNB, access point (AP), radio head, transmit / receive point (TRP), new radio (NR) BS, 5G Node B, etc.

[0005] The above multiple access technologies have been adopted in various telecommunications standards to provide common protocols that enable different user equipment to communicate at city, national, regional, and even global levels. New Radio (NR), sometimes referred to as 5G, is a set of enhancements to the LTE mobile standard promulgated by the Third Generation Partnership Project (3GPP®). NR is designed to increase spectral efficiency, reduce costs, improve service, utilize new spectrum, use Orthogonal Frequency Division Multiplexing (OFDM) with Cyclic Prefix (CP) (CP-OFDM) on the downlink (DL), and CP-OFDM and / or SC-FDM (e.g., also known as Discrete Fourier Transform Spread OFDM (DFT-s-OFDM)) on the uplink (UL) to better integrate with other open standards, as well as support beamforming, multiple-input multiple-output (MIMO) antenna technology, and carrier aggregation. However, as demand for mobile broadband access continues to grow, further improvements to NR and LTE technologies are needed. Preferably, these improvements should be applicable to other multiple access technologies and telecommunications standards that employ these technologies.

[0006] An artificial neural network may comprise an interconnected group of artificial neurons (e.g., neuron models). An artificial neural network may be a computing device or may be represented as a method to be executed by a computing device. A convolutional neural network, such as a deep convolutional neural network, is a type of feedforward artificial neural network. A convolutional neural network may include layers of neurons that may be arranged in tiled receptive fields. It would be desirable to apply neural network processing to wireless communications to achieve greater efficiency. Summary of the Invention [Means for solving the problem]

[0007] In some aspects of the present disclosure, a method of wireless communication by a transmitting device may include transforming a transmit waveform with an encoder neural network to control power amplifier (PA) operation with respect to nonlinearity. The method may also include transmitting the transformed transmit waveform over a propagation channel.

[0008] In another aspect of the present disclosure, a method of wireless communication by a receiving device may include receiving a waveform transformed by an encoder neural network, and recovering encoder input symbols from the received waveform using a decoder neural network.

[0009] In some aspects of the present disclosure, a method in a transmitting device may include calculating a distortion error based on an undistorted digital transmit waveform and a distorted digital transmit waveform. The method may also compress the distortion error using an encoder neural network of an autoencoder. The method may further include transmitting the compressed distortion error to a receiving device to compensate for power amplifier (PA) nonlinearity.

[0010] In yet another aspect of the present disclosure, a method in a receiving device may include decompressing a distortion error caused by a power amplifier (PA) using a decoder neural network of an autoencoder. The method may also include recovering an undistorted signal based on the decompressed distortion error.

[0011] In some aspects of the present disclosure, a transmitting device for wireless communication may include a memory and at least one processor operably coupled to the memory. The memory and the at least one processor may be configured to transform a transmit waveform by an encoder neural network to control power amplifier (PA) operation with respect to nonlinearity. The memory and the processor may also be configured to transmit the transformed transmit waveform over a propagation channel.

[0012] In some aspects of the present disclosure, a receiving device for wireless communication may include a memory and at least one processor operably coupled to the memory. The memory and the at least one processor may be configured to receive a waveform transformed by an encoder neural network. The memory and the processor may also be configured to recover encoder input symbols from the received waveform using a decoder neural network.

[0013] In yet another aspect of the present disclosure, a transmitting device for wireless communication may include a memory and at least one processor operably coupled to the memory. The memory and the at least one processor may be configured to calculate a distortion error based on an undistorted digital transmit waveform and a distorted digital transmit waveform. The memory and the processor may also be configured to compress the distortion error using an encoder neural network of an autoencoder and to transmit the compressed distortion error to a receiving device to compensate for power amplifier (PA) nonlinearity.

[0014] In some aspects of the present disclosure, a receiving device for wireless communication may include a memory and at least one processor operably coupled to the memory. The memory and the at least one processor may be configured to decompress a distortion error caused by a power amplifier (PA) using a decoder neural network of an autoencoder. The memory and the processor may also be configured to restore an undistorted signal based on the decompressed distortion error.

[0015] In another aspect of the present disclosure, a transmit device for wireless communication may include means for transforming a transmit waveform by an encoder neural network to control power amplifier (PA) operation with respect to nonlinearity. The transmit device may also include means for transmitting the transformed transmit waveform over a propagation channel.

[0016] In an aspect of the present disclosure, a receiving device for wireless communication may include means for receiving a waveform transformed by an encoder neural network. The receiving device may also include means for recovering encoder input symbols from the received waveform using a decoder neural network.

[0017] In some aspects of the present disclosure, a transmitting device for wireless communication may include means for calculating a distortion error based on an undistorted digital transmit waveform and a distorted digital transmit waveform. The transmitting device may also include means for compressing the distortion error using an encoder neural network of an autoencoder. The transmitting device may further include means for transmitting the compressed distortion error to a receiving device to compensate for power amplifier (PA) nonlinearity.

[0018] In yet another aspect of the present disclosure, a receiving device for wireless communication may include means for decompressing a distortion error caused by a power amplifier (PA) using a decoder neural network of an autoencoder. The receiving device may also include means for recovering an undistorted signal based on the decompressed distortion error.

[0019] In aspects of the present disclosure, a non-transitory computer-readable medium may include program code executed by a user equipment (UE). The medium may include program code for transforming a transmit waveform by an encoder neural network to control power amplifier (PA) operation with respect to nonlinearity. The medium may also include program code for transmitting the transformed transmit waveform over a propagation channel.

[0020] In some aspects of the present disclosure, a non-transitory computer-readable medium may include program code executed by a receiving device. The medium may include program code for receiving a waveform transformed by an encoder neural network. The medium may also include program code for recovering encoder input symbols from the received waveform using a decoder neural network.

[0021] In a further aspect of the present disclosure, a non-transitory computer-readable medium may include program code executed by a transmitting device. The medium may include program code for calculating a distortion error based on an undistorted digital transmit waveform and a distorted digital transmit waveform. The medium may also include program code for compressing the distortion error using an encoder neural network of an autoencoder. The medium may further include program code for transmitting the compressed distortion error to a receiving device to compensate for power amplifier (PA) nonlinearity.

[0022] In yet a further aspect of the present disclosure, a non-transitory computer-readable medium may include program code executed by a receiving device. The medium may include program code for decompressing distortion errors caused by a power amplifier (PA) using a decoder neural network of an autoencoder. The medium may also include program code for restoring an undistorted signal based on the decompressed distortion errors.

[0023] Aspects generally include methods, apparatus, systems, computer program products, non-transitory computer-readable media, user equipment, base stations, wireless communication devices, and processing systems, as fully described with reference to and as illustrated by the accompanying drawings and this specification.

[0024] The foregoing has outlined rather broadly the features and technical advantages of examples according to the present disclosure so that the following detailed description may be better understood. Additional features and advantages are described below. The concepts and examples disclosed may be readily utilized as a basis for modifying or designing other structures for carrying out the same purposes of the present disclosure. Such equivalent structures do not depart from the scope of the appended claims. The nature of the concepts disclosed herein, both their organization and method of operation, together with associated advantages, will be better understood from the following description when considered in connection with the accompanying figures. Each of the figures is provided for the purpose of illustration and description, and not as a definition of the limits of the claims.

[0025] So that the above-mentioned features of the present disclosure may be understood in detail, a more detailed description briefly summarized above may be had by reference to embodiments, some of which are shown in the accompanying drawings. It should be noted, however, that the present description may admit of other equally effective embodiments, and therefore, that the accompanying drawings illustrate only some typical embodiments of the present disclosure and should not be considered limiting of its scope. The same reference numbers in different drawings may identify the same or similar elements. [Brief explanation of the drawings]

[0026] [Figure 1] FIG. 1 is a block diagram conceptually illustrating an example of a wireless communication network, in accordance with various aspects of the present disclosure. [Figure 2] FIG. 1 is a block diagram conceptually illustrating an example of a base station in communication with user equipment (UE) in a wireless communication network, in accordance with various aspects of the present disclosure. [Figure 3] FIG. 1 illustrates an exemplary implementation of designing a neural network using a system-on-chip (SOC) that includes a general-purpose processor, according to some aspects of the present disclosure. [Figure 4A] FIG. 1 illustrates a neural network according to an aspect of the present disclosure. [Figure 4B] FIG. 1 illustrates a neural network according to an aspect of the present disclosure. [Figure 4C] FIG. 1 illustrates a neural network according to an aspect of the present disclosure. [Figure 4D] FIG. 1 illustrates an exemplary deep convolutional network (DCN), according to aspects of the present disclosure. [Figure 5] FIG. 1 is a block diagram illustrating an example deep convolutional network (DCN), according to aspects of the present disclosure. [Figure 6] FIG. 1 is a block diagram of an example architecture for implementing a first solution for transforming a transmit waveform, according to aspects of the present disclosure. [Figure 7] FIG. 1 is a block diagram of an example architecture for transforming a transmit waveform, according to aspects of the present disclosure. [Figure 8] FIG. 1 is a block diagram of an example architecture for transforming a transmit waveform, according to aspects of the present disclosure. [Figure 9] FIG. 1 is a block diagram of an example architecture for transforming a transmit waveform, according to aspects of the present disclosure. [Figure 10] FIG. 1 is a block diagram of an example architecture for transforming a transmit waveform, according to aspects of the present disclosure. [Figure 11]FIG. 1 is a block diagram of an example architecture for transforming a transmit waveform, according to aspects of the present disclosure. [Figure 12] 1 illustrates an example process performed, for example, by a transmitting device, in accordance with various aspects of the present disclosure. [Figure 13] 1 illustrates an exemplary process performed, for example, by a receiving device, according to various aspects of the present disclosure. [Figure 14] 1 illustrates an example process performed, for example, by a transmitting device, in accordance with various aspects of the present disclosure. [Figure 15] 1 illustrates an exemplary process performed, for example, by a receiving device, according to various aspects of the present disclosure. DETAILED DESCRIPTION OF THE INVENTION

[0027] Various aspects of the present disclosure are described more fully below with reference to the accompanying drawings. However, the present disclosure may be embodied in many different forms and should not be construed as limited to any specific structure or function presented throughout this disclosure. Rather, these aspects are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art. Based on the teachings herein, those skilled in the art will appreciate that the scope of the present disclosure encompasses any aspect of the present disclosure disclosed herein, whether implemented independently or in combination with any other aspect of the present disclosure. For example, an apparatus may be implemented or a method may be practiced using any number of the described aspects. In addition, the scope of the present disclosure encompasses such apparatuses or methods practiced using other structure, functionality, or structure and functionality in addition to or other than the various aspects of the present disclosure described. It should be understood that any aspect of the present disclosure disclosed may be embodied by one or more elements of a claim.

[0028] Several aspects of telecommunications systems are now presented with reference to various apparatus and techniques. These apparatus and techniques are described in the detailed description that follows and illustrated in the accompanying drawings by various blocks, modules, components, circuits, steps, processes, algorithms, etc. (collectively referred to as "elements"). These elements may be implemented using hardware, software, or a combination thereof. Whether such elements are implemented as hardware or software depends on the particular application and design constraints imposed on the overall system.

[0029] Although aspects may be described using terminology generally associated with 5G and later wireless technologies, it should be noted that aspects of the present disclosure may be applied in other generation-based communication systems such as and including 3G and / or 4G technologies.

[0030] 1 is a diagram illustrating a network 100 in which aspects of the present disclosure may be practiced. Network 100 may be a 5G or NR network, or some other wireless network, such as an LTE network. Wireless network 100 may include several BSs 110 (shown as BS 110a, BS 110b, BS 110c, and BS 110d) and other network entities. A BS is an entity that communicates with user equipment (UE) and may also be referred to as a base station, NR BS, Node B, gNB, 5G Node B (NB), access point, transmit / receive point (TRP), etc. Each BS may provide communication coverage for a particular geographic area. In 3GPP, the term "cell" can refer to the coverage area of a BS and / or the BS subsystem serving this coverage area, depending on the context in which the term is used.

[0031] A BS may provide communication coverage for a macrocell, a picocell, a femtocell, and / or another type of cell. A macrocell may cover a relatively large geographic area (e.g., a few kilometers in radius) and may allow unrestricted access by UEs with service subscriptions. A picocell may cover a relatively small geographic area and may allow unrestricted access by UEs with service subscriptions. A femtocell may cover a relatively small geographic area (e.g., a home) and may allow restricted access by UEs that have an association with the femtocell (e.g., UEs in a Closed Subscriber Group (CSG)). A BS for a macrocell may be referred to as a macro BS. A BS for a picocell may be referred to as a pico BS. A BS for a femtocell may be referred to as a femto BS or a home BS. 1, BS 110a may be a macro BS for a macro cell 102a, BS 110b may be a pico BS for a pico cell 102b, and BS 110c may be a femto BS for a femto cell 102c. A BS may support one or more (e.g., three) cells. The terms “eNB,” “base station,” “NR BS,” “gNB,” “TRP,” “AP,” “Node B,” “5G NB,” and “cell” may be used interchangeably herein.

[0032] In some aspects, the cells may not necessarily be stationary, and the geographic area of the cell may move according to the location of the mobile BS. In some aspects, the BSs may be interconnected to each other and / or to one or more other BSs or network nodes (not shown) in wireless network 100 through various types of backhaul interfaces, such as direct physical connections, virtual networks, etc., using any suitable transport network.

[0033] Wireless network 100 may also include relay stations. A relay station is an entity that can receive data transmissions from an upstream station (e.g., a BS or UE) and send the data transmissions to a downstream station (e.g., a UE or BS). A relay station may also be a UE that can relay transmissions for other UEs. In the example shown in FIG. 1, relay station 110d may communicate with macro BS 110a and UE 120d to facilitate communication between BS 110a and UE 120d. A relay station may also be called a relay BS, a relay base station, a relay, etc.

[0034] Wireless network 100 may be a heterogeneous network including different types of BSs, e.g., macro BSs, pico BSs, femto BSs, relay BSs, etc. These different types of BSs may have different transmit power levels, different coverage areas, and different impacts on interference in wireless network 100. For example, a macro BS may have a high transmit power level (e.g., 5-40 watts), while a pico BS, femto BS, and relay BS may have a lower transmit power level (e.g., 0.1-2 watts).

[0035] A network controller 130 may couple to a set of BSs and provide coordination and control for these BSs. The network controller 130 may communicate with the BSs via a backhaul. The BSs may also communicate with each other directly or indirectly, e.g., via wireless or wireline backhaul.

[0036] The UEs 120 (e.g., 120a, 120b, 120c) may be dispersed throughout the wireless network 100, and each UE may be fixed or mobile. A UE may also be referred to as an access terminal, terminal, mobile station, subscriber unit, station, etc. A UE may be a cellular phone (e.g., a smartphone), a personal digital assistant (PDA), a wireless modem, a wireless communication device, a handheld device, a laptop computer, a cordless phone, a wireless local loop (WLL) station, a tablet, a camera, a gaming device, a netbook, a smartbook, an ultrabook, a medical device or equipment, a biometric sensor / device, a wearable device (smart watch, smart clothing, smart glasses, smart wristband, smart jewelry (e.g., smart ring, smart bracelet)), an entertainment device (e.g., a music or video device, or satellite radio), a vehicle component or sensor, a smart meter / sensor, industrial manufacturing equipment, a global positioning system device, or any other suitable device configured to communicate over a wireless or wired medium.

[0037] Some UEs may be considered machine type communication (MTC) UEs or evolved or enhanced machine type communication (eMTC) UEs. MTC UEs and eMTC UEs include, for example, a robot, a drone, a remote device, a sensor, a meter, a monitor, a location tag, etc. that may communicate with a base station, another device (e.g., a remote device), or some other entity. A wireless node may provide connectivity for or to a network (e.g., a wide area network such as the Internet or a cellular network), for example, via a wired or wireless communication link. Some UEs may be considered Internet of Things (IoT) devices and / or may be implemented as narrowband Internet of Things (NB-IoT) devices. Some UEs may be considered customer premises equipment (CPE). The UE 120 may be included inside a housing that houses components of the UE 120, such as processor components, memory components, etc.

[0038] In general, any number of wireless networks may be deployed within a given geographic area. Each wireless network may support a particular RAT and may operate on one or more frequencies. A RAT may also be referred to as a radio technology, air interface, etc. A frequency may also be referred to as a carrier, frequency channel, etc. Each frequency may support a single RAT within a given geographic area to avoid interference between wireless networks of different RATs. In some cases, an NR RAT network or a 5G RAT network may be deployed.

[0039] In some aspects, two or more UEs 120 (e.g., shown as UE 120a and UE 120e) may communicate directly (e.g., without using a base station 110 as an intermediary for communicating with each other) using one or more sidelink channels. For example, the UEs 120 may communicate using peer-to-peer (P2P) communication, device-to-device (D2D) communication, a vehicle-to-everything (V2X) protocol (which may include, e.g., a vehicle-to-vehicle (V2V) protocol, a vehicle-to-infrastructure (V2I) protocol, etc.), a mesh network, etc. In this case, the UEs 120 may perform scheduling operations, resource selection operations, and / or other operations described elsewhere herein as being performed by the base station 110.

[0040] As noted above, Figure 1 is provided as an example only. Other examples may differ from those described with respect to Figure 1.

[0041] 2 shows a block diagram of a design 200 of a base station 110 and a UE 120, which may be one of the base stations and one of the UEs in FIG. 1. Base station 110 may be equipped with T antennas 234a through 234t, and UE 120 may be equipped with R antennas 252a through 252r, where in general T≧1 and R≧1.

[0042] At base station 110, transmit processor 220 may receive data for one or more UEs from data source 212, select one or more modulation and coding schemes (MCSs) for each UE based at least in part on a channel quality indicator (CQI) received from the UE, process (e.g., encode and modulate) the data for each UE based at least in part on the MCS selected for the UE, and provide data symbols to all UEs. Transmit processor 220 may also process system information (e.g., for semi-static resource partitioning information (SRPI), etc.) and control information (e.g., CQI requests, grants, upper layer signaling, etc.) and provide overhead symbols and control symbols. Transmit processor 220 may also generate reference symbols for reference signals (e.g., cell-specific reference signals (CRSs)) and synchronization signals (e.g., primary synchronization signals (PSSs) and secondary synchronization signals (SSSs)). The transmit (TX) multiple-input multiple-output (MIMO) processor 230 may perform spatial processing (e.g., precoding) on the data symbols, control symbols, overhead symbols, and / or reference symbols, if applicable, and may provide T output symbol streams to T modulators (MODs) 232a through 232t. Each modulator 232 may process a respective output symbol stream (e.g., for OFDM, etc.) to obtain an output sample stream. Each modulator 232 may further process (e.g., convert to analog, amplify, filter, and upconvert) the output sample stream to obtain a downlink signal. The T downlink signals from modulators 232a through 232t may be transmitted via T antennas 234a through 234t, respectively. According to various aspects described in more detail below, synchronization signals may be generated using location coding to convey additional information.

[0043] At UE 120, antennas 252a through 252r may receive downlink signals from base station 110 and / or other base stations and may provide received signals to demodulators (DEMODs) 254a through 254r, respectively. Each demodulator 254 may condition (e.g., filter, amplify, downconvert, and digitize) its received signal to obtain input samples. Each demodulator 254 may further process the input samples (e.g., for OFDM, etc.) to obtain received symbols. A MIMO detector 256 may obtain received symbols from all R demodulators 254a through 254r, perform MIMO detection on the received symbols, if applicable, and provide detected symbols. A receive processor 258 may process (e.g., demodulate and decode) the detected symbols and provide decoded data for UE 120 to a data sink 260 and may provide decoded control and system information to controller / processor 280. The channel processor may determine a reference signal received power (RSRP), a received signal strength indicator (RSSI), a reference signal received quality (RSRQ), a channel quality indicator (CQI), etc. In some aspects, one or more components of the UE 120 may be included in a housing.

[0044] On the uplink, at the UE 120, the transmit processor 264 may receive and process data from the data source 262 and control information from the controller / processor 280 (e.g., for reports including RSRP, RSSI, RSRQ, CQI, etc.). The transmit processor 264 may also generate reference symbols for one or more reference signals. The symbols from the transmit processor 264 may be precoded by a TX MIMO processor 266 if applicable, further processed by modulators 254a through 254r (e.g., for DFT-s-OFDM, CP-OFDM, etc.), and transmitted to the base station 110. At the base station 110, uplink signals from the UE 120 and other UEs may be received by the antennas 234, processed by the demodulator 232, detected by a MIMO detector 236 if applicable, and further processed by the receive processor 238 to obtain decoded data and control information sent by the UE 120. The receive processor 238 may provide the decoded data to a data sink 239 and the decoded control information to the controller / processor 240. The base station 110 may include a communication unit 244 and may communicate with the network controller 130 via the communication unit 244. The network controller 130 may include a communication unit 294, a controller / processor 290, and a memory 292.

[0045] The controller / processor 240 of the base station 110, the controller / processor 280 of the UE 120, and / or any other components of FIG. 2 may perform one or more techniques related to machine learning for nonlinearities, as described in more detail elsewhere, the disclosures of which are incorporated herein by reference in their entireties. For example, the controller / processor 240 of the base station 110, the controller / processor 280 of the UE 120, and / or any other components of FIG. 2 may perform or direct the operation of, for example, processes 1200, 1300, 1400, 1500 of FIGS. 12-15, and / or other processes as described. The memory 242 and the memory 282 may store data and program codes for the base station 110 and the UE 120, respectively. The scheduler 246 may schedule UEs for data transmission on the downlink and / or uplink.

[0046] In some aspects, the UE 120 may include means for receiving, means for transmitting, means for calculating, means for compressing, means for decompressing, means for restoring, and means for converting. Such means may include one or more components of the UE 120 or the base station 110 described with respect to FIG.

[0047] As noted above, Figure 2 is provided as an example only. Other examples may differ from those described with respect to Figure 2.

[0048] In some cases, different types of devices supporting different types of applications and / or services may coexist in a cell. Examples of different types of devices include UE handsets, customer premises equipment (CPE), vehicles, Internet of Things (IoT) devices, etc. Examples of different types of applications include ultra-reliable low latency communications (URLLC) applications, massive machine-type communications (mMTC) applications, enhanced mobile broadband (eMBB) applications, vehicle-to-anything (V2X) applications, etc. Furthermore, in some cases, a single device may support different applications or services simultaneously.

[0049] 3 illustrates an example implementation of a system-on-chip (SOC) 300 that may include a central processing unit (CPU) 302 or a multi-core CPU configured for machine learning to address nonlinearities in accordance with some aspects of the present disclosure. The SOC 300 may be included in a base station 110 or a UE 120. Variables (e.g., neural signals and synaptic weights), system parameters associated with a computational device (e.g., a neural network with weights), delays, frequency bin information, and task information may be stored in a memory block associated with a neural processing unit (NPU) 308, a memory block associated with the CPU 302, a memory block associated with a graphics processing unit (GPU) 304, a memory block associated with a digital signal processor (DSP) 306, a memory block 318, or distributed across multiple blocks. Instructions executed in the CPU 302 may be loaded from a program memory associated with the CPU 302 or from the memory block 318.

[0050] The SOC 300 may also include a connectivity block 310 that may include a GPU 304, a DSP 306, fifth-generation (5G) connectivity, fourth-generation long-term evolution (4G LTE) connectivity, Wi-Fi connectivity, USB connectivity, Bluetooth connectivity, etc., as well as additional processing blocks adapted for specific functions, such as a multimedia processor 312 that may detect and recognize gestures. In one implementation, the NPU is implemented in a CPU, DSP, and / or GPU. The SOC 300 may also include a sensor processor 314, an image signal processor (ISP) 316, and / or a navigation module 320 that may include a global positioning system.

[0051] The SOC 300 may be based on the ARM instruction set. In one aspect of the present disclosure, instructions loaded into the general-purpose processor 302 may include code for transforming a transmit waveform with an encoder neural network to control power amplifier (PA) operation with respect to nonlinearity and code for transmitting the transformed transmit waveform over a propagation channel. The instructions loaded into the general-purpose processor 302 may further include code for receiving the waveform transformed by the encoder neural network and code for recovering encoder input symbols from the received waveform with a decoder neural network. The instructions loaded into the general-purpose processor 302 may include code for calculating a distortion error based on the undistorted digital transmit waveform and the distorted digital transmit waveform, code for compressing the distortion error with the encoder neural network of an autoencoder, and code for transmitting the compressed distortion error to a receiving device to compensate for power amplifier (PA) nonlinearity. The instructions loaded into the general-purpose processor 302 may also include code for decompressing distortion errors caused by a power amplifier (PA) using a decoder neural network of the autoencoder, and code for recovering an undistorted signal based on the decompressed distortion errors.

[0052] Deep learning architectures can perform object recognition tasks by learning to represent inputs at successively higher levels of abstraction at each layer, thereby establishing useful feature representations of the input data. In this way, deep learning addresses a major bottleneck of traditional machine learning. Prior to the advent of deep learning, machine learning approaches to object recognition problems may have relied heavily on human-designed features, possibly in combination with shallow classifiers. A shallow classifier may be, for example, a two-class linear classifier in which a weighted sum of feature vector components may be compared to a threshold to predict which class the input belongs to. Human-designed features may be templates or kernels adapted to a particular problem domain by an engineer with domain expertise. In contrast, deep learning architectures may learn, but through training, to represent features similar to those a human engineer might design. Furthermore, deep networks may learn to represent and recognize new types of features that humans may not have considered.

[0053] Deep learning architectures may learn a hierarchy of features. For example, when presented with visual data, a first layer may learn to recognize relatively simple features, such as edges, in the input stream. In another example, when presented with auditory data, the first layer may learn to recognize spectral power at specific frequencies. A second layer may take the output of the first layer as input and learn to recognize combinations of features, such as simple shapes in the case of visual data, or combinations of sounds in the case of auditory data. For example, higher layers may learn to represent complex shapes in visual data or words in auditory data. Even higher layers may learn to recognize common visual objects or spoken phrases.

[0054] Deep learning architectures can perform particularly well when applied to problems that have a natural hierarchical structure. For example, motor vehicle classification can benefit from first learning to recognize wheels, windshields, and other features. These features can be combined in different ways in higher layers to recognize cars, trucks, and airplanes.

[0055] Neural networks can be designed with a variety of connectivity patterns. In feedforward networks, information is passed from lower to higher layers, with each neuron in a given layer transmitting to neurons in higher layers. A hierarchical representation can be established in successive layers of a feedforward network, as described above. Neural networks can also have recurrent or feedback (also called top-down) connections. In recurrent connections, the output from a neuron in a given layer can be transmitted to another neuron in the same layer. Recurrent architectures can be useful in recognizing patterns that span two or more of the input data chunks fed to the neural network in sequence. Connections from neurons in a given layer to neurons in lower layers are called feedback (or top-down) connections. Networks with numerous feedback connections can be useful when recognizing high-level concepts can help distinguish certain low-level features of the input.

[0056] The connections between layers of a neural network can be fully connected or locally connected. FIG. 4A shows an example of a fully connected neural network 402. In a fully connected neural network 402, a neuron in a first layer can transmit its output to every neuron in a second layer, such that each neuron in the second layer receives input from every neuron in the first layer. FIG. 4B shows an example of a locally connected neural network 404. In a locally connected neural network 404, a neuron in a first layer can be connected to a limited number of neurons in the second layer. More generally, the locally connected layers of a locally connected neural network 404 can be configured such that each neuron in a layer has the same or similar connectivity pattern, but with connection strengths that can have different values (e.g., 410, 412, 414, and 416). The connectivity patterns of local connections can give rise to spatially distinct receptive fields in higher layers because higher layer neurons in a given region may receive inputs that are tuned through training to the properties of a limited subset of all inputs to the network.

[0057] One example of a locally connected neural network is a convolutional neural network. Figure 4C shows an example of a convolutional neural network 406. The convolutional neural network 406 may be configured such that the connection strengths associated with the inputs for each neuron in the second layer are shared (e.g., 408). Convolutional neural networks may be suitable for problems in which the spatial location of the inputs is meaningful.

[0058] One type of convolutional neural network is the deep convolutional network (DCN). Figure 4D shows a detailed example of a DCN 400 designed to recognize visual features from input images 426 from an image capture device 430, such as an in-car camera. The DCN 400 in this example may be trained to identify traffic signs and numbers given on traffic signs. Of course, the DCN 400 may be trained for other tasks, such as recognizing lane markings or identifying traffic signals.

[0059] The DCN 400 may be trained using supervised learning. During training, the DCN 400 may be presented with an image, such as an image 426 of a speed limit sign, and then a forward pass may be computed to produce the output 422. The DCN 400 may include a feature extraction section and a classification section. Upon receiving the image 426, the convolutional layer 432 may apply a convolutional kernel (not shown) to the image 426 to generate the first set of feature maps 418. As an example, the convolutional kernel for the convolutional layer 432 may be a 5×5 kernel that generates 28×28 feature maps. In this example, four different feature maps are generated in the first set of feature maps 418, so four different convolutional kernels were applied to the image 426 in the convolutional layer 432. A convolutional kernel may also be referred to as a filter or a convolutional filter.

[0060] The first set of feature maps 418 may be subsampled by a max pooling layer (not shown) to generate a second set of feature maps 420. The max pooling layer reduces the size of the first set of feature maps 418. That is, the size of the second set of feature maps 420, such as 14×14, is less than the size of the first set of feature maps 418, such as 28×28. The reduced size provides similar information to subsequent layers while reducing memory consumption. The second set of feature maps 420 may be further convolved through one or more subsequent convolutional layers (not shown) to generate one or more subsequent sets of feature maps (not shown).

[0061] 4D , the second set of feature maps 420 are convolved to generate a first feature vector 424. Furthermore, the first feature vector 424 is further convolved to generate a second feature vector 428. Each feature in the second feature vector 428 may include a number corresponding to a possible feature of the image 426, such as "sign," "60," and "100." A softmax function (not shown) may convert the numbers in the second feature vector 428 into probabilities. Thus, the output 422 of the DCN 400 is the probability that the image 426 contains one or more features.

[0062] In this example, the probabilities in output 422 for "sign" and "60" are higher than the probabilities for other of the outputs 422, such as "30," "40," "50," "70," "80," "90," and "100." Prior to training, the output 422 produced by DCN 400 may be inaccurate. Thus, an error may be calculated between output 422 and a target output. The target output is the ground truth of image 426 (e.g., "sign" and "60"). The weights of DCN 400 may then be adjusted so that output 422 of DCN 400 is more closely aligned with the target output.

[0063] To adjust the weights, the learning algorithm may calculate a gradient vector for the weights. The gradient may indicate the amount by which the error would increase or decrease if the weights were adjusted. In the top layer, the gradient may correspond directly to the values of the weights connecting activated neurons in the penultimate layer to neurons in the output layer. In lower layers, the gradient may depend on the values of the weights and on the calculated error gradients of the upper layers. The weights may then be adjusted to reduce the error. This method of adjusting weights is sometimes called "backpropagation" because it involves a "backward pass" through the neural network.

[0064] In practice, the error gradient of the weights may be calculated over a small number of examples so that the calculated gradient approximates the true error gradient. This approximation method is sometimes called stochastic gradient descent. Stochastic gradient descent may be repeated until the achievable error rate of the overall system no longer decreases or until the error rate reaches a target level. After training, the DCN may be presented with a new image (e.g., the speed limit sign in image 426), and a forward pass through the network may produce output 422, which may be considered the DCN's inference or prediction.

[0065] A deep belief network (DBN) is a probabilistic model with multiple layers of hidden nodes. DBNs can be used to extract hierarchical representations of a training dataset. DBNs can be obtained by stacking layers of restricted Boltzmann machines (RBMs). RBMs are a type of artificial neural network that can learn probability distributions over a set of inputs. Because RBMs can learn probability distributions when there is no information about the class into which each input should be categorized, RBMs are often used in unsupervised learning. Using a hybrid unsupervised and supervised paradigm, the bottom RBM of a DBN can be trained in an unsupervised manner and act as a feature extractor, and the top RBM can be trained in a supervised manner (on the joint distribution of inputs from previous layers and the target class) and act as a classifier.

[0066] Deep convolutional networks (DCNs) are networks of convolutional networks constructed with additional pooling and normalization layers. DCNs have achieved state-of-the-art performance in many tasks. DCNs can be trained using supervised learning, where both the input and output targets are known for a large number of examples and are used to modify the network weights using gradient descent.

[0067] The DCN may be a feedforward network. Additionally, as described above, connections from neurons in the first layer of the DCN to groups of neurons in the next higher layer are shared across the neurons in the first layer. The feedforward and shared connections of the DCN may be exploited for high-speed processing. The computational burden of the DCN may be much smaller than that of a similarly sized neural network with, for example, recurrent or feedback connections.

[0068] The processing at each layer of a convolutional network can be viewed as a spatially invariant template or basis projection. If the input is first decomposed into multiple channels, such as the red, green, and blue channels of a color image, a convolutional network trained on that input can be viewed as three-dimensional, with two spatial dimensions along the image axes and a third dimension capturing color information. The output of the convolutional connections can be viewed as forming a feature map in a subsequent layer, with each element of the feature map (e.g., 420) receiving input from a range of neurons in the previous layer (e.g., feature map 418) and from each of multiple channels. Values in the feature map can be further processed using nonlinearities such as rectification, max(0,x), etc. Values from neighboring neurons can be further pooled, which corresponds to downsampling and can provide additional local invariance and dimensionality reduction. Normalization, corresponding to whitening, can also be applied through lateral inhibition between neurons in the feature map.

[0069] The performance of deep learning architectures can improve as more labeled data points become available or as computational power increases. Modern deep neural networks are routinely trained using computational resources thousands of times greater than those available to the average researcher just 15 years ago. New architectures and training paradigms can further improve deep learning performance. Rectified linear units can reduce the training problem known as vanishing gradients. New training techniques can reduce overfitting, thus enabling larger models to achieve better generalization. Encapsulation techniques can abstract data within a given receptive field, further improving overall performance.

[0070] 5 is a block diagram illustrating a deep convolutional network 550. The deep convolutional network 550 may include multiple different types of layers based on connectivity and weight sharing. As shown in FIG. 5, the deep convolutional network 550 includes convolutional blocks 554A, 554B. Each of the convolutional blocks 554A, 554B may be configured with a convolutional layer (CONV) 556, a normalization layer (LNorm) 558, and a max pooling layer (MAX POOL) 560.

[0071] The convolutional layer 556 may include one or more convolutional filters that may be applied to input data to generate feature maps. While only two of the convolutional blocks, 554A, 554B, are shown, the present disclosure is not so limited; instead, any number of convolutional blocks 554A, 554B may be included in the deep convolutional network 550 according to design preference. The normalization layer 558 may normalize the outputs of the convolutional filters. For example, the normalization layer 558 may provide whitening or lateral inhibition. The max-pooling layer 560 may provide downsampling aggregation across space for local invariance and dimensionality reduction.

[0072] For example, a parallel filter bank of a deep convolutional network may be loaded onto the CPU 302 or GPU 304 of the SOC 300 to achieve high performance and low power consumption. In alternative embodiments, the parallel filter bank may be loaded onto the DSP 306 or ISP 316 of the SOC 300. Additionally, the deep convolutional network 550 may access other processing blocks that may be present on the SOC 300, such as the sensor processor 314 and navigation module 320, which are dedicated to sensors and navigation, respectively.

[0073] The deep convolutional network 550 may also include one or more fully connected layers 562 (FC1 and FC2). The deep convolutional network 550 may further include a logistic regression (LR) layer 564. Between each layer 556, 558, 560, 562, 564 of the deep convolutional network 550 are weights (not shown) that are to be updated. The output of each of the layers (e.g., 556, 558, 560, 562, 564) may serve as an input for a subsequent one of the layers (e.g., 556, 558, 560, 562, 564) in the deep convolutional network 550 to learn a hierarchical feature representation from the input data 552 (e.g., image, audio, video, sensor data, and / or other input data) provided at the first one 554A of the convolutional blocks. The output of the deep convolutional network 550 is a classification score 566 for the input data 552. The classification score 566 may be a set of probabilities, where each probability is a probability of the input data including a feature from the set of features.

[0074] As noted above, Figures 3-5 are provided as examples. Other examples may differ from those described with respect to Figures 3-5.

[0075] Artificial intelligence (AI) / machine learning (ML) algorithms can improve wireless communications. AI / ML modules can run in the UE, the base station, or jointly across the UE and base station in the case of distributed algorithms. In an autoencoder scenario, joint training occurs across the UE and base station.

[0076] By jointly designing neural networks in the transmitter and receiver, nonlinearities in the transmitter (e.g., from the power amplifier (PA)) can be mitigated while efficiently utilizing the available transmit power. Prior art systems deal with nonlinearities by lowering the power in the transmitter, which reduces power efficiency because not all power is used.

[0077] According to this disclosure, the encoder network resides in a transmitter (e.g., a user equipment (UE)) and the decoder network resides in a receiver (e.g., a gNB). The decoder in the receiver recovers the encoder input.

[0078] According to one aspect of the present disclosure, a first solution employs a neural network to transform the transmit waveform to ensure that the power amplifier (PA) operates in a linear region. In other words, the encoder and decoder neural networks are designed to transform the waveform so that the PA operates in a linear region. The encoding neural network process clips the waveform in the transmitter chain to ensure that the waveform's amplitude does not exceed a threshold. Thus, the encoder anticipates nonlinearities in the PA and allows the decoder to restore accordingly.

[0079] Because PA nonlinearity is avoided, training of both the encoder and decoder can be done at either the transmitter or the receiver. These two approaches are described below. In both approaches, the PA nonlinearity does not need to be considered in the training, i.e., the nonlinearity does not need to be modeled.

[0080] The first approach is transmitter (e.g., UE) driven. In this approach, the transmitter discovers / generates the encoder and decoder. The transmitter conveys the decoder information to the receiver. Synthetic data for the channel is used in training because the actual channel does not exist from the encoder's point of view. A multipath propagation channel model is adopted to train the neural network. In this case, the channel model is that of a linear channel.

[0081] The second approach is receiver (e.g., gNB) driven. In the second approach, the receiver discovers / generates the encoder and decoder and signals the encoder to the transmitter. Actual propagation channels can be observed, measured, and used for training, rather than synthetic channel data. Alternatively, artificial channel data can be used.

[0082] There are two options for encoder / decoder location: In the first option, processing is done in the time domain. Therefore, the encoder is placed after the Inverse Fast Fourier Transform (IFFT), which converts from the frequency domain to the time domain. The decoder is placed before the Fast Fourier Transform (FFT), which converts from the time domain to the frequency domain.

[0083] In the second option, the processing is done in the frequency domain, so the encoder is placed before the IFFT processing and the decoder is placed after the FFT processing.

[0084] 6 is a block diagram of an example architecture 600 for implementing this first solution according to an aspect of the present disclosure. The architecture 600 includes an encoder neural network in a transmitter 610 and a decoder neural network in a receiver 620. The transmitter 610 communicates to the receiver 620 over a propagation channel 630. In the example of FIG. 6, there is no MIMO precoding of the transmit waveform.

[0085] Information bits to be transmitted to the receiver 620 are input to a channel encoder / modulator 640, which may be the same as the modulators 232, 254 of FIG. 2. The information bits may be any number of bits, represented as k bits. The channel encoder / modulator 640 performs channel coding, such as generating a low-density parity-check (LDPC) code, a turbo code, etc. The coded bits X are mapped to n modulation symbols, such as quadrature phase-shift keying (QPSK) or 16 quadrature amplitude modulation (QAM) symbols. In this example, QAM symbols will be assumed, although the disclosure is not so limited.

[0086] The n QAM symbols X are zero-padded by the encoder 650. In other words, a zero-valued symbol is added to the end of the QAM symbols X. The encoder neural network (Tx neural network) generates a complex sequence, which is added to the zero-padded QAM symbols X to create a new waveform. The output of the encoder 650 is a new waveform containing n' complex symbols, where n'>n.

[0087] The new waveform output from the encoder 650 is mapped to the tones of the OFDM symbol, along with the inserted pilot tones. The pilot tones are subcarriers containing known tones to help the receiver 620 estimate the channel H. An inverse fast Fourier transform (IFFT) converts the frequency domain signal to a time domain signal before adding a cyclic prefix and inputting the signal to the power amplifier 660.

[0088] For the purpose of training the neural network, a first loss function is measured here before being input to the power amplifier 660. The first loss function penalizes the peak-to-average power ratio (PAPR) measured at the output of the transmitter 610. That is, the PAPR is as small as possible to ensure that the power amplifier 660 does not operate in a nonlinear region based on the output of the IFFT. Due to the waveform modification by the encoder neural network, the output of the IFFT block is such that it does not cause the power amplifier 660 to enter a nonlinear region.

[0089] The output of the power amplifier 660 is received and amplified by a low noise amplifier (LNA) 670 of the receiver 620 over the channel H. The cyclic prefix is removed and a fast Fourier transform (FFT) recovers the frequency domain symbols. The pilot symbols are removed and a channel estimator estimates the channel H based on the pilot symbols. Channel Estimate

number

[0090] After the QAM symbols X are recovered, a second loss function is calculated when training the neural network, i.e., the second loss function attempts to minimize the mean squared error between the recovered modulation symbols and the modulation symbols input to encoder 650.

[0091] The total loss function for training the neural network may be the sum of a weighted first loss function and a weighted second loss function. By training the neural network using the total loss function, the overall design of the encoder neural network and the decoder neural network ensures that the power amplifier 660 operates in a linear region.

[0092] According to another aspect of the present disclosure, the second solution includes a neural network that transforms the transmit waveform to ensure that the power amplifier (PA) operates near its saturation region. That is, the PA nonlinearly distorts the waveform. Therefore, the encoder and decoder are designed to operate the PA near its saturation region. The structure of the neural network for the first and second solutions may be the same. However, the weights and loss functions within the neural network differ depending on the solution.

[0093] The encoder and decoder must understand what is between them, i.e., in this case, the PA nonlinearity in addition to the propagation channel. To train the encoder / decoder, a PA nonlinearity model (nonlinearity information) is generated. In other words, a neural network that models the PA nonlinearity is derived. Two approaches for this neural network are considered:

[0094] The first approach is transmitter (e.g., UE) driven. In this approach, the transmitter discovers / generates a neural network that models the PA nonlinearity. The transmitter trains the encoder and decoder and signals the decoder to the receiver. Artificial data for the channel is used in the training.

[0095] The second approach is receiver (e.g., gNB) driven. In the second approach, the transmitter discovers / generates a neural network that models the PA nonlinearity. Because the receiver is unaware of any nonlinearity in the transmitter, the transmitter generates a nonlinearity model and signals the neural network to the receiver. After receiving the PA nonlinearity information, the receiver trains an encoder and decoder and signals the encoder to the transmitter. In this approach, a real channel, rather than an artificial channel, can be measured and used in training. Alternatively, an artificial channel can be used for training.

[0096] There are two options for encoder / decoder location for this second solution: In the first option, processing is done in the time domain. Therefore, the encoder is placed after the Inverse Fast Fourier Transform (IFFT), which converts from the frequency domain to the time domain. The decoder is placed before the Fast Fourier Transform (FFT), which converts from the time domain to the frequency domain.

[0097] In the second option, the encoder and decoder operate on the waveform in the frequency domain, so the encoder is placed before the IFFT processing and the decoder is placed after the FFT processing.

[0098] Multiple-input multiple-output (MIMO) processing can also be incorporated into both the first and second solutions. To accommodate MIMO, the precoding block location for the encoder at the transmitter can be either after the encoder or before the encoder. A pilot insertion option is available for the first and second solutions. Also, pilot insertion can be performed either before or after the encoder neural network.

[0099] When the precoding block location is after the encoder, the precoded channel is used as side information in the decoder. The precoded channel is the raw propagation channel and the cascade of precoder coefficients applied at the transmitter. The side information is information input to the decoder to enable the decoder to recover the input to the encoder, for example, the channel information in the first two solutions described previously.

[0100] When a precoding block is placed before the encoder, the encoder input symbols are not quadrature amplitude modulation (QAM) symbols due to precoding. Either an estimate of the raw propagation channel or a precoded channel is used as side information in the decoder. The reconstruction target can be a QAM symbol from the output of the channel encoder / modulator or a precoded symbol. When the side information is the precoded channel, the target is a QAM symbol. When the side information is the estimated propagation channel, the target is a complex symbol.

[0101] 7 is a block diagram of an example architecture 700 for transforming a transmit waveform according to an embodiment of the present disclosure. In this architecture 700, precoding is performed before the encoder 650, and pilot symbols are also inserted before the encoder 650. Zero padding insertion is not shown in this figure, but may be included in some cases. Information bits are coded and modulated similarly to that described with reference to FIG. 6. Pilot symbols are inserted along with these modulated QAM symbols, and then precoding is performed. The pilot symbols are inserted in the frequency domain. After precoding, the complex symbols are no longer QAM symbols due to the precoding transformation.

[0102] The cyclic prefix and IFFT processing is performed as described with respect to Figure 6. At the receiver 620, the signal is processed by an LNA 670 and then further processed to remove the cyclic prefix and return the signal to the frequency domain.

[0103] The decoder 680 receives the frequency domain symbols and also the raw propagation channel, either precoded or estimated from the transmitted pilot tones.

number

number

[0104] 8 is a block diagram of another example architecture 800 for transforming a transmit waveform according to an aspect of the present disclosure, in which precoding occurs before the encoder 650 and pilot symbols are inserted after the encoder 650.

[0105] Information bits are coded and modulated in the same manner as described with respect to Figure 6. Precoding of the modulated symbols is performed so that the complex symbols are no longer QAM symbols due to the precoding transformation. After processing by the encoder 650, pilot symbols are inserted. Cyclic prefix and IFFT processing are then performed as described with respect to Figure 6. Zero padding insertion is not shown in this figure, but may be included in some cases.

[0106] In the receiver 620, the signal is processed by the LNA 670, which then further processes it to remove the cyclic prefix and return the signal to the frequency domain. The pilot signals are then removed. A decoder 680 receives the frequency domain symbols after the removal of the pilot symbols. The decoder 680 also receives the precoded channel (e.g., a cascade of precoding and raw channel estimated from pilot symbols) or the estimated raw propagation channel estimated from the transmitted pilot tones.

number

[0107] The reconstructed target output from the decoder 680 is either a QAM symbol or a complex symbol obtained by applying precoding to the QAM symbol. When the side information is a precoded channel, the target is a QAM symbol. When the side information is a propagation channel

number

[0108] 9 is a block diagram of yet another example architecture 900 for transforming a transmit waveform, in accordance with an aspect of the present disclosure, in which precoding occurs before the encoder 650 and pilot symbols are inserted after the encoder 650.

[0109] Information bits are coded and modulated in a similar manner as described with respect to Figure 6. Precoding of the modulated symbols is performed so that the complex symbols are no longer QAM symbols.

[0110] Pilot symbols are precoded and time division multiplexed (TDM) with the output of encoder 650. That is, the pilot symbols are sent at different times than the data symbols. The pilot symbols are precoded in the same way as the data symbols. Moreover, the pilot symbols are generated from low peak-to-average power ratio (PAPR) sequences. Therefore, the channel estimate is not affected by nonlinearities.

[0111] Cyclic prefix and IFFT processing is then performed as described with respect to Figure 6. Zero padding insertion is not shown in this figure, but may be included in some cases.

[0112] In the receiver 620, the signal is processed by the LNA 670 and then further processed to remove the cyclic prefix and return the signal to the frequency domain. The pilot signal is then removed.

[0113] The decoder 680 receives the frequency-domain data tones after removal of the pilot symbols. The decoder 680 also receives side information, including the precoded channel coefficients. The reconstruction target output from the decoder 680 is either the QAM symbols from the output of the channel encoder / modulator, or a linear combination of the QAM symbols with the precoded channel coefficients, which is the conventional information used by the receiver chain.

[0114] 10 is a block diagram of yet another example architecture 1000 for transforming a transmit waveform, according to an aspect of the disclosure, in which precoding occurs after the encoder 650 and pilot symbols are inserted before the encoder 650.

[0115] Information bits are encoded and modulated into QAM symbols (data symbols) similar to what was described with respect to Figure 6. Pilot symbols are time division multiplexed with the QAM symbols.

[0116] The precoder weights or precoder indices are input as side information to the neural network of the encoder 650. The encoding neural network therefore encodes based on the precoding that occurs at the output of the encoder 650.

[0117] Cyclic prefix and IFFT processing is then performed as described with respect to Figure 6. Zero padding insertion is not shown in this figure, but may be included in some cases.

[0118] In the receiver 620, the signal is processed by the LNA 670 and then further processed to remove the cyclic prefix and return the signal to the frequency domain. The pilot signal is then removed.

[0119] The decoder 680 receives the frequency-domain data tones after removal of the pilot symbols. The precoded channel coefficients (e.g., a cascade of precoding and raw channel estimated from pilot symbols) are input to the decoder 680 as side information. The reconstructed targets output from the decoder 680 are the QAM symbols from the output of the channel encoder / modulator.

[0120] 11 is a block diagram of a further example architecture 1100 for transforming a transmit waveform according to an aspect of the disclosure, in which precoding occurs after the encoder 650 and pilot symbols are inserted after the encoder 650.

[0121] 6, information bits are coded and modulated and input to encoder 650. Since precoding follows the neural network processing of encoder 650, precoder weights or precoder indices are input as side information to the neural network of encoder 650. Thus, the coding neural network codes based on the precoding that occurs at the output of encoder 650.

[0122] Pilot symbols are time-division multiplexed with the data symbols output from the channel encoder / modulator. The pilot tones are generated from low peak-to-average power ratio (PAPR) sequences to isolate the channel estimate from PA nonlinearities.

[0123] Cyclic prefix and IFFT processing is then performed as described with respect to Figure 6. Zero padding insertion is not shown in this figure, but may be included in some cases.

[0124] At the receiver 620, the signal is processed by an LNA 670, which then further processes it to remove the cyclic prefix and return the signal to the frequency domain. The pilot symbols are then removed.

[0125] The decoder 680 receives the frequency-domain data tones after removal of the pilot symbols. The decoder 680 also receives side information, including the precoded channel coefficients. The reconstruction target output from the decoder 680 is either the QAM symbols from the output of the channel encoder / modulator or a linear combination of the QAM symbols with the precoded channel coefficients.

[0126] According to another aspect of the present disclosure, distortion feedback may be used to address PA nonlinearities. In this solution, the transmitter communicates, via compression, to the receiver a distortion error ε=G(x)−x, where x=the undistorted digital transmit waveform in the time domain (the data being transported) and G(x)=the output of (e.g., nonlinear) clipping and filtering.

[0127] The encoder / decoder of the autoencoder compresses the distortion error ε using a compression scheme. The encoder (denoted by the function f(●)) at the transmitter encodes the distortion error ε for compression. The encoder output f(ε) (e.g., the distortion information compressed by the encoder of the autoencoder) is transmitted to the receiver. The encoder is placed after the IFFT.

[0128] The decoder g(●,●) at the receiver decodes the received, compressed distortion error f(ε) to recover the distortion error ε, which is used to recover the undistorted signal x. The function g(f(ε),H) = distortion error ε, where H is the channel and f(ε) is the encoder output. The decoder is placed before the FFT.

[0129] The transmitter trains both the encoder and decoder and signals the discovered / determined decoder to the receiver.

[0130] 12 illustrates an example process 1200 performed, for example, by a transmitting device, in accordance with various aspects of the present disclosure. The example process 1200 is an example of machine learning for addressing transmit (TX) nonlinearities.

[0131] 12, in some aspects, process 1200 may include transforming a transmit waveform by an encoder neural network to control power amplifier (PA) operation with respect to nonlinearity (block 1210). For example, a UE (e.g., using controller / processor 280, memory 282, etc.) or a base station (e.g., using controller / processor 240, memory 242, etc.) can transform the transmit waveform.

[0132] 12, in some aspects, process 1200 may include transmitting the transformed transmit waveform over a propagation channel (block 1220). For example, a UE (e.g., using antennas 252, DEMOD 254, TX MIMO processor 266, transmit processor 264, controller / processor 280, memory 282, etc.) or a base station (e.g., using antennas 234, MOD 232, TX MIMO processor 230, transmit processor 220, controller / processor 240, memory 242, etc.) may transmit the transformed transmit waveform.

[0133] 13 illustrates an example process 1300 performed, for example, by a receiving device, in accordance with various aspects of the present disclosure. The example process 1300 is an example of machine learning for addressing transmit (TX) nonlinearities.

[0134] 13, in some aspects, process 1300 may include receiving a waveform transformed by the encoder neural network (block 1310). For example, a UE (e.g., using antennas 252, DEMOD 254, MIMO detector 256, receive processor 258, controller / processor 280, memory 282, etc.) or a base station (e.g., using antennas 234, MOD 232, MIMO detector 236, receive processor 238, controller / processor 240, memory 242, etc.) may receive the waveform.

[0135] 13, in some aspects, process 1300 may include recovering encoder input symbols from the received waveform using a decoder neural network (block 1320). For example, a UE (e.g., using controller / processor 280, memory 282, etc.) or a base station (e.g., using controller / processor 240, memory 242, etc.) can recover the encoder input symbols.

[0136] 14 illustrates an example process 1400 performed, for example, by a transmitting device, in accordance with various aspects of the present disclosure. The example process 1400 is an example of machine learning for addressing transmit (TX) nonlinearities.

[0137] 14, in some aspects, process 1400 may include calculating a distortion error based on the undistorted digital transmit waveform and the distorted digital transmit waveform (block 1410). For example, a UE (e.g., using controller / processor 280, memory 282, etc.) or a base station (e.g., using controller / processor 240, memory 242, etc.) may calculate the distortion error.

[0138] 14, in some aspects, process 1400 may include compressing the distortion error using an encoder neural network of an autoencoder (block 1420). For example, the UE (e.g., using controller / processor 280, memory 282, etc.) or the base station (e.g., using controller / processor 240, memory 242, etc.) may compress the distortion error.

[0139] 14, in some aspects, process 1400 may include transmitting the compressed distortion error to a receiving device to compensate for power amplifier (PA) nonlinearity (block 1430). For example, a UE (e.g., using antennas 252, DEMOD 254, TX MIMO processor 266, transmit processor 264, controller / processor 280, memory 282, etc.) or a base station (e.g., using antennas 234, MOD 232, TX MIMO processor 230, transmit processor 220, controller / processor 240, memory 242, etc.) may transmit the compressed distortion error.

[0140] 15 illustrates an example process 1500 performed, for example, by a receiving device, in accordance with various aspects of the present disclosure. The example process 1500 is an example of machine learning for addressing transmit (TX) nonlinearities.

[0141] 15, in some aspects, process 1500 may include decompressing distortion errors caused by a power amplifier (PA) using a decoder neural network of an autoencoder (block 1510). For example, a UE (e.g., using controller / processor 280, memory 282, etc.) or a base station (e.g., using controller / processor 240, memory 242, etc.) may decompress the distortion errors.

[0142] 15, in some aspects, process 1500 may include recovering the undistorted signal based on the decompressed distortion error (block 1520). For example, the UE (e.g., using controller / processor 280, memory 282, etc.) or the base station (e.g., using controller / processor 240, memory 242, etc.) may recover the undistorted signal.

[0143] 12-15 illustrate example blocks of processes 1200, 1300, 1400, 1500 in some embodiments, although processes 1200, 1300, 1400, 1500 may include additional, fewer, different, or differently arranged blocks compared to the blocks illustrated in Figures 12-15. Additionally or alternatively, two or more of the blocks of processes 1200, 1300, 1400, 1500 may be performed in parallel.

[0144] The following numbered clauses describe example implementations. Clause 1. A method of wireless communication by a transmitting device, comprising: transforming a transmit waveform with an encoder neural network to control power amplifier (PA) operation with respect to nonlinearities; transmitting the converted transmit waveform over a propagation channel; A method comprising: Clause 2. The method of clause 1, further comprising transforming the transmit waveform to ensure that the PA operates in a linear region. Clause 3. The method of any of the preceding clauses, further comprising transforming the transmit waveform to ensure that the PA operates near its saturation region. Clause 4. Generating an encoder neural network and a decoder neural network; transmitting the decoder neural network to a receiving device; Any of the above clauses, including the following: Clause 5. The method of any of the preceding clauses, further comprising training the encoder neural network and the decoder neural network with artificial channel data. Clause 6. The method of any of the preceding clauses, further comprising encoding in the frequency domain or the time domain by an encoder neural network. Clause 7. The method of any of the preceding clauses, further comprising mapping an output of the encoder neural network to an orthogonal frequency division multiplexing (OFDM) symbol corresponding to the input data. Clause 8. The method of any of the preceding clauses, wherein the loss function of the encoder neural network is based on the peak-to-average power ratio (PAPR) of the transmitted waveform. Clause 9. The method of any of the preceding clauses, further comprising inserting a pilot signal before or after encoding by the encoder neural network. Clause 10. The method of any of the preceding clauses, wherein the pilot signal is pre-coded and inserted after encoding, and the pre-coded pilot signal is time-division multiplexed with the output from the encoder neural network. Clause 11. The method of any of clauses 1 to 9, wherein a pilot signal is inserted before encoding, the pilot signal is time-division multiplexed with the data symbols before encoding by the encoder neural network, and the encoder neural network receives precoder weights or precoder indices as side information. Clause 12. The method of any of clauses 1 to 9, wherein the pilot signal is inserted after encoding, the pilot signal is time-division multiplexed with the output from the encoder neural network, and the encoder neural network receives precoder weights or precoder indices as side information. Clause 13. A method of wireless communication by a receiving device, comprising: receiving a waveform transformed by an encoder neural network; recovering the encoder input symbols from the received waveform using a decoder neural network; A method comprising: Clause 14. Generating an encoder neural network and a decoder neural network; transmitting the encoder neural network to a transmitting device; The method of clause 13 further includes: Clause 15. The method of clause 13 or 14, further comprising training the encoder neural network and the decoder neural network using artificial channel data or using actual channel measurements. Clause 16. The method of any of clauses 13 to 15, further comprising recovering the encoder input symbols based on side information comprising a precoded channel or a raw propagation channel. Clause 17. The method of any of clauses 13 to 16, wherein the reconstruction target comprises symbols output from a channel encoder / modulator at the transmitting device, precoded symbols at the transmitting device, complex symbols obtained by applying precoding to symbols, or a linear combination of the symbols output from the channel encoder / modulator with precoded channel coefficients. Clause 18. A method of wireless communication in a transmitting device, comprising: calculating a distortion error based on the undistorted digital transmit waveform and the distorted digital transmit waveform; compressing the distortion error using an encoder neural network of the autoencoder; transmitting the compressed distortion error to a receiving device to compensate for power amplifier (PA) nonlinearity; A method comprising: Clause 19. Training an encoder neural network and a decoder neural network; transmitting the decoder neural network to a receiving device; The method of clause 18 further includes: Clause 20. A method of wireless communication in a receiving device, comprising: Decompressing distortion errors caused by a power amplifier (PA) using a decoder neural network of the autoencoder; recovering the undistorted signal based on the decompressed distortion error; A method comprising: Clause 21. A transmitting device for wireless communications, comprising: Memory and and at least one processor operably coupled to the memory, wherein the memory and the at least one processor: Transforming a transmit waveform by an encoder neural network to control power amplifier (PA) operation with respect to nonlinearities; and Transmitting the converted transmit waveform over the propagation channel. a transmitting device configured to: Clause 22. The transmit device of clause 21, wherein the at least one processor is further configured to transform the transmit waveform to ensure that the PA operates in a linear region. Clause 23. The transmitting device of clause 21 or 22, wherein the at least one processor is further configured to transform the transmit waveform to ensure that the PA operates near a saturation region. Article 24. At least one processor: generating an encoder neural network and a decoder neural network; and Transmitting the decoder neural network to a receiving device 24. The transmitting device of any of clauses 21 to 23, further configured to: Clause 25. The transmitting device of any of clauses 21 to 24, wherein the at least one processor is further configured to train the encoder neural network and the decoder neural network with the artificial channel data. Clause 26. The transmitting device of any of clauses 21 to 25, wherein the at least one processor is further configured to encode in the frequency domain or the time domain by an encoder neural network. Clause 27. The transmitting device of any of clauses 21 to 26, wherein the at least one processor is further configured to map an output of the encoder neural network to an orthogonal frequency division multiplexing (OFDM) symbol corresponding to the input data. Clause 28. The transmitting device of any of clauses 21 to 27, wherein the loss function of the encoder neural network is based on the peak-to-average power ratio (PAPR) of the transmitted waveform. Clause 29. The transmitting device of any of clauses 21 to 28, wherein the at least one processor is further configured to insert a pilot signal before or after encoding by the encoder neural network. Clause 30. The transmitting device of any of clauses 21 to 29, wherein the pilot signal is pre-coded and inserted after encoding, and the pre-coded pilot signal is time-division multiplexed with the output from the encoder neural network. Clause 31. The transmitting device of any of clauses 21 to 29, wherein the at least one processor is further configured to insert a pilot signal before encoding, the pilot signal being time division multiplexed with the data symbols before encoding by the encoder neural network, and the encoder neural network receiving precoder weights or precoder indexes as side information. Clause 32. The transmitting device of any of clauses 21 to 29, wherein the at least one processor is further configured to insert a pilot signal after encoding, the pilot signal being time division multiplexed with the output from the encoder neural network, and the encoder neural network receiving precoder weights or precoder index as side information. Clause 33. A receiving device for wireless communications, comprising: Memory and and at least one processor operably coupled to the memory, wherein the memory and the at least one processor: receiving a waveform transformed by an encoder neural network; and Recovering the encoder input symbols from the received waveform using a decoder neural network a receiving device configured to: Article 34. At least one processor: generating an encoder neural network and a decoder neural network; and Transmitting the encoder neural network to a transmitting device 34. The receiving device of clause 33, further configured to: Clause 35. The receiving device of any of clauses 33 to 34, wherein the at least one processor is further configured to train the encoder neural network and the decoder neural network using artificial channel data or using actual channel measurements. Clause 36. The receiving device of any of clauses 33 to 35, wherein the at least one processor is further configured to recover the encoder input symbols based on side information comprising a precoded channel or a raw propagation channel. Clause 37. A receiving device according to any of clauses 33 to 36, wherein the reconstruction target comprises symbols output from a channel encoder / modulator in the transmitting device, precoded symbols in the transmitting device, complex symbols obtained by applying precoding to symbols, or a linear combination of the symbols output from the channel encoder / modulator with precoded channel coefficients. Clause 38. An apparatus comprising at least one means for carrying out the method of any of clauses 1 to 20. Clause 39. A non-transitory computer readable medium storing code for wireless communications, the code comprising instructions executable by a processor to perform the method of any of clauses 1 to 20.

[0145] The above disclosure provides illustration and description, and is not intended to be exhaustive or to limit the embodiments to the precise form disclosed. Modifications and variations may be made in light of the above disclosure or may be acquired from practice of the embodiments.

[0146] As used herein, the term "component" shall be interpreted broadly as hardware, firmware, and / or a combination of hardware and software. As used herein, a processor is implemented in hardware, firmware, and / or a combination of hardware and software.

[0147] Some aspects are described herein in terms of thresholds. As used herein, meeting a threshold can refer to a value being greater than the threshold, greater than or equal to the threshold, less than the threshold, less than or equal to the threshold, equal to the threshold, not equal to the threshold, etc., depending on the context.

[0148] It will be apparent that the systems and / or methods described herein may be implemented in different forms of hardware, firmware, and / or combinations of hardware and software. The actual dedicated control hardware or software code used to implement these systems and / or methods is not limiting of the aspects. Accordingly, the operation and behavior of the systems and / or methods are described herein without reference to specific software code. It should be understood that software and hardware may be designed to implement the systems and / or methods based at least in part on the description herein.

[0149] Although particular combinations of features are recited in the claims and / or disclosed herein, these combinations do not limit the disclosure of various aspects. Indeed, many of these features may be combined in ways not specifically recited in the claims and / or disclosed herein. While each dependent claim listed below may depend directly on only one claim, the disclosure of various aspects includes each dependent claim in combination with every other claim in the claim set. A phrase referring to "at least one of" a list of items refers to any combination of those items, including single members. As an example, "at least one of a, b, or c" is intended to encompass a, b, c, ab, ac, bc, and abc, as well as any combination having multiple identical elements (e.g., aa, aaa, aab, aac, abb, acc, bb, bbb, bbc, cc, and ccc, or any other order of a, b, and c).

[0150] No element, act, or instruction used herein should be construed as critical or required unless explicitly described as such. Also, as used herein, the articles "a" and "an" include one or more items and may be used interchangeably with "one or more." Furthermore, as used herein, the terms "set" and "group" include one or more items (e.g., related items, unrelated items, combinations of related and unrelated items, etc.) and may be used interchangeably with "one or more." Where only one item is intended, the phrase "only one" or similar language is used. Also, as used herein, terms such as "has," "have," and "having" are intended to be open-ended terms. Furthermore, the phrase "based on" is intended to mean "based at least in part on," unless expressly specified otherwise. [Explanation of symbols]

[0151] 100 Network, Wireless Network 102a Macrocell 102b Picocell 102c Femtocell 110 BS, base station 110a BS, Macro BS 110b BS, Pico BS 110c BS, Femto BS 110d BS, relay station 120, 120a, 120b, 120c, 120d, 120e UE 130 Network Controller 200 designs 212, 262 data sources 220, 264 Transmit Processor 230 Transmit (TX) Multiple Input Multiple Output (MIMO) Processor, TX MIMO Processor 232 Modulator, Demodulator, MOD 232a~232t Modulator (MOD), Modulator 234, 234a~234t, 252, 252a~252r Antennas 236, 256 MIMO detector 238, 258 Receive Processor 239, 260 Data sink 240, 280, 290 Controller / Processor 242, 282, 292 memory 244, 294 communication unit 246 Scheduler 254 Demodulator, Modulator, DEMOD 254a~254r Demodulator (DEMOD), Demodulator, Modulator 266 TX MIMO Processor 300 System on Chip (SOC), SOC 302 Central Processing Unit (CPU), CPU, General Purpose Processor 304 Graphics Processing Unit (GPU), GPU 306 Digital Signal Processor (DSP), DSP 308 Neural Processing Unit (NPU) 310 Connectivity Blocks 312 Multimedia Processor 314 Sensor Processor 316 Image Signal Processor (ISP) 318 memory blocks 320 Navigation Module 400 DCN 402 Fully Connected Neural Network 404 Locally Connected Neural Networks 406 Convolutional Neural Networks 418 First set of feature maps, feature maps 420 Second set of feature maps, feature maps 422 output 424 First feature vector 426 images 428 Second feature vector 430 Image Capture Device 432 Convolutional Layer 550 Deep Convolutional Networks 552 input data 554A Convolution block, first of the convolution blocks 554B Convolution Block 556 Convolutional Layer (CONV), Convolutional Layer, Layer 558 normalization layer (LNorm), normalization layer, layer 560 Max Pooling Layer (MAX POOL), Max Pooling Layer, Layer 562 fully connected layer, layer 564 Logistic Regression (LR) Layer, Layer 566 classification scores 600, 700, 800, 900, 1000, 1100 architecture 610 Transmitter 620 receiver 630 Propagation Channel 640 Channel Encoder / Modulator 650 Encoder 660 Power Amplifier 670 Low Noise Amplifier (LNA), LNA 680 decoder 1200, 1300, 1400, 1500 processes

Claims

1. 1. A method of wireless communication by a transmitting device, comprising: generating an encoder neural network and a decoder neural network; transmitting the decoder neural network to a receiving device; transforming a transmit waveform with the encoder neural network to control the operation of a power amplifier (PA) with respect to nonlinearities; transmitting the converted transmit waveform over a propagation channel; A method comprising:

2. The step of converting the transmission waveform comprises: Transforming the transmit waveform to ensure that the PA operates in a linear region; or Transforming the transmit waveform to ensure that the PA operates near a saturation region.

2. The method of claim 1, comprising:

3. The method of claim 1 , further comprising training the encoder neural network and the decoder neural network with artificial channel data.

4. The method of claim 1 , further comprising encoding in the frequency domain or the time domain by the encoder neural network.

5. 10. The method of claim 1, further comprising: mapping an output of the encoder neural network to an orthogonal frequency division multiplexing (OFDM) symbol corresponding to input data.

6. The method of claim 1, wherein the loss function of the encoder neural network is based on the peak-to-average power ratio (PAPR) of the transmitted waveform.

7. The method of claim 1 , further comprising inserting a pilot signal before or after encoding by the encoder neural network.

8. the pilot signal is pre-coded and inserted after the encoding, and the pre-coded pilot signal is time-division multiplexed with the output from the encoder neural network; or 8. The method of claim 7, wherein the pilot signal is inserted before the encoding, the pilot signal is time-division multiplexed with data symbols before encoding by the encoder neural network, and the encoder neural network receives precoder weights or precoder indices as side information.

9. 8. The method of claim 7, wherein the pilot signal is inserted after the encoding, the pilot signal being time-division multiplexed with an output from the encoder neural network, and the encoder neural network receiving precoder weights or precoder indices as side information.

10. 1. A method of wireless communication by a receiving device, comprising: generating an encoder neural network and a decoder neural network; transmitting the encoder neural network to a transmitting device; receiving a waveform transformed by the encoder neural network; recovering encoder input symbols from the received waveform using the decoder neural network; A method comprising:

11. The method of claim 10 , further comprising training the encoder neural network and the decoder neural network with artificial channel data or with actual channel measurements.

12. The method of claim 10 , further comprising recovering the encoder input symbols based on side information comprising a precoded channel or a raw propagation channel.

13. 11. The method of claim 10, wherein the reconstruction target comprises a symbol output from a channel encoder / modulator at the transmitting device, a precoded symbol at the transmitting device, a complex symbol obtained by applying precoding to the symbol, or a linear combination of the symbol output from the channel encoder / modulator with precoded channel coefficients.

14. A transmitting device for wireless communication, comprising means for performing the method according to any one of claims 1 to 9.

15. A receiving device for wireless communication, comprising means for performing the method of any one of claims 10 to 13.

16. A computer program comprising instructions which, when executed by a computer, cause the computer program to perform the method of any one of claims 1 to 9.

17. A computer program comprising instructions which, when executed by a computer, cause the computer program to perform a method according to any one of claims 10 to 13.

Images