Abstract: A user equipment (UE) receives a first channel state information reference signal (CSI-RS) on a first set of beams and receives a second CSI-RS on a second set of beams. The apparatus determines a channel quality indicator (CQI), a rank indicator (RI), and a precoding matrix indicator (PMI) based on the first CSI-RS. The apparatus reports the CQI and RI using an uplink control channel and indicates the PMI using a precoded SRS transmission. The precoded SRS transmission may be precoded based on a dissimilarity between the first CSI-RS and the second CSI-RS. The first CSI-RS may include a defined precoder, and the second CSI-RS may include a precoder based on an uplink channel estimate of a non-precoded SRS from the UE.

Abstract: Aspects described herein relate to encoding, based on an estimated channel matrix of a reference signal received from a base station, channel state information (CSI) using a machine learning (ML)-based CSI encoder, transmitting, to the base station, an output of the ML-based CSI encoder and assistance information related to the estimated channel matrix, and receiving, from the base station, a scheduling grant for a downlink channel having at least one parameter that is based on the output of the ML-based CSI encoder and the assistance information. Other aspects relate to receiving the CSI encoder output and assistance information, and transmitting the scheduling grant.

Abstract: Various aspects of the present disclosure generally relate to wireless communication. In some aspects, a first device may encode a data set using one or more extraction operations and compression operations associated with a neural network, the one or more extraction operations and compression operations being based at least in part on a set of features of the data set to produce a compressed data set. The first device may transmit the compressed data set to a second device. Numerous other aspects are provided.

Abstract: Various aspects of the present disclosure generally relate to wireless communication. In some aspects, a user equipment (UE) may receive signaling indicating a transition from a first type of channel state information (CSI) processing to a second type of CSI processing at the UE, wherein at least one of the first type of CSI processing or the second type of CSI processing is a machine-learning-based CSI processing type. The UE may transition from the first type of CSI processing to the second type of CSI processing in accordance with the signaling and a timeline for the transition. Numerous other aspects are described.

Abstract: Methods, systems, and devices for wireless communication are described. A user equipment (UE) may select a non-Discrete Fourier Transform (non-DFT) codebook of a set of non-DFT codebooks associated with a channel state feedback message. The UE may determine a set of singular vectors associated with the non-DFT codebook based on a first machine learning model, where the set of singular vectors corresponds to a subspace associated with the non-DFT codebook. The UE may compress the non-DFT codebook, the set of singular vectors, or both, based on a second machine learning model. The UE may transmit the channel state feedback message including the compressed non-DFT codebook, the compressed set of singular vectors, or both.

Abstract: A method of wireless communication by a user equipment (UE), includes estimating a downlink channel to generate a channel estimate. The method also includes obtaining multiple precoding matrices, by a channel state information (CSI) module including a neural network encoder decoder pair, based on the channel estimate and multiple different multiple input multiple output (MIMO) ranks. The method further includes determining a best rank indicator based on the precoding matrices and spectral efficiency estimates for the different MIMO ranks. The method still further includes reporting, to a base station, the best rank indicator, a channel quality index (CQI), and CSI encoder output.

Abstract: Various aspects of the present disclosure generally relate to wireless communication. In some aspects, a user equipment (UE) may receive one or more channel state information (CSI)-reference signals (RSs). The UE may transmit a compressed CSI report that comprises an indication of a plurality of channel metrics, an indication of a channel metric of the plurality of channel metrics comprising a compressed output of a machine learning model having an input as the channel metric, wherein at least one of the plurality of channel metrics is based at least in part on expected orthogonalization of the compressed CSI report at a network node. Numerous other aspects are described.

Abstract: Various aspects of the present disclosure generally relate to wireless communication. In some aspects, a user equipment (UE) may receive one or more channel state information (CSI) reference signals (CSI-RSs). The UE may transmit a compressed CSI report via layer 2 signaling, the compressed CSI report being compressed using a machine learning model. Numerous other aspects are described.

Abstract: In an aspect, a UE obtains information (e.g., UE-specific information, etc.) associated with a set of triggering criteria for a set of neural network functions, the set of neural network functions configured to facilitate positioning measurement feature processing at the UE, the set of neural network functions being generated dynamically based on machine-learning associated with one or more historical measurement procedure, obtains positioning measurement data associated with a location of the UE, and determines a positioning estimate for the UE based at least in part upon the positioning measurement data and at least one neural network function from the set of neural network functions that is triggered by at least one triggering criterion from the set of triggering criteria.

Abstract: In an aspect, a UE obtains information (e.g., UE-specific information) associated with a set of triggering criteria (e.g., from a server, a serving network, e.g., in conjunction with or separate from a set of neural network functions) for a set of neural network functions, the set of neural network functions configured to facilitate positioning measurement feature processing at the UE, the set of neural network functions being generated dynamically based on machine-learning associated with one or more historical measurement procedures. The UE obtains positioning measurement data associated with a location of the UE, and processes the positioning measurement data into a respective set of positioning measurement features based at least in part upon the positioning measurement data and at least one neural network function from the set of neural network functions that is triggered by at least one triggering criterion from the set of triggering criteria.

Abstract: A method of wireless communication by a user equipment (UE) includes receiving, from a base station, a configuration to train a neural network for multiple different signal to noise ratios (SNRs) of a channel estimate for a wireless communication channel. The method also includes determining a current SNR of the channel estimate is above a first threshold value. The method further includes training the neural network based on the channel estimate, to obtain a first trained neural network. The method still further includes perturbing the channel estimate to obtain a perturbed channel estimate, and training the neural network based on the perturbed channel estimate, to obtain a second trained neural network. The method includes reporting, to the base station, parameters of the first trained neural network along with the channel estimate, and parameters of the second trained neural network.

Abstract: A base station may precode a first reference signal using a first precoder (P1) and at least one second reference signal using at least one second precoder (P2). At least two UEs may measure the first reference signal and the at least one second reference signal, and transmit channel state feedback based on the measuring (m11 and m12 for a first UE, and m21 and m22 for a second UE). The channel state feedback may comprise channel state information compressed by at least one neural network. The base station may derive a higher-level precoder (W) for multiple-input multiple-output (MIMO) downlink transmission to the at least two UEs.

Abstract: Certain aspects of the present disclosure provide techniques for uplink sounding reference signal (SRS) transmission with precoding. A method for wireless communication by a user equipment (UE) includes receiving a SRS configuration that configures one or more SRS resource sets, each resource set including one or more SRS resources, each SRS resource comprising one or more SRS ports. The UE receives a set of precoders via an indicator, the set of precoders including a precoder associated with each SRS resource and determines one or more precoders to apply for one or more SRS transmissions via one or more of the SRS resources.

Abstract: Methods, systems, and devices for wireless communication are described. A user equipment (UE) may receive signaling from a network entity via a cell. The signaling may indicate a state of resource utilization associated with at least the cell. The UE may modify a compression scheme in accordance with the state of resource utilization. The compression scheme may be used at the UE for compressing channel state information (CSI) associated with a layer used for wireless communication between the UE and the network entity via the cell.

Abstract: In an aspect, a UE obtains information (e.g., UE-specific information, etc.) associated with a set of triggering criteria for a set of neural network functions, the set of neural network functions configured to facilitate positioning measurement feature processing at the UE, the set of neural network functions being generated dynamically based on machine-learning associated with one or more historical measurement procedure, obtains positioning measurement data associated with a location of the UE, and determines a positioning estimate for the UE based at least in part upon the positioning measurement data and at least one neural network function from the set of neural network functions that is triggered by at least one triggering criterion from the set of triggering criteria.

Abstract: Various embodiments include methods for demodulating wireless transmission waveforms to reconstruct data tones, performed in receiver circuitry of a wireless communication device. The methods may include receiving time domain wireless transmission waveforms that include time domain peak reduction tones (PRTs) that were generated by a set of PRT neural networks using time domain data signals derived from frequency domain data tones in another wireless communication device, reconstructing the time domain data signals from the time domain wireless transmission waveforms using a receiver neural network that has been trained in conjunction with the set of PRT neural networks to generate a time domain reconstruction of data signals, and transforming the time domain reconstruction of data signals to a frequency domain reconstruction of data tones.

Abstract: Various aspects of the present disclosure generally relate to wireless communication. In some aspects, a user equipment (UE) may receive a sounding reference signal (SRS) configuration that indicates a maximum number of SRS ports and whether the UE is permitted to determine an actual number of SRS ports to be used for SRS transmission that is different from the maximum number of SRS ports. The UE may determine the actual number of SRS ports to be used for SRS transmission based at least in part on the SRS configuration. The UE may transmit one or more SRSs using the actual number of SRS ports. Numerous other aspects are provided.

Abstract: Certain aspects of the present disclosure provide techniques for measurement encoding and decoding using neural networks to compress and decompress measurement data. One example method generally includes: generating, via each of a plurality of neural network encoders operating on measurement data, a compressed measurement based on a respective portion of the measurement data, wherein each of the neural network encoders is based on the same neural network model; generating at least one message indicative of the measurement data based on the compressed measurements; and transmitting the at least one message.

Abstract: Various embodiments include methods for reducing peak to average power ratio of wireless transmission waveforms, performed in transmitter circuitry of a wireless communication. Various embodiments may include receiving frequency domain data tones, transforming the frequency domain data tones to time domain data signals, generating, using a set of peak reduction tone (PRT) neural networks, time domain PRTs using the time domain data signals in which the set of PRT neural networks have been trained in conjunction with an augmentation neural network, and a receiver neural network, generating an output of the augmentation neural network based on an input of final combined time domain signals including the time domain PRTs combined with previous combined time domain signals, and generating time domain wireless transmission waveforms that include the output of the augmentation neural network combined with the final combined time domain signals.

Abstract: A method of wireless communication, by a user equipment (UE), includes receiving multiple neural network training configurations for channel state feedback (CSF). Each configuration corresponds to a different neural network framework. The method also includes training each of a group of neural network decoder/encoder pairs in accordance with the received training configurations. A method of wireless communication, by a base station, includes transmitting multiple neural network training configurations to a user equipment (UE) for channel state feedback (CSF). Each configuration corresponds to a different neural network framework. The method also includes receiving a neural network decoder/encoder pair trained in accordance with the training configurations.