A wireless network can generate candidate signal configurations for physical transmissions to or from a user equipment (UE) in a radio environment. The generation of candidate signal configurations can be performed using a first neural network that is associated with the UE. These signal configurations can then be evaluated using a second neural network that is associated with the radio environment. The second neural network can be trained using measurements from previous physical transmissions in the radio environment. The trained second neural network generates a reward value that is associated with the candidate signal configurations. The first neural network is then trained using the reward values from the second neural network to produce improved candidate signal configurations. When a signal configuration that produces a suitable reward value is generated, this signal configuration can be used for the physical transmission in the radio environment.

An apparatus, method and computing program is described including: receiving one or more received symbols and one or more received bits, wherein the received symbols are received at a receiver of a transmission system including a transmitter, a channel, and the receiver; converting one or more of the received bits that are deemed to be correct into one or more estimated transmission symbols; generating an estimated channel transfer function based on one or more of the estimated transmission symbols and corresponding received symbols; and providing training data pairs, each training data pair including a first element based on the estimated channel transfer function and a second element based on the corresponding received symbols.

There is provided mechanisms for decoding data received from a terminal device. A method is performed by a network node. The method comprises receiving data, from the terminal device, during a set of user conditions prevailing for the terminal device. The set of user conditions comprises a rank indicator value reported by the terminal device and a measurement performed by the network node on at least one reference signal received from the terminal device. The method comprises selecting, by providing the set of user conditions as input to a database, a channel matrix from the database. The database comprises a set of offline trained channel matrices. The method comprises decoding the received data for the terminal device using the selected channel matrix.

Example embodiments of the present disclosure relate to machine learning-based channel estimation. According to an example embodiment, a first device determines a signal quality that is expected in transmission of a reference signal from a second device to the first device and receives the reference signal from the second device. The first device selects, based on the expected signal quality, a channel estimation model from a plurality of channel estimation models that have been trained for a plurality of candidate signal qualities for the reference signal. The first device determines, using the selected channel estimation model and based on the received reference signal, channel state information of a communication channel from the first device to the second device. According to this solution, a channel estimation model is dynamically selected for use, depending on a real-time signal quality expected to be gained in transmission of a certain RS.

A processor-implemented method is presented. The method includes receiving an input sequence comprising a group of channel dynamics observations for a wireless communication channel. Each channel dynamics observation may correspond to a timing of a group of timings. The method also includes determining, via a recurrent neural network (RNN), a residual at each of the group of timings based on the group of channel dynamics observations. The method further includes updating Kalman filter (KF) parameters based on the residual and estimating, via the KF, a channel state based on the updated KF parameters.

In a 5G communication system or a 6G communication system for supporting higher data rates beyond a 4G communication system such as long term evolution (LTE), a method of a first terminal in a wireless communication system is disclosed and may include performing channel measurement, based on one or more first reference signals received from a base station; identifying channel distribution information between the first terminal and the base station, based on the measured channel; selecting one or more representative channel vectors (RCVs), based on the identified channel distribution information; generating one or more constellations corresponding to the selected one or more RCVs; transmitting constellation set information including the generated one or more constellations to the base station; and performing communication with the base station, based on the generated one or more constellations.

A method performed by a radio unit for handling a number of received radio signals over an array of antennas comprised in the radio unit. The radio unit transforms the number of received radio signals into a number of sequences of complex symbols. The radio unit further filters the number of sequences of complex symbols by inputting the number of sequences of complex symbols into a trained computational model comprising an alternating sequence of linear and nonlinear functions and thereby obtaining a reduced number of sequences. The radio unit further transmits the reduced number of sequences to a baseband unit over a front-haul link.

A method includes receiving, by a first user device in a wireless network, an indication of availability of a pre-trained model that estimates a channel between a second user device and a network node; receiving, by the first user device, information relating to the pre-trained model; determining, by the first user device, channel estimation information based at least on the information relating to the pre-trained model; and performing at least one of the following: transmitting, by the first user device, a report to the network node including the channel estimation information; or receiving data, by the first user device from the network node, based on the channel estimation information.

Systems, devices, and methods related to hybrid basis function, neural network-based digital predistortion (DPD) are provided. An example apparatus for a radio frequency (RF) transceiver includes a digital predistortion (DPD) actuator to receive an input signal associated with a nonlinear component of the RF transceiver and output a predistorted signal. The DPD actuator includes a basis-function-based actuator to perform a first DPD operation using a set of basis functions associated with a first nonlinear characteristic of the nonlinear component. The DPD actuator further includes a neural network-based actuator to perform a second DPD operation using a first neural network associated with a second nonlinear characteristic of the nonlinear component. The predistorted signal is based on a first output signal of the basis-function-based actuator and a second output signal of the neural network-based actuator.

An apparatus, method and computer program is described comprising: combining first features extracted from an echo signal using a convolutional encoder of a convolutional encoder-decoder having first weights, wherein the echo signal is obtained in response to a transmission over a channel or a simulation thereof; and using a convolutional decoder of the convolutional encoder-decoder to generate an estimate of a frequency response of the channel based on the echo signal, wherein the convolutional decoder has second weights.