Telefonaktiebolaget Lm Ericsson (Publ)

Heunchul Lee

Robert Mark Harrison

Niklas Wernersson

Sebastian Faxér

Maksym Girnyk

Abdulrahman Alabbasi

Alexandros Palaios

Jaeseong Jeong

Henrik RydÉn

Methods and nodes in a communications network for using a machine learning model that has been trained using training data from a first radio environment, in a second radio environment, wherein the machine learning model takes a first set of features as input and outputs a label based on a first labelling scheme, the first labelling scheme having been determined for the first radio environment. A method comprises obtaining () a dataset in the second radio environment comprising a plurality of data points comprising values of the first set of features, labelling () each data point in the dataset using the first labelling scheme, and determining () from the labelled dataset whether the first labelling scheme is suitable for use in the second radio environment. Responsive to determining that the first labelling scheme is unsuitable for use in the second radio environment, the method further comprises determining () a second labelling scheme for the machine learning model, using an unsupervised labelling method.

Dynamic labeling for machine learning models for use in dynamic radio environments of a communications network

A method, system and apparatus for beamforming techniques that use approximate channel decomposition are disclosed. According to one aspect, a wireless device is provided. The wireless device includes processing circuitry configured to use a non-constant modulus, NCM, precoder of a NCM codebook to transmit on a physical channel where the NCM codebook is based at least in part on a constant modulus, CM, codebook and a plurality of NCM component matrices, and where the plurality of NCM component matrices are based at least in part on a NCM component matrix mathematical expression, and where an approximate decomposition element-wise product of a CM component matrix mathematical expression and the NCM component matrix mathematical expression corresponds to a channel decomposition matrix.

Beamforming technique using approximate channel decomposition

There is disclosed a method of operating a beam-forming wireless communication system, the system has a plurality of radio nodes, an actor neural network being associated to each radio node, wherein further to each actor neural network, there is associated a critic network. The method includes training each actor neural network, for controlling at least one associated radio node, based on learning feedback provided by its associated critic network, the learning feedback being based on operation information provided be the actor neural network for the critic network. The disclosure also pertains to related devices and methods.

Multi-agent policy machine learning

Method and device(s) for providing precoder selection policy for a multi-antenna transmitter arranged to transmit data over a communication channel of a wireless communication network. Machine learning in the form of reinforcement learning is applied involving adaptation of an action value function configured to compute an action value based on information indicative of a precoder of the multi-antenna transmitter and of a state relating to at least the communication channel. The adaptation being further based on reward information provided by a reward function, indicative of how successfully data is transmitted over the communication channel, and the precoder selection policy is provided based on the adapted action value function resulting from the reinforcement learning.

Provision of precoder selection policy for a multi-antenna transmitter

A user equipment (UE) transmits over multiple antenna ports and receives a control message from a base station in a wireless communication network. The control message comprises a precoding matrix indication field configurable to a first, second, and third configuration. The first configuration identifies precoding matrices in both a first set of precoding matrices and a second set of precoding matrices. The second configuration identifies precoding matrices in the second set of precoding matrices but not in the first set of precoding matrices. The third configuration identifies precoding matrices in a third set of precoding matrices in addition to the first and second sets. The precoding matrices in the sets are precoding matrices for transmissions from the UE. Each set of precoding matrices corresponds to a respective coherence capability. For a maximum of four spatial layers, the first, second, and third configurations occupy 5, 4, and 6 information bits, respectively.

Variable coherence adaptive antenna array

A user equipment (UE) () transmits data to a base station () in a wireless communication network. The UE () comprises multiple antenna ports, and selects a precoding matrix from a respective first, second, third, or fourth set of precoding matrices according to a number of spatial layers. The first, second, third, and fourth sets of precoding matrices are available for all coherence capabilities and are comprised within a larger set of precoding matrices. The larger set comprises precoding matrices that are not available for all coherence capabilities. The first, second, third, and fourth sets of precoding matrices correspond to one, two, three, or four spatial layers, respectively. The number of columns in the selected precoding matrix is equal to the number of spatial layers and each column comprises a single non-zero element and one or more zero elements. The UE () transmits data on the number of spatial layers according to the selected precoding matrix.

According to certain embodiments, a method for use in a wireless transmitter for simultaneously transmitting a plurality of wireless signals using a plurality of antenna elements comprises selecting a non-constant modulus (NCM) spatial precoder from a NCM codebook comprising a plurality of NCM spatial precoders. The plurality of the NCM spatial precoders are based on a unit-power beamforming vector optimized so that the wireless transmitter transmits up to the full power of each antenna element of the plurality of antenna elements while an overall transmitted power remains constant. The method further comprises applying the selected NCM spatial precoder to a data signal for generating a plurality of coded signals; and transmitting the plurality of coded signals as wireless signals via a power amplifier operating at full power.

Growing community of inventors

Stockholm, Sweden

Heunchul Lee