Method and apparatus for transceiving signal using artificial intelligence in wireless communication system
The present disclosure relates to 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 transmitting or receiving a signal by a user equipment (UE) in a mobile communication system is provided. The method may include: identifying a neural network model for transmitting first information to a base station (BS); learning a connection weight of the neural network model using the first information; transmitting, to the base station, second information for updating a weight of a second partial neural network corresponding to the base station based on a result of the learning; and updating a weight of a first partial neural network corresponding to the UE based on the result of the learning.
1. A method of transmitting or receiving a signal by a user equipment (UE) in a mobile communication system, the method comprising:
selecting a neural network model corresponding to an autoencoder among a plurality of autoencoders based on at least one parameter associated with a wireless link between the UE and a base station (BS), the at least one parameter including an operating frequency, a UE bandwidth, and antenna configurations of the UE and the BS, wherein the neural network model includes a transmission (Tx) neural network disposed in the UE and a reception (Rx) neural network disposed in the BS;
learning a connection weight of the neural network model using a shadow neural network disposed in the UE based on a channel matrix, the channel matrix being estimated by the UE or being received from the BS, wherein a first neural network corresponding to the Tx neural network and a second neural network corresponding to the Rx neural network are included in the shadow neural network for shadow training;
transmitting, to the BS, a new connection weight of the Rx neural network based on result of the learning;
receiving, from the BS, an acknowledgement (ACK) signal indicating that the new connection weight is received by the BS, wherein a weight of the Rx neural network is updated based on the new connection weight;
updating a weight of the Tx neural network based on the result of the learning after the ACK signal is received from the BS; and
transmitting, to the BS, channel state information (CSI) feedback information based on the updated weight of the Tx neural network.
2. The method of claim 1 ,
wherein the neural network model further comprises at least one hidden layer with a number of nodes less than a number of nodes of an input layer and less than a number of nodes of an output layer.
3. The method of claim 1 , wherein the new connection weight is periodically transmitted with a first periodicity,
wherein the first periodicity is larger than a second periodicity for a channel state information (CSI) report.
4. The method of claim 1 , wherein the new connection weight is aperiodically transmitted in response to a request from the BS.
5. The method of claim 1 , wherein the new connection weight is aperiodically transmitted in response to a performance of the neural network model being greater than or equal to a threshold.
6. The method of claim 5 , wherein the performance of the neural network model is defined based on a degree of a difference between an input and an output of the neural network model.
7. A method of transmitting or receiving a signal by a base station (BS) in a mobile communication system, the method comprising:
selecting a neural network model corresponding to an autoencoder among a plurality of autoencoders based on at least one parameter associated with a wireless link between a user equipment (UE) and the BS, the at least one parameter includes an operating frequency, a UE bandwidth, and antenna configurations of the UE and the BS, wherein the neural network model includes a transmission (Tx) neural network disposed in the UE and a reception (Rx) neural network disposed in the BS;
receiving, from the UE, a new connection weight of the Rx neural network, based on a result of a learning for a connection weight of the neural network model,
transmitting, to the UE, an acknowledgement (ACK) signal indicating that the new connection weight is received by the BS;
updating a weight of the Rx neural network based on the new connection weight after the ACK signal is transmitted to the UE, wherein a weight of the Tx neural network is updated based on the result of the learning; and
receiving, from the UE, channel state information (CSI) feedback based on the updated weight of the Rx neural network,
wherein the learning is performed by a shadow neural network disposed in the UE based on a channel matrix, the channel matrix being estimated by the UE or being transmitted to the UE by the BS, and
wherein a first neural network corresponding to the Tx neural network and a second neural network corresponding to the Rx neural network are included in the shadow neural network for shadow training.
8. The method of claim 7 ,
wherein the neural network model further comprises at least one hidden layer with a number of nodes less than a number of nodes of an input layer and less than a number of nodes of an output layer.
9. The method of claim 7 , wherein a weight of the neural network model is learned in the UE based on channel information estimated by the UE.
10. A user equipment (UE) which transmits or receives a signal in a mobile communication system, the UE comprising:
a transceiver configured to transmit or receive a signal; and
a controller configured to:
select a neural network model corresponding to an autoencoder among a plurality of autoencoders based on at least one parameter associated with a wireless link between the UE and a base station (BS), the at least one parameter includes an operating frequency, a UE bandwidth, and antenna configurations of the UE and the BS, wherein the neural network model includes a transmission (Tx) neural network disposed in the UE and a reception (Rx) neural network disposed in the BS,
learn a connection weight of the neural network model using a shadow neural network disposed in the UE based on a channel matrix, the channel matrix being estimated by the UE_or being received from the BS, wherein a first neural network corresponding to the Tx neural network and a second neural network corresponding to the Rx neural network are included in the shadow neural network for shadow training,
transmit, to the BS, a new connection weight of the Rx neural network based on a result of the learning,
receive, from the BS, an acknowledgement (ACK) signal indicating that the new connection weight is received by the BS, wherein a weight of the Rx neural network is updated based on the new connection weight,
update a weight of the Tx neural network based on the result of the learning after the ACK signal is received from the BS, and
transmit, to the BS, channel state information (CSI) feedback information based on the updated weight of the Tx neural network.
11. The UE of claim 10 ,
wherein the neural network model further comprises at least one hidden layer with a number of nodes less than a number of nodes of an input layer and less than a number of nodes of an output layer.
12. A base station (BS) which transmits or receives a signal in a mobile communication system, the BS comprising:
a transceiver configured to transmit or receive a signal; and
a controller configured to:
select a neural network model corresponding to an autoencoder among a plurality of autoencoders based on at least one parameter associated with a wireless link between a user equipment (UE) and the BS, the at least one parameter includes an operating frequency, a UE bandwidth, and antenna configurations of the UE and the BS, wherein the neural network model includes a transmission (Tx) neural network disposed in the UE and a reception (Rx) neural network disposed in the BS,
receive, from the UE, a new connection weight of the Rx neural network, based on a result of a learning for a connection weight of the neural network model,
transmit, to the UE, an acknowledgement (ACK) signal indicating that the new connection weight is received by the BS,
update a weight of the Rx neural network based on the new connection weight after the ACK signal is transmitted to the UE, wherein a weight of the Tx neural network is updated based on the result of the learning, and
receive, from the UE, channel state information (CSI) feedback information based on the updated weight of the Rx neural network,
wherein the learning is performed by a shadow neural network disposed in the UE based on a channel matrix, the channel matrix being estimated by the UE or being transmitted to the UE by the BS, and
wherein a first neural network corresponding to the Tx neural network and a second neural network corresponding to the Rx neural network are included in the shadow neural network for shadow training.
13. The BS of claim 12 ,
wherein the neural network model further comprises at least one hidden layer with a number of nodes less than a number of nodes of an input layer and less than a number of nodes of an output layer.