Machine learning based channel state information estimation and feedback configuration

Patent metadata

Field	Value
Publication number	US-11949465-B2
Application number	US-202117180061-A
Country	US
Kind code	B2
Filing date	Feb 19, 2021
Priority date	Feb 19, 2021
Publication date	Apr 2, 2024
Grant date	Apr 2, 2024

Abstract

Systems, methods, apparatuses, and computer program products for machine learning based channel state information (CSI) estimation and feedback configuration are provided. A method may include learning channel state information feedback from one or more user equipment as time series data. The method may also include building a predictive model for user equipment feedback based on the learned channel state information feedback. The method may further include configuring a channel state information trigger with the one or more user equipment based on the predictive model. In addition, the method may include signaling predicted channel state information to the one or more user equipment.

First claim

We claim:

1. A method, comprising:
learning channel state information feedback from one or more user
equipment as time series data, wherein the time series data comprises data transmitted by the one or more user equipment which is associated with corresponding time information;
building a predictive model for user equipment feedback based on the learned channel state information feedback;
configuring a channel state information trigger with the one or more user equipment based on the predictive model, wherein the channel state information trigger configures the one or more user equipment to send channel state information feedback when a difference between predicted channel state information and an observed or computed value satisfies a learned threshold value that is learned depending on previous instances of radio link failures and an increase in packet loss rate, wherein channel state information includes CRI-RI-CQI and L1-RSRP values;
transmitting learned channel state information feedback information to the one or more user equipment periodically using a new radio resource control configuration including CSI-PredictInfo;
signaling the predicted channel state information to the one or more user equipment,
generating a learned model from the learned channel state information feedback, wherein the learned model includes time series data using LSTMs and polynomial regression, wherein signaling the predicted channel state information comprises configuring the one or more user equipment with the learned model; and
determining a difference between the predicted and actual channel state information; and
signaling to the one or more user equipment the difference between the predicted and actual channel state information.

2. The method according to claim 1 , wherein signaling the predicted channel state information comprises signaling the learned channel state information feedback to the one or more user equipment.

3. The method according to claim 1 , wherein when the predicted channel state information is different from an observed value, the method further comprises receiving channel state information feedback from the one or more user equipment.

4. The method according to claim 3 , further comprising updating the predictive model based on the received channel state information feedback from the one or more user equipment.

5. An apparatus, comprising:
at least one processor; and
at least one memory comprising computer program code,
the at least one memory and the computer program code are configured, with the at least one processor, to cause the apparatus at least to:
learn channel state information feedback from one or more user equipment as time series data, wherein the time series data comprises data transmitted by the one or more user equipment which is associated with corresponding time information;
build a predictive model for user equipment feedback based on the learned channel state information feedback;
configure a channel state information trigger with the one or more user equipment based on the predictive model, wherein the channel state information trigger configures the one or more user equipment to send channel state information feedback when a difference between predicted channel state information and an observed or computed value satisfies a learned threshold value that is learned depending on previous instances of radio link failures and an increase in packet loss rate, wherein channel state information includes CRI-RI-CQI and L1-RSRP values;
transmit learned channel state information feedback information to the one or more user equipment periodically using a new radio resource control configuration including CSI-PredictInfo;
signal the predicted channel state information to the one or more user equipment,
generate a learned model from the learned channel state information feedback, wherein the learned model includes time series data using LSTMs and polynomial regression, wherein signaling the predicted channel state information comprises configuring the one or more user equipment with the learned model; and
determine a difference between the predicted and actual channel state information; and
signal to the one or more user equipment the difference between the predicted and actual channel state information.

6. The apparatus according to claim 5 , wherein when the predicted channel state information is different from an observed value, the at least one memory and the computer program code are further configured, with the at least one processor, to cause the apparatus at least to:
receive channel state information feedback from the one or more user equipment.

7. The apparatus to claim 6 , wherein the at least one memory and the computer program code are further configured, with the at least one processor, to cause the apparatus at least to:
update the predictive model based on the received channel state information feedback from the one or more user equipment.

Assignees

• Nokia Technologies Oy

Inventors

• Prasad Athul
• Ghosh Amitabha
• Lee Gilsoo
• Chen Jie
• Tan Jun

Classifications

• G06N3/0442

  characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU] · CPC title

• G06N3/098

  Distributed learning, e.g. federated learning · CPC title

• G06N3/09

  Supervised learning · CPC title

• H04B17/3913 Primary

  Predictive models, e.g. based on neural network models · CPC title

• G06N20/00

  Machine learning · CPC title