AI-BASED CSI PREDICTION WITH WEIGHT SHARING

Organization Name

Samsung Electronics Co., Ltd.

Inventor(s)

Daoud Burghal of San Jose CA (US)

Yang Li of Plano TX (US)

Pranav Madadi of Sunnyvale CA (US)

Jeongho Jeon of San Jose CA (US)

Joonyoung Cho of Portland OR (US)

Jianzhong Zhang of Dallas TX (US)

AI-BASED CSI PREDICTION WITH WEIGHT SHARING - A simplified explanation of the abstract

This abstract first appeared for US patent application 18181447 titled 'AI-BASED CSI PREDICTION WITH WEIGHT SHARING

Simplified Explanation

The method described in this patent application involves receiving a pilot signal or a measurement report from a user equipment (UE) or a base station (BS). The received channel state information (CSI) is then stored in a CSI buffer, which holds previous uplink or downlink channel estimates. A portion of the CSI buffer is provided to a CSI predictor, which is an artificial intelligence (AI) model that utilizes weight sharing mechanisms. The AI model consists of a sequence of layers and is used to predict temporal CSI. Additionally, the method can be used for denoising and frequency extrapolation, depending on the desired output.

The method involves receiving a pilot signal or a measurement report from a UE or a BS.
The received CSI is stored in a CSI buffer, which holds previous channel estimates.
A portion of the CSI buffer is provided to an AI-based CSI predictor.
The AI model utilizes weight sharing mechanisms and consists of a sequence of layers.
The CSI predictor is used to predict temporal CSI.
The method can also be used for denoising and frequency extrapolation.

Potential Applications:

Wireless communication systems
Mobile networks
Internet of Things (IoT) devices
5G and beyond networks

Problems Solved:

Efficient utilization of channel state information
Improved prediction of temporal CSI
Enhanced denoising and frequency extrapolation techniques

Benefits:

Better performance in wireless communication systems
Improved accuracy in predicting channel state information
Enhanced data transmission and reception in mobile networks
More efficient utilization of network resources

Original Abstract Submitted

A method includes receiving a pilot signal or a measurement report from a user equipment (UE) or a base station (BS). The method also includes updating a CSI buffer with channel state information (CSI) obtained from the pilot signal or the measurement report, the CSI buffer configured to store previous uplink or downlink channel estimates. The method also includes providing at least a portion of the CSI buffer to a CSI predictor comprising an artificial intelligence (AI) model that utilizes one or more weight sharing mechanisms, the AI model comprising a sequence of layers. The method also includes predicting temporal CSI using the CSI predictor. Depending on the configured output, the method can also include and/or be used for denoising and frequency extrapolation.