US 12,250,035 B2
Codebook for AI-assisted channel estimation
Yeqing Hu, Allen, TX (US); Xiaowen Tian, Raleigh, NC (US); Yang Li, Plano, TX (US); Tiexing Wang, Plano, TX (US); and Jianzhong Zhang, Plano, TX (US)
Assigned to Samsung Electronics Co., Ltd., Suwon-si (KR)
Filed by Samsung Electronics Co., Ltd., Suwon-si (KR)
Filed on Aug. 4, 2023, as Appl. No. 18/365,874.
Claims priority of provisional application 63/397,740, filed on Aug. 12, 2022.
Prior Publication US 2024/0056138 A1, Feb. 15, 2024
Int. Cl. H04B 7/0456 (2017.01); H04L 25/02 (2006.01)
CPC H04B 7/0456 (2013.01) [H04L 25/0204 (2013.01); H04L 25/0256 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A method for generating a codebook, the method performed by at least one processor, the method comprising:
identifying auto-correlation function (ACF) information by:
obtaining channel information that includes multiple channels of expected operation scenarios; and
based on the channel information for each of the channels, determining minimum mean-square error (MMSE) channel estimation (CE) weights expressed in a form of auto-correlation functions (ACFs) and a signal-to-noise ratio (SNR), and covariance matrices;
clustering the MMSE CE weights from the ACF information into K clusters, wherein a center ACF weight of each of the K clusters represents a codeword for computing channel estimation weights;
determining a classification distance metric based on a cluster distance after a re-clustering;
in response to a determination that cluster distances before and after the clustering differ from each other by a non-negligible value, iteratively re-clustering the ACF information thereby updating the center ACF weights and the cluster distances; and
generating the codebook to include an index k of each of the K clusters and the center ACF weight of each of the K clusters.