IP Library › Granted Patent US 12,695,526
Granted Patent B2
US 12,695,526 · App. 18/316,028 · Granted Jul 28, 2026

Electronic device for predicting channel in MIMO communication system and method of predicting the same

Inventors: Junil Choi (Daejeon, KR); Beomsoo Ko (Daejeon, KR); Hwanjin Kim (Daejeon, KR)
Assignees: Samsung Electronics Co., Ltd.; Korea Advanced Institute of Science and Technology
H04B17/3913H04B17/309H04B7/0413
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,695,526
App. No.
18/316,028
Filed
May 11, 2023
Granted
Jul 28, 2026
Kind
B2
Examiner
LE, LANA N
Art Unit
2648
USPC
455/67.11
Abstract

An electronic device which includes processing circuitry configured to acquire training data corresponding to multiple antennas, the training data having a label and features, the label including a first channel value for a first time slot, and the features including a plurality of second channel values for a plurality of time slots before the first time slot, train a channel prediction model based on the training data to obtain a trained channel prediction model, and obtain a channel prediction value for a prediction time based on the trained channel prediction model.

Claims (95)

1 . An electronic device comprising:

processing circuitry configured to,

acquire training data corresponding to multiple antennas, the training data having a label and features, the label including a first channel value for a first time slot, and the features including a plurality of second channel values for a plurality of time slots before the first time slot;

train a channel prediction model based on the training data to obtain a trained channel prediction model; and

obtain a channel prediction value for a prediction time based on the trained channel prediction model,

wherein the processing circuitry is configured to:

estimate a respective channel value for each of a plurality of resource blocks (RBs) based on an uplink signal, the uplink signal including a pilot signal, and the pilot signal being received on the plurality of RBs through the multiple antennas; and

acquire the label and the features from a row vector of a matrix, a column vector of the matrix having the respective channel value for each of the plurality of RBs.

2 . The electronic device of claim 1 , wherein

the respective channel value for each of the plurality of RBs has a dimension of M×1, M being a number of the multiple antennas; and

the label and each of the features has a dimension of 1×L, L being a number of the plurality of RBs.

3 . The electronic device of claim 1 , wherein the processing circuitry is configured to train the channel prediction model to minimize an error between a result value of the channel prediction model and the label.

4 . The electronic device of claim 1 , wherein the channel prediction model is based on a Multi-Layer Perceptron (MLP).

5 . The electronic device of claim 1 , wherein a number of the plurality of time slots is set based on a moving speed of a user equipment that transmits the uplink signal.

6 . The electronic device of claim 1 , wherein the channel prediction model includes only one model.

7 . The electronic device of claim 1 , wherein

the first channel value for the first time slot is

q

n

+

1

α

;

and

the plurality of second channel values are

q

n

-

n

0

+

1

α

,

…

,

q

n

α

,

α being a value from 1 to M, M being a number of the multiple antennas, and n being a natural number.

8 . The electronic device of claim 1 , wherein the processing circuitry is configured to reduce a time dimension of the training data.

9 . A method of predicting a channel, the method comprising:

acquiring training data corresponding to multiple antennas, the training data having a label and features, the label including a first channel value for a first time slot, and the features including a plurality of second channel values for a plurality of time slots before the first time slot;

training a channel prediction model based on the training data to obtain a trained channel prediction model; and

obtaining a channel prediction value for a prediction time based on the trained channel prediction model,

wherein the acquiring of the training data comprises:

estimating a respective channel value defined for each of a plurality of resource blocks (RBs) based on an uplink signal, the uplink signal including a pilot signal, and the pilot signal being received on the plurality of RBs through the multiple antennas; and

acquiring the label and the features from a row vector of a matrix, a column vector of the matrix having the respective channel value for each of the plurality of RBs.

10 . The method of claim 9 , wherein

the respective channel value for each of the plurality of RBs has a dimension of M×1, M being a number of the multiple antennas; and

the label and each of the features has a dimension of 1×L, L being a number of the plurality of RBs.

11 . The method of claim 9 , wherein the training of the channel prediction model includes training the channel prediction model to minimize an error between a result value of the channel prediction model and the label.

12 . The method of claim 9 , wherein a number of the plurality of time slots is set based on a moving speed of a user equipment that transmits the uplink signal.

13 . The method of claim 9 , wherein the channel prediction model includes only one model.

14 . The method of claim 9 , wherein

the first channel value for the first time slot is

q

n

+

1

α

;

and

the plurality of second channel values are

q

n

-

n

0

+

1

α

,

…

,

q

n

α

,

α being a value from 1 to M, M being a number of the multiple antennas, and n being a natural number.

15 . The method of claim 9 , further comprising:

reducing a time dimension of the training data.

16 . A base station comprising:

a transceiver configured to receive an uplink signal via multiple antennas, the uplink signal including a pilot signal on a plurality of resource blocks (RBs); and

processing circuitry connected to the transceiver, the processing circuitry being configured to,

acquire training data corresponding to the multiple antennas, the training data having a label and features, the label including a first channel value for a first time slot, and the features including a plurality of second channel values for a plurality of time slots before the first time slot,

train a channel prediction model based on the training data to obtain a trained channel prediction model, and

obtain a channel prediction value for a prediction time based on the trained channel prediction model,

wherein the processing circuitry is configured to:

estimate a respective channel value for each of the plurality of RBs based on the uplink signal; and

acquire the label and the features from a row vector of a matrix, a column vector of the matrix having the respective channel value for each of the plurality of RBs.

17 . The base station of claim 16 , wherein

the respective channel value for each of the plurality of RBs has a dimension of M×1, M being a number of the multiple antennas; and

the label and each of the features has a dimension of 1×L, L being a number of the plurality of RBs.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded May 19, 2023
From: CHOI, JUNIL; KO, BEOMSOO; KIM, HWANJIN
To: SAMSUNG ELECTRONICS CO., LTD.; KOREA ADVANCED INSTITUTE OF SCIENCE AND TECHNOLOGY
Reel/Frame 063697/0843 →
Priority Claims (1)
KR 10-2022-0096359 · Aug 2, 2022 · national
Continuity (1)
Related Publication 20240048257A1 · Feb 8, 2024
References Cited (20)
US 10805120B2 · Li et al. · 2020 [cited by applicant]
US 10892810B2 · Zhan et al. · 2021 [cited by applicant]
US 10903870B1 · Smyth et al. · 2021 [cited by applicant]
US 20030017835A1 · Bergel · 2003 [cited by examiner]
US 20140119225A1 · Hasegawa · 2014 [cited by examiner]
US 20180159602A1 · Tsai · 2018 [cited by examiner]
US 20210006989A1 · Lee · 2021 [cited by examiner]
US 20210194733A1 · Huangfu · 2021 [cited by examiner]
US 20210256428A1 · Düll · 2021 [cited by examiner]
US 20210314036A1 · Baknina et al. · 2021 [cited by applicant]
US 20210314198A1 · Kwon · 2021 [cited by applicant]
US 20230222324A1 · Iwata · 2023 [cited by examiner]
US 20240340942A1 · Saily · 2024 [cited by examiner]
CN 111917446A · 2020 [cited by applicant]
KR 1020210128841A · 2021 [cited by applicant]
Dong, P. “Deep CNN for wideband mmwave massive MIMO channel estimation using frequency correlation” [cited by applicant]
Jiang, W. “Recurrent neural network-based frequency-domain channel prediction for wideband communications” [cited by applicant]
Kim, H. “Massive MIMO channel prediction: Kalman filtering vs. machine learning” [cited by applicant]
Yuan, J. “Machine learning-based channel estimation in massive MIMO with channel aging” [cited by applicant]
Ko, B. “Machine Learning-Based Channel Prediction Exploiting Frequency Correlation in Massive MIMO Wideband Systems” IEEE. [cited by applicant]