SYSTEM AND METHOD FOR GENERATING CODEBOOK FOR ANALOG BEAMFORMING

• US 20200136694A1
• Filed: 03/14/2019
• Published: 04/30/2020
• Est. Priority Date: 10/30/2018
• Status: Active Grant

• Current Assignee: Samsung Electronics Co. Ltd.
• Original Assignee: Samsung Electronics Co. Ltd.
• Inventors: Ganji, Mehdi, Cheng, Hongbing, Song, Kee-Bong, Zhan, Qi
• Granted Patent: US 10,742,282 B2
• Time in Patent Office: Days
• CPC Class Codes: H04B 7/0456 Selection of precoding matrices or codebooks, e.g. using matrices antenna weighting
• CPC Class Codes: H04B 7/0482 Adaptive codebooks
• CPC Class Codes: H04B 7/0617 for beam forming
• CPC Class Codes: H04B 7/0695 using beam selection

Abstract

A method of designing a codebook for analog beamforming. In some embodiments, the method includes: generating a plurality of training points, based on a predefined distribution, each training point being a channel vector; generating, for each of a plurality of codewords, a respective initial value; assigning, in a first assignment operation, each of the training points to a respective desired codeword, the desired codeword maximizing a metric function of a beamforming gain for the training point over the codewords; updating, in a first updating operation, each of the codewords according to the training points assigned to it; and determining whether convergence is achieved; and in response to determining that convergence is achieved, determining a final codebook.

Claims

1. A method of designing a codebook for analog beamforming, the method comprising:

generating a plurality of training points, each training point being a channel vector;

generating, for each of a plurality of codewords, a respective initial value;

assigning, in a first assignment operation, each of the training points to a respective desired codeword, the desired codeword maximizing a metric function of a beamforming gain for the training point over the codewords;

updating, in a first updating operation, each of the codewords according to the training points assigned to each corresponding codeword, the updating comprising executing a plurality of iterations of a gradient descent updating rule; and

repeating the assigning and the updating until commence is achieved.

2. The method of claim 1, wherein the generating of a plurality of training points comprises generating a plurality of training points, based on a predefined distribution.

3. The method of claim 1 further comprising determining whether convergence is achieved.

4. The method of claim 3, further comprising, in response to determining that convergence is achieved, determining the codebook containing the updated codewords.

5. The method of claim 4, further comprising,

in response to determining, after the first updating operation, that convergence is not achieved,

assigning, in a second assignment operation, each of the training points to a respective desired codeword, the desired codeword maximizing a metric function of a beamforming gain for the training point over the codewords;

and

updating, in a second updating operation, each of the codewords according to the training points assigned to it.

6. The method of claim 3, wherein the determining whether convergence is achieved comprises determining one from a combination of:

whether a number of iterations equals a set number;

whether a metric is larger than a set value;

whether a change in a metric, between two successive iterations, is smaller than a set threshold; and

whether a change in the plurality of codewords between two iterations is smaller than a set threshold.

7. The method of claim 1, wherein the generating, for each of the plurality of codewords, the respective initial value, comprises performing a discrete Fourier transform.

8. The method of claim 1, wherein the metric function returns one or more measures selected from a combination of:

a measure of average beamforming, gain, a measure of capacity, a measure of reciprocal bit error rate, and a measure of coverage.

10. The method of claim 1, wherein the updating, of a codeword of the plurality of codewords, comprises updating a value of the codeword to:

9. A method of designing a codebook for analog beamforming, the method comprising:

generating a plurality of training points, each training point being a channel vector;

generating, for each of a plurality of codewords, a respective initial value;

assigning, in a first assignment operation, each of the training points to a respective desired codeword, the desired codeword maximizing a metric function of a beamforming gain for the training point over the codewords; and

updating, in a first updating operation, each of the codewords according to the training points assigned to each corresponding codeword, the updating comprising executing a plurality of iterations of a gradient descent updating rule,

wherein the training points include one selected from a combination of;

empirical points;

points uniformly distributed over a sphere;

pseudorandom points generated based on a uniform spherical distribution;

points uniformly distributed over a range of horizontal angles and uniformly distributed over a range of down-tilt angles; and

pseudorandom points generated based on a distribution that is uniform over a range of horizontal angles and uniform over a range of down-tilt angles.

11. A system for communicating, comprising:

an antenna system, comprising a plurality of antennas and a corresponding plurality of phase shifters,

a processing circuit electrically coupled to the antenna system, wherein the processing circuit is configured to;

generate a plurality of training points, based on a predefined distribution, each training point being a channel vector;

generate, for each of a plurality of codewords, a respective initial value;

assign, in a first assignment operation, each of the training points to a respective desired codeword, the desired codeword maximizing a metric function of beamforming gain for the training point over the codewords;

update, in a first updating operation, each of the codewords according to the training points assigned to each corresponding codeword; and

repeat the assigning and the updating until convergence is achieved.

12. The system of claim 11, wherein the predefined distribution includes a uniform distribution.

13. The system of claim 11, wherein the processing circuit is further configured to determine whether convergence is achieved.

14. The system of claim 13, wherein the processing circuit is further configured to, in response to determining that convergence is achieved, determine the codebook containing the updated codewords.

15. The system of claim 14, wherein the processing circuit is further configured to:

in response to determining, after the first updating operation, that convergence is not achieved,

assign, in a second assignment operation, each of the training points to a respective desired codeword, the desired codeword maximizing a metric function of a beamforming gain for the training point over the codewords; and

update, in a second updating operation, each of the codewords according to the training points assigned to it.

16. The system of claim 13, wherein the determining whether convergence is achieved comprises determining one from a combination of:

whether a number of iterations equals a set number;

whether a metric is larger than a set value;

whether a change in a metric, between two successive iterations, is smaller than a set threshold; and

whether a change in the plurality of codewords between two iterations smaller than a set threshold.

17. The system of claim 11, wherein the generating, for each of the plurality of codewords, the respective initial value, comprises performing a discrete Fourier transform.

18. The system of claim 11, wherein the metric function returns one or more measures selected from a combination of:

a measure of average beamforming gain, a measure of capacity, a measure of reciprocal bit error rate, and a measure of coverage.

19. The system of claim 11, wherein the training points include one selected from a combination of:

empirical points;

points uniformly distributed over a sphere;

pseudorandom points generated based on a uniform spherical distribution;

points uniformly distributed over a range of horizontal angles and uniformly distributed over a range of down-tilt angles; and

pseudorandom points generated based on a distribution that is uniform over a range of horizontal angles and uniform over a range of down-tilt angles.

20. The system of claim 11, wherein the updating, of a codeword of the plurality of codewords, comprises updating a value of the codeword to: