Granted Patent B2

US 12,301,387 · App. 18/446,386 · Granted May 13, 2025

Inventors: Yeqing Hu (Allen, TX); Yang Li (Plano, TX); Jianzhong Zhang (Dallas, TX)

Assignee: Samsung Electronics Co., Ltd.

H04L25/0242H04L25/021

Abstract

Claims (82)

1. A base station (BS) comprising:

a transceiver configured to receive a reference signal from a user equipment (UE); and

a processor operably coupled to the transceiver, the processor configured to:

perform a linear transformation based on the received reference signal;

select a basis set based on the linear transformation;

select a set of kernels based on the selected basis set and the linear transformation; and

reconstruct a channel based on the selected set of kernels.

2. The BS of claim 1 , wherein:

the basis set is selected based on one of a time correlation corresponding with the linear transformation and a kernel power density corresponding with the linear transformation;

the set of kernels is selected based on at least one of the time correlation corresponding with the linear transformation and an energy parameter; and

the processor is further configured to:

generate a predicted set of kernels based on the selected set of kernels; and

construct a predicted channel based on the predicted set of kernels, wherein the channel is reconstructed based on the predicted channel.

3. The BS of claim 2 , wherein the processor is further configured to:

categorize the set of kernels as fast changing or slow changing; and

perform a prediction complexity reduction operation based on the categorization, wherein the predicted set of kernels are based on the complexity reduction operation.

4. The BS of claim 2 , wherein the processor is further configured to:

categorize the set of kernels based on at least one of a Doppler spectrum estimation and a time domain correlation estimation; and

perform a prediction complexity reduction operation based on the categorization, wherein the predicted set of kernels are based on the complexity reduction operation.

5. The BS of claim 2 , wherein the processor is further configured to:

determine a set of residual unselected kernels corresponding with the linear transformation;

apply a scaling factor to the set of residual unselected kernels; and

construct a projected channel based on the set of scaled set of residual unselected kernels,

wherein reconstructing the channel comprises adding the projected channel to the predicted channel.

6. The BS of claim 2 , wherein the processor is further configured to:

determine a set of residual unselected kernels corresponding with the linear transformation;

apply a scaling factor to the set of residual unselected kernels; and

add the scaled set of residual unselected kernels to the predicted set of kernels,

wherein the predicted channel is based on the addition of the scaled set of residual unselected kernels to the predicted set of kernels.

7. The BS of claim 1 , wherein:

the basis set is selected based on one of a constant initial offset related to the linear transformation and a kernel power density related to the linear transformation;

the set of kernels is selected based on an energy parameter; and

the processor is further configured to perform a de-noising operation on the set of kernels, wherein the channel is reconstructed based on the de-noised set of kernels.

8. The BS of claim 1 , wherein the linear transformation is a canonical model (CM) basis projection.

9. A method performed by a base station (BS), the method comprising:

receiving a reference signal from a user equipment (UE);

performing a linear transformation based on the received reference signal;

selecting a basis set based on the linear transformation;

selecting a set of kernels based on the selected basis set and the linear transformation; and

reconstructing a channel based on the selected set of kernels.

10. The method of claim 9 , wherein:

the basis set is selected based on one of a time correlation corresponding with the linear transformation and a kernel power density corresponding with the linear transformation;

the set of kernels is selected based on at least one of the time correlation corresponding with the linear transformation and an energy parameter, wherein the method further comprises:

generating a predicted set of kernels based on the selected set of kernels; and

constructing a predicted channel based on the predicted set of kernels, wherein the channel is reconstructed based on the predicted channel.

11. The method of claim 10 , further comprising:

categorizing the set of kernels as fast changing or slow changing; and

performing a prediction complexity reduction operation based on the categorization, wherein the predicted set of kernels are based on the complexity reduction operation.

12. The method claim 10 , further comprising:

categorizing the set of kernels based on at least one of a Doppler spectrum estimation and a time domain correlation estimation; and

performing a prediction complexity reduction operation based on the categorization, wherein the predicted set of kernels are based on the complexity reduction operation.

13. The method of claim 10 , further comprising:

determining a set of residual unselected kernels corresponding with the linear transformation;

applying a scaling factor to the set of residual unselected kernels; and

constructing a projected channel based on the set of scaled set of residual unselected kernels,

wherein reconstructing the channel comprises adding the projected channel to the predicted channel.

14. The method of claim 10 , further comprising:

determining a set of residual unselected kernels corresponding with the linear transformation;

applying a scaling factor to the set of residual unselected kernels; and

adding the scaled set of residual unselected kernels to the predicted set of kernels,

wherein the predicted channel is based on the addition of the scaled set of residual unselected kernels to the predicted set of kernels.

15. The method of claim 9 , wherein:

the basis set is selected based on one of a constant initial offset related to the linear transformation and a kernel power density related to the linear transformation;

the set of kernels is selected based on an energy parameter; and

the method further comprises performing a de-noising operation on the set of kernels, wherein the channel is reconstructed based on the de-noised set of kernels.

16. The method of claim 9 , wherein the linear transformation is a canonical model (CM) basis projection.

17. A non-transitory computer readable medium embodying a computer program, the computer program comprising program code that, when executed by a processor of a device, causes the device to:

receive a reference signal from a user equipment (UE);

perform a linear transformation based on the received reference signal;

select a basis set based on the linear transformation;

select a set of kernels based on the selected basis set and the linear transformation; and

reconstruct a channel based on the selected set of kernels.

18. The non-transitory computer readable medium of claim 17 , wherein:

the basis set is selected based on one of a time correlation corresponding with the linear transformation and a kernel power density corresponding with the linear transformation;

the set of kernels is selected based on at least one of the time correlation corresponding with the linear transformation and an energy parameter, wherein the computer program further comprises program code that, when executed by the processor of the device, causes the device to:

generate a predicted set of kernels based on the selected set of kernels; and

construct a predicted channel based on the predicted set of kernels, wherein the channel is reconstructed based on the predicted channel.

19. The non-transitory computer readable medium of claim 17 , wherein:

the basis set is selected based on one of a constant initial offset related to the linear transformation and a kernel power density related to the linear transformation;

the set of kernels is selected based on an energy parameter; and

the processor is further configured to perform a de-noising operation on the set of kernels, wherein the channel is reconstructed based on the de-noised set of kernels.

20. The non-transitory computer readable medium of claim 17 , wherein the linear transformation is a canonical model (CM) basis projection.

Assignments (1)

Continuity (2)

References Cited (11)