Granted Patent B2

US 11,418,239 · App. 17/213,490 · Granted Aug 16, 2022

Inventors: Logeshwaran Vijayan (Bengaluru, IN); Karthik Muralidhar (Bengaluru, IN)

Assignee: Mavenir Systems, Inc.

H04B7/0456G06N5/04G06N20/00H04L5/0048H04L25/0224

Abstract

Claims (93)

1. A method of decoding an uplink in a multiple input multiple output wireless communication system in an open radio access network having an open distributed unit (O-DU) and an open radio unit (O-RU), said method comprising:

said O-DU:

constructing a combining matrix for a resource block,

wherein said constructing comprises multiplying a received signal vector with a whitening vector, thus converting colored interference in said received signal vector into white interference, and

wherein said whitening vector is based on channel estimates from past sounding reference signal symbols; and

sending said combining matrix to said O-RU; and

said O-RU:

utilizing said combining matrix to compress signals on N R antennas per subcarrier into M values, where N R is a number of antennas, and M is less than N R ; and

sending said M values per subcarrier to said O-DU.

2. The method of claim 1 ,

wherein said constructing comprises reducing interference by multiplying a received signal vector with a nulling vector, and

wherein said nulling vector is based on channel estimates from past sounding reference signal symbols.

3. The method of claim 1 , wherein said constructing comprises:

estimating channel values from past sounding reference signals, thus yielding estimated channels; and

using a conjugate of said estimated channels in said combining matrix.

4. The method of claim 1 , wherein said method further comprises sending demodulation reference signals from said O-RU to said O-DU.

5. The method of claim 1 , further comprising using said combining matrix to send said M values for demodulation reference signals from said O-RU to said O-DU.

6. The method of claim 1 , further comprising:

learning linear prediction (LP) coefficients from sounding reference signals;

using said LP coefficients for said constructing said combining matrix; and

using said combining matrix to send said M values for demodulation reference signals from said O-RU to said O-DU.

7. The method of claim 1 , wherein said constructing comprises:

learning linear prediction (LP) coefficients from sounding reference signals (SRS);

using said LP coefficients to predict a channel in a current subframe based on channel values from past SRS symbols, thus yielding a predicted channel for said current subframe;

computing a nulling vector from said predicted channel;

computing a conjugated channel vector from conjugated values of said predicted channel; and

constructing said combining matrix from said whitening vector, said nulling vector, and said conjugated channel vector.

8. The method of claim 1 , wherein said constructing comprises:

learning linear prediction (LP) coefficients from sounding reference signals (SRS) and demodulation reference signals;

using said LP coefficients to predict a channel in a current subframe based on channel values from past SRS symbols, thus yielding a predicted channel for said current subframe;

computing a nulling vector from said predicted channel;

computing a conjugated channel vector from conjugated values of said predicted channel; and

constructing said combining matrix from said whitening vector, said nulling vector, and said conjugated channel vector.

9. The method of claim 1 , wherein said constructing comprises:

learning linear prediction (LP) coefficients from sounding reference signals (SRS) and compressed demodulation reference signals;

using said LP coefficients to predict a channel in a current subframe based on channel values from past SRS symbols, thus yielding a predicted channel for said current subframe;

computing a nulling vector from said predicted channel;

computing a conjugated channel vector from conjugated values of said predicted channel; and

constructing said combining matrix from said whitening vector, said nulling vector, and said conjugated channel vector.

10. A system comprising:

an open distributed unit (O-DU); and

an open radio unit (O-RU),

wherein said O-DU performs operations of:

constructing a combining matrix for a resource block,

wherein said constructing comprises multiplying a received signal vector with a whitening vector, thus converting colored interference in said received signal vector into white interference, and

wherein said whitening vector is based on channel estimates from past sounding reference signal symbols; and

sending said combining matrix to said O-RU; and

wherein said O-RU performs operations of:

utilizing said combining matrix to compress signals on N R antennas per subcarrier into M values, where N R is a number of antennas, and M is less than N R ; and

sending said M values per subcarrier to said O-DU.

11. The system of claim 10 ,

wherein said constructing comprises reducing interference by multiplying a received signal vector with a nulling vector, and

wherein said nulling vector is based on channel estimates from past sounding reference signal symbols.

12. The system of claim 10 , wherein said constructing comprises:

estimating channel values from past sounding reference signals, thus yielding estimated channels; and

using a conjugate of said estimated channels in said combining matrix.

13. The system of claim 10 , wherein said O-RU performs a further operation of sending demodulation reference signals to said O-DU.

14. The system of claim 10 , wherein said O-RU performs a further operation of using said combining matrix to send said M values for demodulation reference signals to said O-DU.

15. The system of claim 10 ,

wherein said constructing comprises:

learning linear prediction (LP) coefficients from sounding reference signals; and

using said LP coefficients for said constructing said combining matrix, and

wherein said O-DU performs a further operation of:

using said combining matrix to send said M values for demodulation reference signals to said O-DU.

16. The system of claim 10 , wherein said constructing comprises:

learning linear prediction (LP) coefficients from sounding reference signals (SRS);

using said LP coefficients to predict a channel in a current subframe based on channel values from past SRS symbols, thus yielding a predicted channel for said current subframe;

computing a nulling vector from said predicted channel;

computing a conjugated channel vector from conjugated values of said predicted channel; and

constructing said combining matrix from said whitening vector, said nulling vector, and said conjugated channel vector.

17. The system of claim 10 , wherein said constructing comprises:

learning linear prediction (LP) coefficients from sounding reference signals (SRS) and demodulation reference signals;

using said LP coefficients to predict a channel in a current subframe based on channel values from past SRS symbols, thus yielding a predicted channel for said current subframe;

computing a nulling vector from said predicted channel;

computing a conjugated channel vector from conjugated values of said predicted channel; and

constructing said combining matrix from said whitening vector, said nulling vector, and said conjugated channel vector.

18. The system of claim 10 , wherein said constructing comprises:

learning linear prediction (LP) coefficients from sounding reference signals (SRS) and compressed demodulation reference signals;

using said LP coefficients to predict a channel in a current subframe based on channel values from past SRS symbols, thus yielding a predicted channel for said current subframe;

computing a nulling vector from said predicted channel;

computing a conjugated channel vector from conjugated values of said predicted channel; and

constructing said combining matrix from said whitening vector, said nulling vector, and said conjugated channel vector.

19. A system comprising:

an open distributed unit (O-DU) having a first processor and a first memory that contains first instructions; and

an open radio unit (O-RU) having a second processor and a second memory that contains second instructions that are readable by said second processor,

wherein said first instructions are readable by said first processor to cause said first processor to perform operations of:

constructing a combining matrix for a resource block,

wherein said constructing comprises multiplying a received signal vector with a whitening vector, thus converting colored interference in said received signal vector into white interference, and

wherein said whitening vector is based on channel estimates from past sounding reference signal symbols; and

sending said combining matrix to said O-RU; and

wherein said second instructions are readable by said second processor to cause said second processor to perform operations of:

utilizing said combining matrix to compress signals on N R antennas per subcarrier into M values, where N R is a number of antennas, and M is less than N R ; and

sending said M values per subcarrier to said O-DU.

Assignments (16)

Priority Claims (1)

Continuity (1)