Generalized decision feedback equalizer precoder with receiver beamforming for matrix calculations in multi-user multiple-input multiple-output wireless transmission systems

• US 8,396,107 B2
• Filed: 09/04/2009
• Issued: 03/12/2013
• Est. Priority Date: 09/04/2009
• Status: Expired due to Fees

• Current Assignee: Hitachi, Ltd.
• Original Assignee: Hitachi, Ltd.
• Inventors: Gaur, Sudhanshu
• Primary Examiner(s): Payne, David C.
• Assistant Examiner(s): SHAH, TANMAY K
• Application Number:
  US12/554,082
• Publication Number:
  US 20110058599A1
• Time in Patent Office: 1,285 Days
• Field of Search: 375/232, 375/233
• US Class Current: 375/232
• CPC Class Codes: H04B 7/0452 Multi-user MIMO systems
• CPC Class Codes: H04B 7/0465 taking power constraints at power amplifier or emission constraints, e.g. constant modulus, into account
• CPC Class Codes: H04B 7/086 using weights depending on external parameters, e.g. direction of arrival [DOA], predetermined weights or beamforming
• CPC Class Codes: H04L 25/03057 with a recursive structure H04L25/03031 takes precedence
• CPC Class Codes: H04L 25/03343 Arrangements at the transmitter end
• CPC Class Codes: H04W 52/346 distributing total power among users or channels
• CPC Class Codes: H04W 52/42 in systems with time, space, frequency or polarisation diversity

Abstract

To realize a GDFE precoder for multi-user MIMO systems, which significantly reduces the computational cost while resulting in no capacity loss, one method comprises obtaining an effective downlink (DL) channel matrix H for the DL channel after receiver processing at the user terminals; computing an uplink (UL) covariance matrix D by assuming there are as many user terminals as a number of rows in the effective DL channel matrix H; computing a filter matrix C based on the UL covariance matrix D; computing a feedforward filter matrix F based on the filter matrix C; computing an interference pre-cancellation matrix G, based on the feedforward filter matrix F and the filter matrix C, used in a transmitter at an interference pre-cancellation stage of the GDFE precoder; and processing user symbols by a decision feedback equalizing stage of the GDFE precoder to produce filtered vector symbols.

Claims

1. A method for processing user symbols with a generalized decision feedback equalizer (GDFE) based precoder in a base station of a multi-user multiple-input multiple-output (MU-MIMO) wireless system having K user terminals (UTs) which communicate with the base station via an uplink (UL) channel and a corresponding downlink (DL) channel, the method comprising:

a processor obtaining an effective downlink (DL) channel matrix H for the DL channel after receiver processing at the user terminals;

computing an uplink (UL) covariance matrix D by assuming there are as many user terminals as a number of rows in the effective DL channel matrix H, the UL covariance matrix D being a diagonal matrix;

computing a filter matrix C based on the UL covariance matrix D;

computing a feedforward filter matrix F based on the filter matrix C;

computing an interference pre-cancellation matrix G, based on the feedforward filter matrix F and the filter matrix C, used in a transmitter at an interference pre-cancellation stage of the GDFE precoder; and

processing by the processor user symbols by a decision feedback equalizing stage of the GDFE precoder to produce filtered vector symbols;

wherein the effective downlink (DL) channel matrix is H=

H=[Ĥ

₁^H,Ĥ

₁^H, . . . , Ĥ

_K^H],

where Ĥ

_k is an effective DL channel sub-matrix for the k^th UT

Ĥ

_k=S_kV_k^H,

where S_k and V_k are matrices obtained from a singular value decomposition (SVD) of and estimated DL channel matrix H_k for the k^th UT

H_k=U_kS_kV_k^H.

2. The method of claim 1, wherein U_k^H is used for receiver processing at the k^th UT and the effective DL channel sub-matrix for the k^th UT, Ĥ

_k, is provided to the base station by the k^th UT.

3. The method of claim 1, wherein

C=[√

{square root over (Σ

^†

)}−

λ

√

{square root over (Σ

)}]V^HH^HD,

where superscript †

denotes a Moore-Penrose Generalized Inverse, and where V is a unitary matrix and Σ

is a diagonal matrix with non-negative entries in an Eigen Value Decomposition (EVD) of an input covariance matrix S_xx represented as

S_xx=VΣ

V^H,

and λ

is a UL/DL duality variable for a given total transmit power P_t,

λ

=trace(I−

[H_DL^HDH_DL+I]<sup>−

1</sup>)/P_t,

where V is a unitary matrix and Σ

is a diagonal matrix with non-negative entries, and the input covariance matrix S_xx is computed as

4. The method of claim 3,

wherein C=MR denotes QR decomposition (QRD) of C, M being a unitary matrix, and R being an upper right triangular matrix;

wherein the feedforward filter matrix F is

F=Diagonal(R); and

wherein the interference pre-cancellation matrix G is

G=FR^†

,

where superscript †

denotes a Moore-Penrose Generalized Inverse.

5. The method of claim 4, further comprising:

computing a transmit filter matrix B for a transmit filter

B=VΣ

^1/2M;

passing the filtered vector symbols through the transmit filter to produce an output of transmitted signals;

directing the output of the transmit filter to the channel represented by the effective DL channel matrix H through which communications occur in the wireless system with the user terminals.

6. The method of claim 1, wherein processing user symbols by a decision feedback equalizing stage of the GDFE precoder to produce filtered vector symbols comprises:

directing the user symbols through a modulo unit disposed in a feedforward path to produce the filtered vector symbols which are fed back through an interference pre-cancellation block disposed in a feedback path, the interference pre-cancellation block being denoted by I-G; and

subtracting an output signal of the interference pre-cancellation block from the user symbols which are applied to the modulo unit in the feedforward path.

7. A generalized decision feedback equalizer (GDFE) based precoder in a base station (BS) of a multi-user multiple-input multiple-output (MU-MIMO) wireless system having K user terminals (UTs) which communicate with the base station via an uplink (UL) channel and a corresponding downlink (DL) channel, the GDFE precoder comprising:

a processor for calculating;

a feedforward path;

a feedback path; and

an interference pre-cancellation block denoted by I-G disposed in the feedback path, I being an identity matrix, G being an interference pre-cancellation matrix;

wherein the interference pre-cancellation matrix G is computed based on a feedforward filter matrix F and a filter matrix C, the feedforward filter matrix F is computed based on the filter matrix C, the filter matrix C is computed based on an uplink (UL) covariance matrix D, the UL covariance matrix D is computed by assuming there are as many user terminals as a number of rows in an effective downlink (DL) channel matrix H, the UL covariance matrix D is a diagonal matrix, and the effective DL channel matrix H is obtained after receiver processing at the user terminals;

wherein the effective downlink (DL) channel matrix is

H=[Ĥ

₁^H,Ĥ

₂^H, . . . , Ĥ

_K^H]

where Ĥ

_k is an effective DL channel sub-matrix for the k^th UT

Ĥ

_k=S_kV_k^H,

where S_k and V_k are matrices obtained from a singular value decomposition (SVD) of and estimated DL channel matrix H_k for the k^th UT

H_k=U_kS_kV_k^H.

8. The GDFE precoder of claim 7, wherein U_k^H is used for receiver processing at the k^th UT and the effective DL channel sub-matrix for the k^th UT, Ĥ

_k, is provided to the base station by the k^th UT.

9. The GDFE precoder of claim 7, wherein

C=[√

{square root over (Σ

^†

)}−

λ

√

{square root over (Σ

)}]V^HH^HD,

where superscript †

denotes a Moore-Penrose Generalized Inverse, and where V is a unitary matrix and Σ

is a diagonal matrix with non-negative entries in an Eigen Value Decomposition (EVD) of an input covariance matrix S_xx represented as

S_xx=VΣ

V^H,

and λ

is a UL/DL duality variable for a given total transmit power P_t,

λ

=trace(I−

[H_DL^HDH_DL+I]<sup>−

1</sup>)/P_t,

where V is a unitary matrix and Σ

is a diagonal matrix with non-negative entries, and the input covariance matrix S_xx is computed as

10. The GDFE precoder of claim 9,

wherein C=MR denotes QR decomposition (QRD) of C, M being a unitary matrix, and R being an upper right triangular matrix;

wherein the feedforward filter matrix F is

F=Diagonal(R); and

wherein the interference pre-cancellation matrix G is

G=FR^†

,

where superscript †

denotes a Moore-Penrose Generalized Inverse.

11. The GDFE precoder of claim 7, further comprising:

a modulo unit disposed in the feedforward path to produce a stream of filtered vector symbols X which are fed back through the interference pre-cancellation block disposed in the feedback path.

12. The GDFE precoder of claim 11,

wherein an output signal of the interference pre-cancellation block is subtracted from a stream of user symbols and applied to the modulo unit in the feedforward path.

13. The GDFE precoder of claim 12, further comprising:

a transmit filter represented by the transmit filter matrix B for filtering the stream of filtered vector symbols X produced by the modulo unit disposed in the feedforward path;

wherein the transmit filter matrix B is

B=VΣ

^1/2M.

14. A MU-MIMO wireless system comprising:

a base station including the GDFE precoder of claim 13;

a plurality of K user terminals; and

a channel, represented by the DL channel matrix H through which communications occur in the wireless system with the user terminals, to receive an output of the transmit filter.

15. A generalized decision feedback equalizer (GDFE) based precoder in a base station (BS) of a multi-user multiple-input multiple-output (MU-MIMO) wireless system having K user terminals (UTs) which communicate with the base station via an uplink (UL) channel and a corresponding downlink (DL) channel, the GDFE precoder comprising:

a decision feedback equalizing stage for processing, by a processor, user symbols to produce filtered vector symbols, the decision feedback equalizing stage including an interference pre-cancellation stage having an interference pre-cancellation matrix G used in a transmitter at the interference pre-cancellation stage; and

a transmit filter represented by a transmit filter matrix B for processing the filtered vector symbols after the decision feedback equalizing stage to produce an output of transmitted signals to be directed to the DL channel represented by the effective DL channel matrix H through which communications occur in the wireless system with the user terminals;

wherein the interference pre-cancellation matrix G is computed based on a feedforward filter matrix F and a filter matrix C, the feedforward filter matrix F is computed based on the filter matrix C, the filter matrix C is computed based on an uplink (UL) covariance matrix D, the UL covariance matrix D is computed by assuming there are as many user terminals as a number of rows in an effective downlink (DL) channel matrix H, the UL covariance matrix D is a diagonal matrix, and the effective DL channel matrix H is obtained after receiver processing at the user terminals;

wherein the effective downlink (DL) channel matrix is

H=[Ĥ

₁^H,Ĥ

₂^H, . . . , Ĥ

_K^H]

where Ĥ

_k is an effective DL channel sub-matrix for the k^th UT

Ĥ

_k=S_kV_k^H,

where S_k and V_k are matrices obtained from a singular value decomposition (SVD) of and estimated DL channel matrix H_k for the k^th UT

H_k=U_kS_kV_k^H.

16. The GDFE precoder of claim 15, wherein U_k^H is used for receiver processing at the k^th UT and the effective DL channel sub-matrix for the k^th UT, Ĥ

_k, is provided to the base station by the k^th UT.

17. The GDFE precoder of claim 15, wherein

C=[√

{square root over (Σ

^†

)}−

λ

√

{square root over (Σ

)}]V^HH^HD,

where superscript †

denotes a Moore-Penrose Generalized Inverse, and where V is a unitary matrix and Σ

is a diagonal matrix with non-negative entries in an Eigen Value Decomposition (EVD) of an input covariance matrix S_xx represented as

S_xx=VΣ

V^H,

and λ

is a UL/DL duality variable for a given total transmit power P_t,

λ

=trace(I−

[H_DL^HDH_DL+I]<sup>−

1</sup>)/P_t,

where V is a unitary matrix and Σ

is a diagonal matrix with non-negative entries, and the input covariance matrix S_xx is computed as

18. The GDFE precoder of claim 17,

wherein C=MR denotes QR decomposition (QRD) of C, M being a unitary matrix, and R being an upper right triangular matrix;

wherein the feedforward filter matrix F is

F=Diagonal(R); and

wherein the interference pre-cancellation matrix G is

G=FR^†

,

where superscript †

denotes a Moore-Penrose Generalized Inverse.

19. The GDFE precoder of claim 15,

wherein the decision feedback equalizing stage includes a modulo unit disposed in a feedforward path to produce a stream of filtered vector symbols X which are fed back through an interference pre-cancellation block disposed in a feedback path, the interference pre-cancellation block denoted by I-G disposed in the feedback path, wherein an output signal of the interference pre-cancellation block is subtracted from a stream of user symbols and applied to the modulo unit in the feedforward path.

20. A MU-MIMO wireless system comprising:

a base station including the GDFE precoder of claim 15;

a plurality of K user terminals; and

a channel, represented by the effective DL channel matrix H through which communications occur in the wireless system with the user terminals, to receive the output of the transmit filter.