Efficient reduced complexity windowed optimal time domain equalizer for discrete multitone-based DSL modems

• US 7,180,969 B1
• Filed: 04/18/2006
• Issued: 02/20/2007
• Est. Priority Date: 02/28/2001
• Status: Expired due to Fees

• Current Assignee: AT&T Corporation (AT&T, Inc.)
• Original Assignee: AT&T Corporation (AT&T, Inc.)
• Inventors: Erving, Richard H., Sankaranarayanan, Lalitha
• Primary Examiner(s): Corrielus; Jean B.
• Application Number:
  US11/405,860
• Time in Patent Office: 308 Days
• Field of Search: 375229-232, 375/285, 375/316, 375/346, 375/348, 375/350
• US Class Current: 375/348
• CPC Class Codes: H04L 1/0057 Block codes H04L1/0061, H04L1/0064 take precedence
• CPC Class Codes: H04L 1/006 Trellis-coded modulation
• CPC Class Codes: H04L 1/0065 Serial concatenated codes
• CPC Class Codes: H04L 2025/03414 Multicarrier
• CPC Class Codes: H04L 2025/03445 Time domain
• CPC Class Codes: H04L 25/03114 non-adaptive, i.e. not adjustable, manually adjustable, or adjustable only during the reception of special signals
• CPC Class Codes: H04L 5/0007 the frequencies being orthogonal, e.g. OFDM(A), DMT

Abstract

An algorithm for computing an efficient, reduced complexity, windowed optimal linear time domain equalizer for a dispersive channel comprises the steps of determining a window of maximum energy in the impulse response of length equal to or less than a number of cyclic prefix samples associated with a received digital data signal, computing the corresponding inside and outside matrices, performing an inverse Cholesky decomposition of the inside matrix, creating a resultant matrix as the product of the outer and the upper and lower square root inner matrix, followed by Householder reduction and QL transformation to thereby compute the time domain equalizer as the linear transformation of the eigenvector corresponding to the smallest eigenvalue at the receiver. The smallest eigenvalue is determined using the aforementioned orthogonal transformations without determining all the eigenvalues efficiently but without the loss accuracy associated with iterative methods like the conventional power method. The algorithm may be most conveniently implemented, for example, in the form of a thirty-two bit digital signal processor at a data receiver.

Claims

1. A method for suppressing noise resulting from inter-symbol interference (ISI) in a receiver for use in a digital data transmission system, comprising the steps of:

during an initialization period;

determining a time domain impulse response of a channel from where said receiver receives received data;

developing two square matrices from the impulse response;

computing an eigenvector that corresponds to a smallest eigenvalue of a matrix derived from the two square matrices; and

transforming the eigenvector to develop a time domain equalizer at said receiver; and

during period of data transmission;

passing said received data through the time domain equalizer.

2. The method according to claim 1, wherein the initialization period is also used for handshaking, synchronization and channel parameter initialization between a transmitter that is the source of said received data and the receiver.

3. The method of claim 1 wherein the step of developing two square matrices comprises the step of selecting a tap length of a filter of the equalizer.

4. The method according to claim 1 where the step of computing includes use of an orthogonal transformation method to determine the smallest eigenvalue and the eigenvector.

5. The method of claim 1 where the step of computing includes use of an inverse Cholesky decomposition function.

6. The method of claim 1 where the step of computing includes use of a Householder reduction function.

7. The method of claim 1 where the step of computing includes use of a QL transform function.

8. The method of claim 1 wherein said smallest eigenvalue is determined without determining an inverse matrix.

9. The method of claim 3 wherein said tap length is chosen to be equal to or less than the length of a guard band of transmitted digital data.

10. The method of claim 1 wherein said smallest eigenvalue is determined without utilizing a power method or other power iteration process.

11. The method of claim 1 where the step of computing includes performing a QL transform on ascending values of a diagonal of a matrix.

12. The method of claim 1 implemented via a digital signal processor of a digital subscriber line communication device.