Multiple-filter equalizer for structured digitally modulated signals

• US 5,524,124 A
• Filed: 11/17/1993
• Issued: 06/04/1996
• Est. Priority Date: 11/17/1993
• Status: Expired due to Term

• Current Assignee: Exelis Incorporated (L3Harris Technologies, Inc.)
• Original Assignee: Signal Science, Inc. (L3Harris Technologies, Inc.)
• Inventors: Koenig, Carolyn T.
• Primary Examiner(s): Chin, Stephen
• Assistant Examiner(s): Bocure, Tesfaldet
• Application Number:
  US08/153,451
• Time in Patent Office: 930 Days
• Field of Search: 375/11, 375/14, 375/13, 375/229, 375/232, 375/233, 375/235, 364/724.19, 364/724.2, 395/24, 333/18, 333/28 R
• US Class Current: 375/229
• CPC Class Codes: H04L 25/03038 with a non-recursive structure H04L25/03031 takes precedence

Abstract

An equalization and demodulation method for a structured digitally modulated signal provides a multiple filter equalizer, which comprises multiple, parallel, automatically adjustable processors. The multiple filter equalizer is applicable to a structured digitally modulated signal, such as a signal from time division multiplexing (TDM) of multiple data sources. The multiple filter equalizer exploits the repetitive structure of TDM signal data by employing multiple parallel processors each constructed according to the specific requirements at the position in the frame of a symbol to be demodulated. Each processor comprises one or more adaptive digital transversal filters, one or more nonlinear threshold operators, and a symbol decision operator. The transversal filters equalize the data, remove interfering signals, reduce intersymbol interference, and mitigate multipath and other propagation effects. After the samples corresponding to one frame of information symbols are equalized in parallel, the equalized samples are demodulated in parallel to generate demodulated information symbols for that frame. The method is applicable to adaptive equalization both with and without a known training sequence.

Claims

1. A structure for equalizing and demodulating a structured signal, said structured signal having a predetermined number of symbols transmitted in a frame, said structure comprising a separate processor for each symbol position in said frame, each processor comprising:

a transversal digital filter computing a weighted sum of a predetermined number of samples of said structured signal and providing said weighted sum as an equalized sample; and

a decision device, coupled to receive said equalized sample, for determining from said equalized sample, independently of symbol decisions at other symbol positions in said frame, a symbol transmitted at said symbol position in said frame.

2. A structure as in claim 1, wherein each processor further comprises means, coupled to receive one or more of said samples, said symbol and said weighted sum, for updating the weights of said transversal digital filter.

3. A structure as in claim 1, wherein each processor further comprises means, coupled to receive said symbol and said weighted sum, for correcting a phase error in said weighted sum.

4. A structure as in claim 1, wherein, within each processor, each transversal filter is one of a plurality of transversal filters in a first level structure of said processor, said processor further comprising:

a plurality of non-linear operators, each non-linear operator operating on an output value of a corresponding one of said plurality of transversal filters to provide a first level equalized output value; and

a transversal digital filter in a second level structure receiving said first level equalized output values for computing a weighted sum of said equalized output values to obtain said equalized sample.

5. A structure as in claim 2, wherein said weights are updated in accordance with an adaptive algorithm.

6. A structure as in claim 4, wherein said weights are updated in accordance with a backpropagation algorithm.

7. A structure as in claim 4, wherein said non-linear operators each comprise means for providing a differentiable function.

8. A method for equalizing and demodulating a structured signal, said structured signal having a predetermined number of symbols transmitted in a frame, said method comprises, for each symbol position in said frame, the steps of:

computing in a separate transversal digital filter a weighted sum of a predetermined number of samples of said structured signal and providing said weighted sum as an equalized sample; and

determining in a decision device, coupled to receive said equalized sample, independently of symbol decisions at other symbol positions in said frame, a symbol transmitted at said symbol position in said frame.

9. A method as in claim 8, further comprising the step of updating the weights of said transversal digital filter.

10. A method in claim 8, further comprising the step of correcting a phase error in said weighted sum using said symbol and said weighted sum.

11. A method as in claim 8, wherein said step of computing in a separate transversal digital filter provides a plurality of transversal filters in a first level structure, said method further comprising the steps of:

providing a plurality of non-linear operators, each non-linear operator operating on an output value of a corresponding one of said plurality of transversal filters to provide a first level equalized output value; and

computing in a transversal digital filter in a second level a weighted sum of said first level equalized output values to obtain said equalized sample.

12. A method as in claim 9, wherein said weights are updated in accordance with an adaptive algorithm.

13. A method as in claim 11, wherein said weights are updated in accordance with a backpropagation algorithm.

14. A method as in claim 11, wherein said non-linear operator applies a differentiable function.

15. A structure for equalizing and demodulating a structured signal, said structured signal having a predetermined number of symbols transmitted in a frame, said structure comprising a first processor and a second processor, each processor being provided for a corresponding one of a first symbol position and a second symbol position in said frame, each of said first and second processors comprising:

a transversal digital filter computing a weighted sum of a predetermined number of samples of said structured signal and providing said weighted sum as an equalized sample; and

a decision device, coupled to receive said equalized sample, for determining from said equalized sample, independently of symbol decisions at other symbol positions in said frame, a symbol transmitted at said symbol position corresponding to said processor in said frame.

16. A structure as in claim 15, wherein each processor further comprises means, coupled to receive one or more of said samples, said symbol and said weighted sum, for updating the weights of said transversal digital filter.

17. A structure as in claim 15, wherein each processor further comprises means, coupled to receive said symbol and said weighted sum, for correcting a phase error in said weighted sum.

18. A structure as in claim 15, wherein, within each processor, each transversal digital filter is one of a plurality of transversal filters in a first level structure of said processor, said processor further comprising:

a plurality of non-linear operators, each non-linear operator operating on an output value of a corresponding one of said plurality of transversal filters to provide a first level equalized output value; and

a transversal digital filter in a second level structure receiving said first level equalized output values for computing a weighted sum of said equalized output values to obtain said equalized sample.

19. A structure as in claim 16, wherein said weights are updated in accordance with an adaptive algorithm.

20. A structure as in claim 18, wherein said weights are updated in accordance with a backpropagation algorithm.

21. A structure as in claim 18, wherein said non-linear operators each comprise means for providing a differentiable function.

22. A method for equalizing and demodulating a structured signal, said structured signal having a predetermined number of symbols transmitted in a frame, said method comprises, for each of a first symbol position and a second symbol position in said frame, the steps of:

computing in a separate transversal digital filter a weighted sum of a predetermined number of samples of said structured signal and providing said weighted sum as an equalized sample; and

determining in a decision device, coupled to receive said equalized sample, independently of symbol decisions at other symbol positions in said frame, a symbol transmitted at said symbol position in said frame.

23. A method as in claim 22, further comprising the step of updating the weights of said transversal digital filter.

24. A method as in claim 22, further comprising the step of correcting a phase error in said weighted sum using said symbol and said weighted sum.

25. A method as in claim 22, wherein said step of computing in a separate transversal digital filter provides a plurality of transversal filters in a first level structure, said method further comprising the steps of:

providing a plurality of non-linear operators, each non-linear operator operating on an output value of a corresponding one of said plurality of transversal filters to provide a first level equalized output value; and

computing in a transversal digital filter in a second level a weighted sum of said first level equalized output values to obtain said equalized sample.

26. A method as in claim 23, wherein said weights are updated in accordance with an adaptive algorithm.

27. A method as in claim 25, wherein said weights are updated in accordance with a backpropagation algorithm.

28. A method as in claim 25, wherein said non-linear operator applies a differentiable function.