Adaptive recursive least mean square error filter

• US 4,038,536 A
• Filed: 03/29/1976
• Issued: 07/26/1977
• Est. Priority Date: 03/29/1976
• Status: Expired due to Term

• Current Assignee: Rockwell International Corporation
• Original Assignee: Rockwell International Corporation
• Inventors: Feintuch, Paul L.
• Primary Examiner(s): Smith, Jerry
• Application Number:
  US05/671,889
• Time in Patent Office: 484 Days
• Field of Search: 235/152, 235/156, 325/38 B, 325/42, 328/167, 333/18, 333/28 R, 333/70 T, 179/170.2
• US Class Current: 708/322
• CPC Class Codes: H03H 21/0043 Adaptive algorithms

Abstract

An adaptive recursive filter is disclosed which, in a preferred embodiment, comprises first and second adaptive transversal filters selectively coupled together to minimize the mean square error of the output data of the recursive filter based upon observations of input data to the recursive filter. Each transversal filter includes a tapped delay line with a variable weight on each tap. The output data of the recursive filter is developed by combining the outputs of the first and second transversal filters. The input data is applied to the first transversal filter, while the output data is applied to the second transversal filter. The output data is also combined with a reference signal to provide an error signal. A function of that error signal is utilized to update the weights of all of the taps in both transversal filters in order to cause the weights to automatically adapt themselves to minimize the mean square error of the output data of the recursive filter.

Claims

1. An adaptive filter comprising:

first means for receiving a first input signal;

a first adaptive transversal fiter, coupled to said first means, being responsive to the first input signal for developing a first filter signal;

a second adaptive transversal filter responsive to an output signal of the adaptive recursive filter for developing a second filter signal;

second means for combining the first and second filter signals to form said output signal;

third means for receiving the second input signal;

fourth means coupled to said second and third means for combining the second input signal with the output signal to generate an error signal; and

fifth means responsive to the error signal for causing said first and second transversal filters to adaptively modify the first and second filter signals to minimize the mean square error in the output signal.

2. The adaptive recursive filter of claim 1 wherein said means includes:

sixth and seventh means respectively responsive to first and second scalars and commonly responsive to the error signal for respectively developing first and second functions of the error signal, said first and second transversal filters being respectively responsive to said first and second functions of the error signal for automatically adapting themselves to respectively modify the first and second filter signals to minimize the mean square error in the output signal.

3. The adaptive recursive filter of claim 1 wherein said fifth means includes:

sixth means responsive to a preselected scalar and to the error signal for developing one function of the error signal, said first and second transversal filters being responsive to the function of the error signal for automatically adapting themselves to modify the first and second filter signals to minimize the mean square error in the output signals.

4. The adaptive recursive filter of claim 1 wherein:

said fifth means develops a function of the error signal equal to the error signal, said first and second transversal filters being responsive to said function of the error signal for automatically adapting themselves to modify the first and second filter signals to minimize the mean square error in the output signal.

5. The adaptive recursive filter of claim 1 wherein each of said first and second transversal filters includes:

a delay line having a plurality of taps for delaying the associated one of the input and output data signals applied thereto so that a plurality of differently delayed data signals is respectively developed at the plurality of taps;

a summing circuit for summing a plurality of updated terms to develop the associated one of the first and second signal; and

a plurality of sixth means respectively coupled to said plurality of taps, said plurality of sixth means being responsive to a function of the error signal and also being respectively responsive to the plurality of differently delayed data signals for developing the plurality of updated terms.

6. The adaptive recursive filter of claim 5 wherein said of said sixth means includes:

a first multiplier responsive to the function of the error signal and to an associated one of the differently delay data signals for developing a first product signal;

means for summing the first product signal with a presently updated signal component to develop a summed signal;

means for applying the summed signal to develop the next updated signal component; and

a second multiplier responsive to the presently updated signal component and to the associated one of the differently delayed data signals for developing an associated said updated term.

7. A machine method for rendering a recursive filter adaptive comprising the steps of:

combining the filter output signal with a first input signal to develop an error signal; and

utilizing the filter output signal, a second input signal and the error signal to enable the filter to automaticaly adapt itself to minimize the mean square error in the filter output signal.

8. A machine method for enabling a recursive filter to automatically adapt comprising the steps of:

combining the filter output signal with a first input signal to develop an error signal;

generating a function of the error signal; and

enabling the filter to minimize the mean square error in the output signal in response to a second input signal, the output signal and to the function of the error signal.

9. A machine method for causing a recursive filter to automatically adapt comprising the steps of:

combining a first input signal with the output signal from the recursive filter to develop an error signal;

scaling the error signal; and

applying a second input signal, the output signal and the scaled error signal to the filter to enable the filter to automatically adapt itself to minimize the mean square error in the ouput signal.

10. A machine method for enabling a recursive filter to automatically adapt comprising the steps of:

combining the output signal of the filter with a first input signal to develop an error signal;

developing at last one function signal from the error signal;

utilizing a second input signal, the output signal and said at least one function signal to cause the recursive filter to automatically adapt itself to minimize the mean square error in the output signal.

11. The machine method of claim 10 wherein:

said developing step consists of generating first and second function signals from the error signal; and

said utilizing step consists of using both the first and second function signals to enable the recursive filter to automatically adapt itself to minimize the mean square error in the output signal.

12. The machine method of claim 10 wherein:

said developing step consists of generating one function

said developing step consists of generating one function signal from the error signal; and

said utilizing step consists of using the one function signal to enable the recursive filter to automatically adapt itself to minimize the mean square error in the output signal.

13. The machine method of claim 12 wherein:

said generating step consists of making the one function signal equal to the error signal.

14. A machine method for making a recursive filter adaptive comprising the steps of:

providing a first input signal to a first transversal filter;

combining the output of the first transversal filter with the output of a second transversal filter to develop the output signal to the recursive filter;

feeding back the output signal to the second transversal filter;

enabling a second input signal to be combined with the output signal to develop an error signal;

utilizing the error signal to develop at least one function signal; and

applying said at least one function signal to the first and second transversal filter to enable the first and second transversal filters to automatically adapt themselves to minimize the mean square error in the output signal.

15. The machine method of claim 14 wherein:

said utilizing step consists of generating first and second function signals from the error signal; and

said applying step consists of sending the first and second function signals to the first and second transversal filters respectively to enable the first and second transversal filters to automatically adapt themselves to minimize the mean square error of the output signal.

16. The machine method of claim 14 wherein:

said utilizing step consists of generating only one function signal from the error signal; and

said applying step consists of using the one function signal to enable both of the first and second transversal filters to automatically adapt themselves to minimize the mean square error in the output signal.

17. The method of claim 16 wherein:

the generating step consists of making the function signal equal to the error signal.