Adaptive equalizer, acoustic echo canceller device, and active noise control device

Patent metadata

Field	Value
Publication number	US-9830900-B2
Application number	US-201213984136-A
Country	US
Kind code	B2
Filing date	Mar 29, 2012
Priority date	May 10, 2011
Publication date	Nov 28, 2017
Grant date	Nov 28, 2017

Abstract

A variable update step size is determined in proportion to a magnitude ratio or magnitude difference between a first residual signal and a second residual signal. The first residual signal is obtained by using adaptive filter coefficient sequence, where the adaptive filter coefficient sequence has been obtained in previous operations of the adaptive equalizer. The second residual signal is obtained by using a prior update adaptive filter coefficient sequence, where the prior update adaptive filter coefficient sequence is obtained by performing a coefficient update with an arbitrary prior update step size on the adaptive filter coefficient sequence having been obtained in previous operations of the adaptive equalizer.

First claim

The invention claimed is:

1. An adaptive equalizer that generates an output signal by performing a filtering process on an input signal using an adaptive filter coefficient sequence having been subjected to a coefficient update process using a variable update step size,
wherein the variable update step size is determined based on a magnitude ratio between or a magnitude difference between:
a first residual signal obtained by using a first estimated response signal which is obtained through a filtering process on the input signal using a filter coefficient sequence, the filter coefficient sequence having been obtained in previous operations of the adaptive equalizer; and
a second residual signal obtained by using a second estimated response signal which is obtained through a filtering process on the input signal using a prior update adaptive filter coefficient sequence, the prior update adaptive filter coefficient sequence being obtained by updating the filter coefficient sequence used for obtaining the first estimated response signal with a prior update step size.

2. The adaptive equalizer according to claim 1 , wherein
the second residual signal is a plurality of residual signals obtained by using a plurality of prior update adaptive filter coefficient sequences, the plurality of prior update adaptive filter coefficient sequences being obtained by performing coefficient updates with different prior update step sizes on the adaptive filter coefficient sequence having been obtained in previous operations of the adaptive equalizer, and
the variable update step size is determined based on a magnitude ratio or magnitude difference between at least two residual signals selected from the first residual signal and the plurality of second residual signals.

3. The adaptive equalizer according to claim 1 , wherein the determination of the variable update step size is performed by using the prior update step size as additional information.

4. The adaptive equalizer according to claim 1 , wherein
the input signal is a signal to be input to an unknown system being a target of learning identification, and
the determination of the variable update step size is performed by using the input signal as additional information.

5. The adaptive equalizer according to claim 1 , wherein the variable update step size is determined to be any of values in a predetermined range, the predetermined range being determined based on the first residual signal and the second residual signal.

6. The adaptive equalizer according to claim 3 , wherein the variable update step size is determined to be any of values in a predetermined range, the predetermined range being determined based on the first residual signal, the second residual signal and the prior update step size.

7. The adaptive equalizer according to claim 1 , wherein
the adaptive equalizer calculates variances or average power with respect to each of the first residual signal and/or the second residual signal in a predetermined signal block length basis through a block adaptive algorithm that performs the coefficient update on the predetermined signal block length basis, and
the determination of the variable update step size is performed by using the variances or average power as additional information.

8. The adaptive equalizer according to claim 1 , wherein the prior update step size is a variable update step size obtained in previous operations.

9. The adaptive equalizer according to claim 3 , wherein the variable update step size is determined for each frequency component by using the first residual signal and the second residual signal having been subjected to a frequency transform, and the prior update step size.

10. The adaptive equalizer according to claim 1 , comprising:
a subtraction filter that generates a first estimated response signal by performing a filtering process on the input signal using the adaptive filter coefficient sequence obtained in previous operations, and generates the first residual signal by subtracting the generated first estimated response signal from an observed response signal obtained by adding a disturbance to a response signal of the unknown system to the input signal;
an update filter that calculates a prior update filter coefficient sequence by performing, using the first residual signal generated by the subtraction filter and the prior update step size, a coefficient update on the adaptive filter coefficient sequence obtained in previous operations of the adaptive equalizer, generates a second estimated response signal by performing a filtering process on the input signal using the calculated prior update filter coefficient sequence, and generates the second residual signal by subtracting the second estimated response signal from the observed response signal; and
a step size controller that determines the variable update step size from at least the first residual signal generated by the subtraction filter, the second residual signal generated by the update filter, and the prior update step size.

11. The adaptive equalizer according to claim 10 , wherein
the update filter generates a plurality of second estimated response signals by performing a filtering process on the input signal using a plurality of prior update filter coefficient sequences calculated using a plurality of prior update step sizes, and generates a plurality of second residual signals by subtracting the plurality of second estimated response signals from the observed response signal, and
the step size controller determines the variable update step size from at least two residual signals selected from the first residual signal and the plurality of second residual signals and the prior update step size.

12. The adaptive equalizer according to claim 10 , further comprising a coefficient update filter that performs a coefficient update on the adaptive filter coefficient sequence by using the variable update step size.

13. The adaptive equalizer according to claim 1 , comprising:
a first subtraction filter that generates a first estimated response signal by performing a filtering process on the input signal using the adaptive filter coefficient sequence obtained in previous operations, and generates the first residual signal by subtracting the generated first estimated response signal from an observed response signal obtained by adding a disturbance to a response signal of the unknown system to the input signal;
a second subtraction filter that generates a second estimated response signal by performing a filtering process on the input signal using a prior update filter coefficient sequence obtained in previous operations, and generates the second residual signal by subtracting the generated second estimated response signal from the observed response signal;
a step size controller that determines the variable update step size from at least the first residual signal, the second residual signal, and the prior update step size;
a coefficient update filter that updates the adaptive filter coefficient sequence based on an amount of coefficient update obtained in previous operations and the variable update step size determined by the step size controller, generates an output estimated response signal by performing a filtering process on the input signal using the updated adaptive filter coefficient sequence, and obtains an output residual signal by subtracting the generated output estimated response signal from the observed response signal;
an coefficient update amount calculator that calculates the amount of coefficient update from the input signal and the output residual signal obtained by the coefficient update filter; and

Assignees

• Yano Atsuyoshi
• Awano Tomoharu
• Mitsubishi Electric Corp

Inventors

• Yano Atsuyoshi
• Awano Tomoharu

Classifications

• G10K11/17854

  the filter being an adaptive filter · CPC title

• H03G5/165

  Equalizers; Volume or gain control in limited frequency bands · CPC title

• H04B3/235

  combined with adaptive equaliser · CPC title

• H04R3/02

  for preventing acoustic reaction {, i.e. acoustic oscillatory feedback (specially adapted for hearing aids H04R25/453)} · CPC title

• G10K2210/3014

  Adaptive noise equalizers [ANE], i.e. where part of the unwanted sound is retained · CPC title