Tuned feedforward LMS filter with feedback control

• US 6,996,241 B2
• Filed: 05/10/2004
• Issued: 02/07/2006
• Est. Priority Date: 06/22/2001
• Status: Expired due to Term

• Current Assignee: The Trustees of Dartmounth College (Dartmouth College)
• Original Assignee: The Trustees of Dartmounth College (Dartmouth College)
• Inventors: Streeter, Alexander D., Ray, Laura R.
• Primary Examiner(s): Grier, Laura A.
• Application Number:
  US10/842,714
• Publication Number:
  US 20040264706A1
• Time in Patent Office: 638 Days
• Field of Search: 381/71.11, 381/71.12, 381/71.1, 381/71.6, 381/94.1
• US Class Current: 381/71.11
• CPC Class Codes: G10K 11/17854 the filter being an adaptive filter
• CPC Class Codes: G10K 11/17861 using additional means for damping sound, e.g. using sound absorbing panels
• CPC Class Codes: G10K 11/17881 the reference signal being an acoustic signal, e.g. recorded with a microphone
• CPC Class Codes: H04R 1/1008 Earpieces of the supra-aural or circum-aural type
• CPC Class Codes: H04R 1/1083 Reduction of ambient noise active noise reduction per se G10K11/175; protective devices for the ear, e.g. providing acoustic protection A61F11/06
• CPC Class Codes: H04R 2420/01 Input selection or mixing for amplifiers or loudspeakers
• CPC Class Codes: H04R 3/005 for combining the signals of two or more microphones specially adapted for hearing aids H04R25/407

Abstract

A method to automatically and adaptively tune a leaky, normalized least-mean-square (LNLMS) algorithm so as to maximize the stability and noise reduction performance in feedforward adaptive noise cancellation systems. The automatic tuning method provides for time-varying tuning parameters λ_k and μ_k that are functions of the instantaneous measured acoustic noise signal, weight vector length, and measurement noise variance. The method addresses situations in which signal-to-noise ratio varies substantially due to nonstationary noise fields, affecting stability, convergence, and steady-state noise cancellation performance of LMS algorithms. The method has been embodied in the particular context of active noise cancellation in communication headsets. However, the method is generic, in that it is applicable to a wide range of systems subject to nonstationary, i.e., time-varying, noise fields, including sonar, radar, echo cancellation, and telephony. Further, the hybridization of the disclosed Lyapunov-tuned feedforward LMS filter with a feedback controller as also disclosed herein enhances stability margins, robustness, and further enhances performance.

Claims

1. A method of tuning an adaptive feedforward noise cancellation algorithm, comprising the acts of:

providing a feedback active noise reduction (ANR) circuit, for providing an ANR error signal;

providing a feedforward LMS tuning algorithm including at least first and second time varying parameters wherein said feedforward LMS tuning algorithm includes the formulas;

y_k=W_k^TX_k and

W_k+1=λ

_kW_k+μ

_kX_ke_k

adjusting said at least first and second time varying parameters as a function of instantaneous measured acoustic noise, a weight vector length and measurement noise variance, wherein said time varying parameters include;

$\begin{array}{c}{\mathrm{\mu <br><br>}}_k=<br><br>\frac{{\mathrm{\mu <br><br>}}_o<br><br>{\mathrm{\lambda <br><br>}}_k}{{\left(X_k+Q_k\right)}^T<br><br>\left(X_k+Q_k\right)}\\ {\mathrm{\lambda <br><br>}}_k=<br><br>\frac{{\left(X_k+Q_k\right)}^T<br><br>\left(X_k+Q_k\right)-2<br><br>L<br><br><br><br>{\mathrm{\sigma <br><br>}}_q^2}{{\left(X_k+Q_k\right)}^T<br><br>\left(X_k+Q_k\right)}\end{array}$

 

wherein X_k=X_k+Q_k is a measured reference signal;

Q_k is measurement noise, including electronic noise and quantization noise;

 

σ

_q² is the known or measured variance of the measurement noise;

 

L is the length of the LMS weight vector W_k; and

 

e_k is an error signal which is the net result of both the feedforward method and the feedback circuit.

2. The method of claim 1 wherein the output of the filter y_k is multiplied by a feedforward proportionality constant K_ff to produce a feedforward acoustic noise cancellation signal K_ffy_k and wherein the error signal e_k is acted on by a digital infinite impulse response filter so as to produce a cancellation signal r_k, which is multiplied by a feedback proportionality constant K_fb and the sum of the feedforward and feedback components K_ffy_k+K_fbr_k provides a total noise cancellation signal.

3. The method of claim 2 wherein said ANR error signal of feedback active noise reduction (ANR) circuit and said feedforward LMS tuning algorithm each provide an active noise reduction performance value that is greater than a sum of said ANR circuit and said LMS tuning algorithm.

4. A method of tuning an algorithm for providing noise cancellation, comprising the acts of:

receiving a measured reference signal, the measured reference signal including a measurement noise component having a measurement noise value of known variance; and

generating an acoustic noise cancellation signal according to the formulas;

y_k=W_k^TX_k

W_k+1=λ

_kW_k+μ

_kX_ke_k

 

wherein time varying parameters λ

_k and μ

_k are determined according to the formulas;

$\begin{array}{c}{\mathrm{\mu <br><br>}}_k=<br><br>\frac{{\mathrm{\mu <br><br>}}_o<br><br>{\mathrm{\lambda <br><br>}}_k}{{\left(X_k+Q_k\right)}^T<br><br>\left(X_k+Q_k\right)}\\ {\mathrm{\lambda <br><br>}}_k=<br><br>\frac{{\left(X_k+Q_k\right)}^T<br><br>\left(X_k+Q_k\right)-2<br><br>L<br><br><br><br>{\mathrm{\sigma <br><br>}}_q^2}{{\left(X_k+Q_k\right)}^T<br><br>\left(X_k+Q_k\right)}\end{array}$

 

wherein X_k=X_k+Q_k is a measured reference signal;

 

Q_k is electronic noise and quantization;

 

σ

_q² is a known variance of the measurement noise;

 

L is the length of weight vector W_k; and

 

e_k is an error signal which is the net result of both a feedforward tuning method and a feedback active noise reduction method.

5. A method of tuning a least mean square (LMS) filter comprising the acts of:

providing a feedback active noise reduction (ANR) circuit, for providing an ANR error signal;

formulating a Lyapunov function of a LMS filter weight vector, a reference input signal, a measurement noise on the measured reference input signal, a time varying leakage parameter λ

_k, and a step size parameter μ

_k;

using the resultant Lyapunov function to identify formulas for computing the time varying leakage parameter λ

_k and step size parameter μ

_k that maximize stability and performance of the resultant LMS filter weight vector update equation

W_k+1=λ

_kW_k+μ

_ke_kX_k

 

wherein said time varying parameters determined are ${\mathrm{\mu <br><br>}}_k=\frac{{\mathrm{\mu <br><br>}}_o<br><br>{\mathrm{\lambda <br><br>}}_k}{{\left(X_k+Q_k\right)}^T<br><br><br><br>\left(X_k+Q_k\right)}$${\mathrm{\lambda <br><br>}}_k=\frac{{\left(X_k+Q_k\right)}^T<br><br><br><br>\left(X_k+Q_k\right)-2<br><br>L<br><br><br><br>{\mathrm{\sigma <br><br>}}_q^2}{{\left(X_k+Q_k\right)}^T<br><br><br><br>\left(X_k+Q_k\right)}$

 

wherein X_k=X_k+Q_k is a measured reference signal;

 

Q_k is electronic noise and quantization;

 

σ

_q² is a known variance of the measurement noise;

 

L is the length of weight vector W_k; and

 

e_kis an error signal which is the net result of both the ANR circuit and the LMS filter.

6. A method of tuning an adaptive feedforward noise cancellation algorithm, comprising the acts of:

providing a feedback active noise reduction (ANR) circuit, for providing an ANR error signal;

providing a feedforward LMS tuning algorithm including at least first and second time varying parameters wherein said feedforward LMS tuning algorithm includes the formulas;

y_k=W_k^TX_{k; and}

W_k+1=λ

_kW_k+μ

_kX_ke_k

adjusting said at least first and second time varying parameters as a function of instantaneous measured acoustic noise, a weight vector length and measurement noise variance, wherein said time varying parameters include;

$\begin{array}{c}{\mathrm{\mu <br><br>}}_k=<br><br>\frac{{\mathrm{\mu <br><br>}}_o<br><br>{\mathrm{\lambda <br><br>}}_k}{{\left(X_k+Q_k\right)}^T<br><br>\left(X_k+Q_k\right)}\\ {\mathrm{\lambda <br><br>}}_k=<br><br>\frac{{\left(X_k+Q_k\right)}^T<br><br>\left(X_k+Q_k\right)-2<br><br>L<br><br><br><br>{\mathrm{\sigma <br><br>}}_q^2}{{\left(X_k+Q_k\right)}^T<br><br>\left(X_k+Q_k\right)}\end{array}$

 

wherein X_k=X_k+Q_k is a measured reference signal;

Q_k is measurement noise, including electronic noise and quantization noise;

σ

_q² is the known or measured variance of the measurement noise;

L is the length of the LMS weight vector W_k; and

e_k is an error signal which is the net result of both the feedforward method and the feedback circuit, and further wherein the output of the filter y_k is multiplied by a feedforward proportionality constant K_ff to produce a feedforward acoustic noise cancellation signal K_ffy_k and the error signal e_k is acted on by a digital infinite impulse response filter so as to produce a cancellation signal r_k, which is multiplied by a feedback proportionality constant K_fb and the sum of the feedforward and feedback components K_ffy_k+K_fbr_k provides a total noise cancellation signal.

7. The method of claim 6 wherein said ANR error signal of feedback active noise reduction (ANR) circuit and said feedforward LMS tuning algorithm each provide an active noise reduction performance value that is greater than a sum of said ANR circuit and said LMS tuning algorithm.