Adaptive Noise Cancellation Using RLS Adaptive Filtering

This example shows how to use an RLS filter to extract useful information from a noisy signal. The information bearing signal is a sine wave that is corrupted by additive white gaussian noise.

The adaptive noise cancellation system assumes the use of two microphones. A primary microphone picks up the noisy input signal, while a secondary microphone receives noise that is uncorrelated to the information bearing signal, but is correlated to the noise picked up by the primary microphone.

Note: This example is equivalent to the Simulink® model 'rlsdemo' provided.

The model illustrates the ability of the Adaptive RLS filter to extract useful information from a noisy signal.

priv_drawrlsdemo
axis off

The information bearing signal is a sine wave of 0.055 cycles/sample.

signal = sin(2*pi*0.055*(0:1000-1)');
signalSource = dsp.SignalSource(signal,'SamplesPerFrame',100,...
    'SignalEndAction','Cyclic repetition');

plot(0:199,signal(1:200));
grid; axis([0 200 -2 2]);
title('The information bearing signal');

The noise picked up by the secondary microphone is the input for the RLS adaptive filter. The noise that corrupts the sine wave is a lowpass filtered version of (correlated to) this noise. The sum of the filtered noise and the information bearing signal is the desired signal for the adaptive filter.

nvar  = 1.0;                  % Noise variance
noise = randn(1000,1)*nvar;   % White noise
noiseSource = dsp.SignalSource(noise,'SamplesPerFrame',100,...
    'SignalEndAction','Cyclic repetition');

plot(0:999,noise);
title('Noise picked up by the secondary microphone');
grid; axis([0 1000 -4 4]);

The noise corrupting the information bearing signal is a filtered version of 'noise'. Initialize the filter that operates on the noise.

lp = dsp.FIRFilter('Numerator',fir1(31,0.5));% Low pass FIR filter

Set and initialize RLS adaptive filter parameters and values:

M      = 32;                 % Filter order
delta  = 0.1;                % Initial input covariance estimate
P0     = (1/delta)*eye(M,M); % Initial setting for the P matrix
rlsfilt = dsp.RLSFilter(M,'InitialInverseCovariance',P0);

Running the RLS adaptive filter for 1000 iterations. As the adaptive filter converges, the filtered noise should be completely subtracted from the "signal + noise". Also the error, 'e', should contain only the original signal.

scope = timescope('TimeSpan',1000,'YLimits',[-2,2], ...
	                  'TimeSpanOverrunAction','Scroll');
for k = 1:10
    n = noiseSource(); % Noise
    s = signalSource();
    d = lp(n) + s;
    [y,e]  = rlsfilt(n,d);
    scope([s,e]);
end
release(scope);

The plot shows the convergence of the adaptive filter response to the response of the FIR filter.

H  = abs(freqz(rlsfilt.Coefficients,1,64));
H1 = abs(freqz(lp.Numerator,1,64));

wf = linspace(0,1,64);

plot(wf,H,wf,H1);
xlabel('Normalized Frequency  (\times\pi rad/sample)');
ylabel('Magnitude');
legend('Adaptive Filter Response','Required Filter Response');
grid;
axis([0 1 0 2]);

References

[1] S.Haykin, "Adaptive Filter Theory", 3rd Edition, Prentice Hall, N.J., 1996.

×

예제 열기

이 예제의 수정된 버전이 있습니다. 사용자가 편집한 내용을 반영하여 이 예제를 여시겠습니까?

아니요, 수정된 버전을 덮어씁니다. 예