LMS adaptive Viner filter

I. background

The Gini filter parameters are fixed and suitable for Stable Random Signals. Kalman filter parameters are time-varying and suitable for non-Stable Random Signals. However, the two filters can obtain optimal filtering only when the statistical characteristics of signal and noise are a prior known. In practical applications, we often cannot obtain a prior knowledge of the statistical characteristics of signals and noise. In this case, the adaptive filtering technology can achieve excellent filtering performance, so it has good application value. Common adaptive filtering technologies include least mean square (LMS) adaptive filter, Recursive Least Square (RLS) filter, lattice filter, and Infinite Impulse Response (IIR) filter. The application of these adaptive filtering technologies includes Adaptive Noise Cancellation, adaptive spectral line enhancement, and ripple.

Ii. Theoretical knowledge of LMS adaptive question filter

 

 

 

Iii. MATLAB simulation using the above conclusions

First, use the built-in Matlab function to read and save a piece of audio, and then add Gaussian white noise interference to it to filter the processed signal LMS algorithm. And generate and save the corresponding audio files. We can view the effect of LMS Filtering in two ways: audio audition and image observation.

M script code

1% -- 14-10-25 -- % 2% function: LMS adaptive noise offset 3% inputs: 4% outputs: 5 clc; 6 clear; 7 [ref, FS] =wavread('in.wav '); 8 APM = 1; % noise margin 9 firlen = 160; % defined filter length 10 mu = 0.002; % iteration step 11 A = zeros (1, firlen ); % generate a row with a 160 column vector 12 datalen = length (REF); % assign a value to ref signal vector 13 eout = zeros (1, datalen ); % generates a column of datalen vectors with a value of 14 E = zeros (1, datalen); 15 firin = zeros (1, firlen); 16 firout = zeros (1, datalen ); 17 noise = APM * rand (1, datalen); % generate random noise 18 near = ref '+ noise; % add noise 19 for I = firlen: datalen20 firout (I) = A * (near (I-firlen + 1: I) '; % FIR filter, realize Horizontal Filter operation 21 E (I) = REF (I) -firout (I); % returns the error signal 22 A = a + 2 * mu * E (I) * (near (I-firlen + 1: I )); % LMS coefficient updated 23 eout (I) = E (I) * E (I); 24 end25 wavwrite(firout,fs,'firout.wav '); 26 wavwrite(eout,fs,'eout.wav'); 27 '); 28 wavplay (firout, FS); 29 figure (1) 30 subplot (411); 31 plot (REF); 32 title ('original sign'); 33 subplot (412 ); 34 plot (near); 35 title ('voice plus Noise'); 36 subplot (413); 37 plot (Noise); 38 title ('noise '); 39 subplot (414); 40 plot (firout); 41 title ('Adaptive denoising sign ');

Wave. m

Run

4. Download source files

LMS filter1.0.rar
115 online storage package code: 5lbagoebt972
Http://115.com/lb/5lbagoebt972

LMS adaptive Viner filter