| LTPDA Toolbox | contents | ![]() |
Whitening filter We will now use a whitening filter to whiten a data stream. The test data for this you will find in your data packet in 'topic2'. We start by loading the xml file:
>> a = ao('topic2/white.xml');
a is now an AO containing coloured noise. Let's have a look at this times series using iplot.
>> iplot(a);
We have to define the parameter list for the whitening tool:
>> pl = plist(...
'model', [], ...
'MaxIter', 30, ...
'PoleType', 1, ...
'MinOrder', 2, ...
'MaxOrder', 9, ...
'Weights', 2, ...
'Plot', false,...
'Disp', false,...
'RMSEVar', 5,...
'FitTolerance', 0.6); % tolerance on fit residuals spectral flatness
No we can call the whitening function 'whiten1D' with our AO, a and the parameter list pl:
>> aw = whiten1D(a,pl);
To compare the whitened data with the coloured noise we compute the power spectrum (details see Power spectral density estimation):
>> awxx = aw.psd; >> axx = a.psd;
and finally plot our result in the frequency domain:
The whitened data (awxx) compared to the coloured noise that was our input (axx).
>> iplot(axx, awxx);
|
Remove trends from a time-series AO | Selecting data from an AO | ![]() |
©LTP Team