TFA Adaptive Transform

[M.Siddeq]

Hello

I have  questions regarding the implementation of the TFA Adaptive Transform VI, the Advanced Signal Processing Toolkit help stated that " The TFA Adaptive Transform provides an implementation of the adaptive transform that is more efficient and accurate than the matching pursuit method. This implementation uses the matching pursuit method with a small dictionary size as a coarse estimation step and then follows with a refinement step to achieve an accurate estimation."

http://zone.ni.com/reference/en-XX/help/371419D-01/lvasptconcepts/tfa_adapt_trans_exp/

My questions are: What is the more efficient implementation algorithm? what are the principles of the  refinement step that follows the transform?

Unlike most of LabVIEW VIs and algorithms that I know, there is nothing in the documentation about the mentioned questions and there are no reference material.

[Bob_Schor]

It is possible that this is a proprietary algorithm that NI created -- I certainly haven't seen it described.  You might generate some known test signals and compare it with the other transforms (Matching Pursuit?) to judge for yourself how they compare.

Maybe someone from NI will notice this query and get back to you ...

Bob Schor

[M.Siddeq]

Thank you Bob-Schor

Yes it is mentioned that " The adaptive transform and the adaptive expansion are unique to the  LabVIEW Time Frequency Analysis Tools.", yet; I beieve it could be also included in the documentation among the other applications.

For the performance evaluation, yes I'm gonna do it, I have to decide whether I build a LabVIEW code for the matching pursuit algorithm or I use a code and functions of other programing environment. Still, the results of the performance evaluation need to be justified, which is not easy in the absence of algorithm description.

[Bob_Schor]

I agree with you.  I hope NI sees this post ...

Bob Schor

[M.Siddeq]

Thank you  🙂

Spoiler

[Johnnyg17]

It seems the details of the algorithm used in the TFA Adaptive Transform VI are limited to that explained in the Advanced Signal Processing Toolkit help. If you open up the TFA Adaptive Transform VI you can see it makes a Call Library Function Node to the function SptReAdaptTransH() in the library LVASPT_TFA.dll; this indicates that the actual algorithm used is proprietary. I believe other NI toolkits with proprietary functionality have a similar structure, whereas open source ones can be stepped through their entire VI/subVI structure. 

[M.Siddeq]

Hello Johnny G17
Thank you
I don't think the dll indicates a proprietary algorithm, for example the "AFT create FIR LMS vi" uses dll although the algorithm is not a proprietary of LabVIEW, moreover; the LabVIEW implementation of FIR LMS filter is documented in NI website.
I believe that even it is a proprietary, it could/should be reported somewhere, where the copyright is preserved and granted (patent for example).