modulation

[Aaron_W.]

lavalava,

You are right, unless you have the RF equipment the RFSA drivers don't do you much good. I apologize for the confusion. I was able to download the examples that you posted, but and look through the two different .wav files. I was unable to run the Ideal PSK VI because it was missing to sml_interpolation VIs. I am guessing these are custom VIs that you made. If you attach those I am guessing I will be able to see the difference between the two VIs and the problem will become clearer. 

When you say that you can replicate this in RT are you refering to on-the-fly, or do you mean a Real-Time system? 

When looking at the Real Equalization.VI, I do not see much difference between the w/ or w/o EVM plots, I will have to look into these items a little more to figure out exactly what we would expect to see as a difference between the two.  Could you specify exactly what I should be looking at between these two graphs, so I can look over the functionality with some other individuals. Thanks!

[lavalava]

Hi Aarron,

No, I did not custom made those sml files.  I'm not sure why you didn't have it installed on your computer because it came with my version.  Sorry about that, we must be using different versions.   So here is my spec:

Modulation Toolkit v3.0

NI LV Developer Suite 8.61

In case you still don't have acess to it.  Here, try downloading NISML.zip and extract them to this folder:

C:\Program Files (x86)\National Instruments\LabVIEW 8.6\vi.lib\addons\NISML

Thanks.  When you run Real Equalizer.vi, you'll notice that it would not converge.  It should converge just like the one in Ideal Equalizer.VI.  I suspect the fractional sampler.VI might have something to do with it but I'm not too sure.

PS.  RT means in real time sampling whether it's directly continuously sampling live feed from an audio sound card or playback (I know play back isn't true RT but it's the only way for us two to be on the same page).  And I also updated those two previously posted files too.  Thanks

Message Edited by lavalava on 10-30-2009 04:13 PM

[lavalava]

P.S.  I just made a small change to these files again because I had made some typos in previous posts.  Thanks

Message Edited by lavalava on 10-30-2009 04:27 PM

[Aaron_W.]

lavalava,

I looked over your VIs with one of our modulation toolkit guys and the only main difference between the two VIs was where you were getting the signal from. With the first VI you were generating ideal bits from the MT Generate Bits VI and the other you were getting the data from the .wav file. When using the generate bits the equalization works well but that could be due to the fact that Generate Bits produces an ideal signal, or relatively considering you can manipulate the signal to be less clean. 

Overall, we were not completely sure why the filter was not cleaning up the signal the same as the MT Generate Bits signal, but when looking at the post-filtered signal we did notice some filtering that was taking place, just not as much as what we saw with the ideal case. Is the signal you are generating and filtering still too noisy to work properly for your application? 

[lavalava]

Hi Aaron,

Actually, the Ideal Equalizer.VI was mixed with AGWN so it was not exactly ideal.  Even when the noises was increased so you could still see the equalizer vi converging.  Unfortunately, I can't seem to replicate this at all with my captured signal.  Yes, that was too noisy because sometimes my clients will increase the power level and it will distort.  I suspect that the modulation guys might have not tested the equalizer using real time captured signal.  He must've only tested it in ideal cases when he coded.  Have you guys tried using your RF hardware to upconvert, mixing with noises, and downconverting it and see if it equalizes while writing these codes?  I suspect it was never completed.  If you have tried those Agilent Vector Signal Analyzer such as model 89441A, you'll notice Agilent people coded their equalizer perfectly.  You specify the convergence rate and it adapts immediately.  The reason why I think the toolkit guys have never tested their code with actually RF hardware is when I looked at the supplied example, I noticed they fed their training bits directly from the bit generator (before modulation and upconvertion) then into the equalizer.   In real time, you just can't make such assumption because you don't know what is being generated.  You can only assume LMS algorithm to do the work.  And you should be able to derive the convergence rate from LMS.  Yeah, I have a few more projects involving RF (after this modem projects) so it would be nice if you guys can convince those toolkit guys to look into this problem and possibly providing updates?  That way, I can convince my boss to let us buy those hardwares from you guys.  With the whole NI hardware/software suite it will allow us to run our test in real time whereas with these Agilent instruments,  you can't, you can only do it in steady state.  Which is a major drawback for us.  And it will also benefit you guys too because your goal is to sell a full flow of products (hardware and software), not just software limiting to ideal academia stuffs.

Message Edited by lavalava on 11-03-2009 12:18 PM

[Aaron_W.]

lavalava,

I can ask them to take a closer look at it and see what they have to say. However this might take some time as those guys are very busy from day to day. I can give you an update as soon as I hear back from them. Thanks!

[lavalava]

Say Aaron.  Since this will take a while and these adaptive equalizers involve using adaptive filter toolkits.  Would you happen to have a trail version of the adaptive filter toolkits avaliable for trying?  I just want to know if I can give it a shot, I doubt I can do much but it's probably a good idea to try it and see if it does work then we'll buy it.  Could you please post it on your FTP site and if you need to password encrypt it, could you please email it to me?  Thanks.

[Aaron_W.]

lavalava,

Are you referring to the RFSA and RFSG drivers, or one of our other toolkits? If you are looking for the RFSA and RFSG drivers they are already online and are the full versions of the drivers. 

Drivers and Update Search

If you search for RFSA and then RFSG you would be able to find those drivers. If you were referring to the WLAN, WiMAX, or GPS toolkits there are no trial software for those toolkits. 

[lavalava]

The equalizer VI involves adaptive filter toolkit, this is what I would like to tryout.  It's addon that was not avaliable when I installed the developer edition.

[Aaron_W.]

lavalava,

We apparently do have a eval version of the Adaptive Filter Toolkit, but you must have either the full version of LabVIEW 2009 or the eval version of it. There is also a link for the eval version of 2009 on the page. 

NI LabVIEW Modules and Toolkits for Image and Signal Processing