OFDM Pilot Detection

Started by KanaBaba 1 month ago7 replieslatest reply 2 weeks ago102 views

I have captured OFDM I-Q data at IF. I have down-converted to it to baseband and then simply performed autocorrleation. Based on autocorrelation-based techniques I was able to estimate following parameters:

1. Useful data duration

2. Cyclic prefix duration

3. Total symbol duration

4. Number of subcarriers used

5. FFT size

Now I am trying to find/estimate the locations of active/data subcarriers and pilot tones. Any thoughts on how to approach the problem?

FYI, although most of the time the waveform follows IEEE standards such a Wi-Fi, WiMAX or DVB, but not all the times. I am interested to find approaches for both standard specific case and blind generic case (if possible). 

Any suggestions will be appreciated. Thanks.

[ - ]
Reply by Detlef _AOctober 30, 2020

Hi, identify a FFT bin with constant amplitude and phase. This is a pilot tone.



[ - ]
Reply by Mannai_MuraliNovember 8, 2020

For pilot tones phase will not be constant from ofdm symbol to next symbol.Also in pilot group phase rotation of pilots due to ADC sampling clock offset will be there.So even in one symbol from pilo to pilots phase will not be constrant.

[ - ]
Reply by Mannai_MuraliOctober 31, 2020

Assume like DVB-T continuous pilots are available every Symbol.Let 'M' maximum spacing between pilots.Let 'N' minimum spacing between pilots.Let 'P' be starting non Null tone.

Starting at 'P' take FFT of tone P,P+M,P+2M etc for symbol 'l'.Multiply by complex conjucate of symbol l-1.Find the argument of product at P,P+M,P+2*M etc for 2 sets.Fit a straight line with tone index and the angle for first half set.Do for same fro second half sets seperated by 'M.If they are pilos slope and intercept will be same for both the sets.Also fit lines for the set P+M,P+3*M etc.Verify if the slope and intercepts are same for both the sets.

Repeat for P+1,P+1+M, etc till P+M-1,P+2*M-1 etc of 2 sets.

Reduce M by one and do the same.Repeat till N

For correct pilot sets both first set and second set will be same.For data it will be different.This search needs a lot of computation.But you may try to reduce it.

If you rely on single tone amlitude and phase noise my mislead it.

[ - ]
Reply by KanaBabaNovember 13, 2020

Sounds promising! Do you have any references like journals, technical report or thesis for this algorithm?

[ - ]
Reply by Mannai_MuraliNovember 13, 2020

There are 3 IEEE papers I can give you as reference.Two by Heinrich Meyr and another by different authors.If you give your email I can mail you those papers.Generally for synchronization problems I suggest you read 'Digital Communication Receivers ' By Heinrich Meyr.That is a very good book.All the mathematical baground covered.For further detail you may contact me


[ - ]
Reply by Mannai_MuraliNovember 8, 2020

Are you interested to work in blind algorithms.If you are working whatever knowledge I have I can share with you.

[ - ]
Reply by KanaBabaNovember 13, 2020

Well, even though I will start working with few common IEEE standards, I want to eventually implement practical algorithms for generic pilot location detection/estimation etc. Eventually blind algorithm will be helpful, if it is computationally less complex.