[Discuss-gnuradio] 802.11 and filter bank scaling

[Colby Boyer]

Hey All,

With barker code problem mostly solved, I want to look more in depth at higher modulation (CCK) rates, 5.5 and 11 Mbps.  I've read through the IEEE specs and for the most part understand the scheme.  Since the sample rate is not high enough to send/receive enough samples per chip, the same estimation method used for the barker sequence is necessary. Basically, the chipped sequence is smoothed with a sinc and sampled to fit inside a code word window(25 samples).

The downside of that is I don't think efficient methods(FWT) can be used to decode CCK due to this hack.  To properly decode the CCK sequence there needs to be 64 correlation filters with 25 taps to detect the CCK code words.  My worry is that the computation cost would be too high and a computer would not be able to process it.

Does anyone have experience with high rate filter banks in gnuradio?  How did the process scale with filter order and bank size?

I know this question is a bit vague and narrow.

--Colby