On Sun, Mar 29, 2009 at 11:17 AM, Andrew Buck <address@hidden> wrote:
Hi,
I'm trying to use a USRP to decode a CPFSK signal at 980MHz and 1Mb/s. The
decoder successfully detects the sync pattern, but only about once a minute
when the pattern is transmitted once a second. And when the decoder does
detect the sync pattern, the data following the pattern is incorrect. I
think the issue may be that the sample rate is too low. When I record the
raw signal data to my hard drive, graph it with Matlab, and zoom way in,
it's very jagged and I can't see the uniform changes in frequency I would
expect with CPFSK. Clearly there is some data there since my code detects
the sync pattern once in awhile, but I feel like I may be right on the edge
of the minimum sampling rate. When I tell the USRP to sample at 4MHz, I do
get periodic overflow errors and when I sample at 2MHz I don't, but
obviously that's only 2 samples per bit.
What I don't understand is how if the USRP is capable of recording an HDTV
signal, which has a bitrate well about 1MB/s, why I can't seem to get a
sample rate fast enough for my slower signal. Can anyone tell me what I'm
doing wrong? Maybe it's not sample rate, but something else? If so, why is
the data so jagged?
I am pretty sure the ATSC processing is done after the fact and
definitely not in real-time.
You may want to try something similar. Capture samples to a file
first and process the data. Once you're happy with the result,
profile the code to see where the bottlenecks are in the processing
chain.
The data being "jagged" might just be due to the low sample/symbol rate.
Hope this helps, and good luck!
Brian