Re: [Discuss-gnuradio] Questions about bbn_80211b_rx.py

[Alex Teo]

> I was told that this code can only work with a BBN code running as the 
> transmitter. Not a commercial AP.
> Did anyone use the USRPs to successfully decode the beacons from APs?

I worked on the BBN code before, and I was able to decode Probe
Requests from mobile phones as well as the BBN transmitter.

> Actually, I am going to use the code to receive the 802.11b beacons 
> transmitted by the commercial APs. However, I always get a bad CRC.

I think I found a bug in bbn_slicer_cc.cc that degrades its
performance but I am not too sure how to report the bug.

Anyway, there is a buffer that collects the sum of the magnitude of
each sample in a symbol, in order to determine the offset error for
symbol synchronization. However, the slicer only resets it to zero
during initialization, but it should also be reset every block size.
My fix is as follows:

Index: bbn_slicer_cc.cc
===================================================================
--- bbn_slicer_cc.cc (revision 1066)
+++ bbn_slicer_cc.cc (working copy)
@@ -126,6 +126,10 @@
           max_idx = i;
         }
       }
+
+      for(i=0; i<d_samples_per_symbol; ++i) {
+        d_sums[i] = 0;
+      }

       delta = (float)max_idx - d_f_offset;
 #if BBN_SLICER_DEBUG

After i applied this fix, the packet error rates seem to be reduced
significantly.

Regards,
Alex