Re: [Discuss-gnuradio] False Packets for benchmark TX/RX

[Tom Rondeau]

On Thu, Dec 6, 2012 at 10:44 AM, Yahya Ezzeldin <address@hidden> wrote:

Thank you Tom,

I have been experimenting with the working setting that I have now.

Why do other modulations like bpsk, qpsk fail while gmsk works perfectly ?

That's a very open-ended question and can't be answered. There are so many physical realities to getting signals to work, so it's likely not just one thing (and certainly won't be as easy as it sounds when you read a textbook on the difference in BER performance of modulation schemes).

One thing to keep in mind, though, is that you don't want to use BPSK or QPSK. Use DBPSK and DQPSK instead. There is no mechanism for resolving the phase ambiguity at the receivers. So when running these signals, make sure that they are differentially encoded (use the -v option to get a verbose output of the parameters).

Tom

 

Best Regards,

--

Yahya Ezzeldin

On Thu, Dec 6, 2012 at 5:18 PM, Tom Rondeau <address@hidden> wrote:

On Wed, Dec 5, 2012 at 2:23 AM, Yahya Ezzeldin <address@hidden> wrote:

Thank you Tom and Marcus,

I finally managed to make it work with 100% delivery rate using this configuration.

Transmitter:

*************

Tx-Amplitude = 0.4

Rate = 1M

Samples per Symbol = 4

Receiver:

**********

Rate =1M

Receiver Gain = 10

Samples per Symbol = 4

 

I do have some question that  I don't understand :

---------------------------------------------------------------------

1) The main issue turned out to be the rate. When I increased the rate from 250k to 0.5M some packets started to pass as true. Increasing it to 1M made most of them be accepted as True (pass the CRC32 check if I understand correctly). What is the explanation for this ?

Often, the reason higher rates work better is that the relative frequency offset is lower. You said that you couldn't find a frequency offset that helped with your 250 kbps signal, so I'm not sure if this is really the right answer for you. But that's generally the case.

 

2) Increasing the samples per symbol further improved the reception, why is that ?

Off the top of my head here... GMSK actually introduces ISI, but how that happens is based on the shaping filter used, which we determine by the number of sps. Increasing this is probably making a better shaped signal.

 

3) The two FFT screenshots at 250k and 1M are as follows. How does the difference between them solve the nonlinearity problem, Tom ?

Both of these looked good. I'm not sure it was really nonlinearities, that was just a guess. The original signal you showed us had a strange bulge in left half of the signal. These signals here look symmetric and properly shaped.

 

Tom

 

--

Yahya Ezzeldin