Re: [Discuss-gnuradio] GLFSR transmission with BPSK Mod

[Tom Rondeau]

On Mon, May 28, 2012 at 1:39 PM, Nazmul Islam
<address@hidden> wrote:
> Hello,
>
> I am trying to transmit a GLFSR sequence using bpsk (non-differential
> modulation). I am using gnuradio-companion to construct the blocks. My
> transmit path is is simply the following: GLFSR source -> BPSK Mod
> (Non-differential) -> USRP Sink. I am currently simulating the output of the
> GLFSR source and BPSK Mod source before connecting the blocks to USRP. I
> have attached the screenshot of the blocks and the corresponding grc file
> with the email.
>
> I have two questions:
>
> 1. How can I control the bit transmission rate of the GLFSR source?  I want
> to transmit the GLFSR source at 3.5 Mbits/sec (500 kilo sequence per second
> with degree 3). The GLFSR block does not have any option for transmission
> rate. The UHD USRP sink has a option of sample rate. But I don't see how
> that will control the bit transmission rate of the GLFSR source.

Hi Nazmul,
In the computer, sample rate is meaningless. There is no concept of
"samples/seconds" just operations per second. Without a rate-limiting
device, the computer will process samples as quickly as possible.

The USRP provides the real rate that samples will be running at since
it clocks everything according the its clock and the
interpolation/decimation rate you use. This then provides a rate
limiting to the GNU Radio scheduler to keep the computer processing
and generating samples at an appropriate rate. Simply put, if the
USRP's UHD source block is processing samples, the scheduler will wait
until it has room to accept another chunk of samples.

> 2. I can't understand the relation between the output of the GLFSR source
> with that of the PSK Mod. The GLFSR source is producing an expected output:
> 1110110..(repeat). But the BPSK Mod output is producing strange output in
> the following two ways:
>
> a) The length of the PSK Mod output is almost 8 times with respect to that
> of the GLFSR output. I am using 2 samples/symbol. Why is the PSK Mod
> creating so many output?

The PSK Mod block thinks you are passing it bytes, which is then
converts to bits. So every output of the GLFSR block is treated as 8
individual bits by the modulator.

When running this on my machine, I get a 128x expansion factor between
the two files. You have 8x for the byte-to-bit, 8x going from char to
complex float, and 2x because sps=2.

> b) The PSK Mod is producing output with different complex values, e.g., 1.2,
> 0.95, 1.05, -0.2, -1.2, etc. I have 2 samples/symbol in the PSK mod block.
> Since the PSK Mod output length is not twice the output length of the GLSR
> block, how can I find which two samples correspond to which GLFSR bit?

The PSK modulator will apply an RRC filter to the samples. An RRC
filter is not a Nyquist filter, so it produces ISI, which is what is
causing your values. If you pass this output through another
equivalent RRC filter, you should see points right at +1 and -1.

> Feedback on any of these questions and the overall GRC block will be highly
> appreciated.
>
> Thanks,
>
> Nazmul

Hope that clears things up.

Tom