Forums

question on convolutional interleaver

Started by philgo August 3, 2006
I am using convolutional interleaver for DVB-T.

DVB-T uses shortened (204, 188) Reed Solomon code. The output is sent
to a convolutional interleaver with depth 12 and slope 17. I tested it
in Matlab and am using the function 'convintrlv'.
After the convolutional interleaver, the interleaved symbols contain
many zero symbols because of the delay registers.

my question is, are we gonna send those zero symbols? if so, aren't
they some overhead that we can avoid?

thanks.

Furthermore, what decides the slope?
It seems to be 17*12=204. Is this a coincidence? If not, why shall we
choose depth to be 17?

can we choose 2 or 3 and still have the same performance? I am asking
this because I think interleaver depth (12) should determines the final
performance, not the slope.

thanks.


philgo wrote:
> I am using convolutional interleaver for DVB-T. > > DVB-T uses shortened (204, 188) Reed Solomon code. The output is sent > to a convolutional interleaver with depth 12 and slope 17. I tested it > in Matlab and am using the function 'convintrlv'. > After the convolutional interleaver, the interleaved symbols contain > many zero symbols because of the delay registers. > > my question is, are we gonna send those zero symbols? if so, aren't > they some overhead that we can avoid? > > thanks.