Hello all,

I found some paper about LDPC said they approach Shannon limit to
within 0.6-1dB. But what BER are they using, 1e-5 or 1e-6 or other?

And what's the relationship with Shannon limit and error floor?

Regards,
Davy

>>>>> "Davy" == Davy  <zhushenli@gmail.com> writes:

    Davy> I found some paper about LDPC said they approach Shannon limit to
    Davy> within 0.6-1dB. But what BER are they using, 1e-5 or 1e-6 or other?

I've always interpreted this to mean that as the SNR gets larger, we
get to within 0.6-1dB of the Shannon Limit.  Or as the BER gets
smaller and smaller we eventually get to 0.6 dB of the limit.

    Davy> And what's the relationship with Shannon limit and error floor?

None, I think.

Ray

On 19 Apr 2005 07:32:26 -0700, "Davy" <zhushenli@gmail.com> wrote:

>Hello all,
>
>I found some paper about LDPC said they approach Shannon limit to
>within 0.6-1dB. But what BER are they using, 1e-5 or 1e-6 or other?
>
>And what's the relationship with Shannon limit and error floor?
>
>Regards,
>Davy

Capacity approaching codes tend to have steep waterfall curves, so
sometimes it doesn't really matter what BER is used...they drop so
fast that you can measure the difference almost anywhere.

But it's really a good question and there is no solid answer.
Different applications tend to consider different BERs as the most
important reference point, so you won't really know the answer to this
unless it's stated specifically.

There is no relationship between the Shannon limit and the so-called
error floor exhibited by capacity approaching codes.  


Eric Jacobsen
Minister of Algorithms, Intel Corp.
My opinions may not be Intel's opinions.
http://www.ericjacobsen.org