On Wed, Mar 6, 2013 at 2:48 PM, David Rowe <da...@rowetel.com> wrote: > I've been talking to some Turbo Coding and LPDC guys and I now think > they would be worth testing, for example with block sizes of 256 bits. > While these codes are optimal at blocks with several thousand bits (a > few seconds delay of Codec 2) they might still be useful at block > sizes/delays of a few 100ms. > > We could also develop a "weak signal" mode which has all the delay we > need to interleave over time, use the best codes available, and just put > up with 5-10 seconds delay and no SSB-type break in experience. If it > out-performs SSB (say error free, noise free voice at SNR=-3dB) then it > might be useful in some applications.
Please don't use turbo-codes there are patents all over that space, though some of them are available for HAM ONLY usage. LDPC itself is unpatentably old, asymptotically optimal, and though it's not super fast we're talking about very low data rates. You should still get useful gain even with single frame delays, assuming that the code is well matched to the channel. I'm not aware of anything in theory that makes it ineffective with smaller sizes (though obviously you can't get great fading immunity with smaller blocks). (reasonable tools for construction LDPC codes are at http://www.cs.utoronto.ca/~radford/ldpc.software.html ) ------------------------------------------------------------------------------ Symantec Endpoint Protection 12 positioned as A LEADER in The Forrester Wave(TM): Endpoint Security, Q1 2013 and "remains a good choice" in the endpoint security space. For insight on selecting the right partner to tackle endpoint security challenges, access the full report. http://p.sf.net/sfu/symantec-dev2dev _______________________________________________ Freetel-codec2 mailing list Freetel-codec2@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/freetel-codec2
