Could such a cognitive system also be useful on the encoding side? Whenever I see postprocessing ideas like this I tend to think "what if you put that model in the coding loop?". Like the loop filter in H.264 for example.
/Tomas On Wed, 2017-11-01 at 13:19 +0100, Matei Alexandru Coltoiu wrote: > Hello, > > I am trying to find a different approach regarding the poor error > handling issue. > My Idea is to write a demodulator from scratch. Instead of either > having an output consisting of approximated values (later to be fed > into the FEC decoder) or having none, we could write the demodulator > so that it adds a "tail" after each symbol so that we have > floating point values in the end. Take this example: The demodulator > would either aproximate a value between 00 and 01 or we would have an > error. With my example, it would output a value between 0000000001 > and 0011111111. We would than use a cognitive decoder to "guess" wich > value it should use. I am aware that a cognitive decoder is not a > piece of cake and that we ar fighting with highly compressed data. > But, given the overhead reduction from the FEC, we should bbe able to > get a decent gain. The result in low SNR should be a noisy output but > inteligible, similar to an analog radio (sort of).
signature.asc
Description: This is a digitally signed message part
------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________ Freetel-codec2 mailing list Freetel-codec2@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/freetel-codec2
