Re: [MT63] What is MT63 good for?
- dalite01@... wrote:
> Your description is most likely much more technically correct.Yes, Hamish has got it right. :)
> The end result is latency or decreased throughput.This isn't exactly true either. It is possible to implement
interleaving so that it does not decrease throughput. And with
packetized data, it is possible to do it so that there is no
(extra) hit on latency either.
That said, the way interleaving is done in MT63 does increase
latency and, with the short messages passed in typical QSOs,
it has a significant impact on throughput as well.
> FWIW, my seat of the pants relating FEC to redundancy would probably notreceiving
> pass muster in the technical arena either :) . I have always equated FEC
> with sending multiple instances of the same information, which the
> station must reassemble multiple instances of verification beforeprinting
> the data to the screen. In the case of MT-63, (If I remembercorrectly) the
> authors say that up to a 25% loss of redundancy will not affect accurateseems to
> copy. I am sure that there is a more factual way to present the process,
> but the concept of redundancy and verification through redundancy
> be easier to comprehend to those using and not writing the software.FEC is always done by adding redundancy to the sent data. Sending
multiple copies of the same message is one way of adding redundancy
but it so happens that it is very inefficient way of doing it. Using
clever math, it is possible to add redundancy in much better ways,
giving greater "coding gain".
Tomi Manninen / OH2BNS / KP20JF60TP