On Jan 26, 2010, at 1:25 PM, Jim Reisert AD1C wrote:
> MTBF, or mean time between/before failure, is a fixed quantity. When
> you INCREASE the number of transmitted characters, the MTBF stays the
> same, but the likelihood of experiencing a failure INCREASES.
The only time I can think of, when what Jim mentioned above is not true is when
the extra characters are part of an error correction scheme.
The simplest code is repeating. Repetition is even used in hardware, for
example in early JPL spacecrafts in the form of "triple modular redundancy"
(Avizienis' papers come to mind).
If you send an exchange twice, such repetition forms an error *detection*
scheme. I.e., if you receive two responses that are different, you know there
is an error and can ask for a repeat. If you send an exchange three times, it
forms an error *correction* scheme for many kind of errors (it does not provide
correction in an exchange where all three numbers are different :-).
Notice that repeating an exchange also has the (good) property of temporal
diversity. Frequency-, polarization- and spatial-diversity are also powerful
tools to counter channel problems on HF -- all RTTY users already take
advantage of frequency-diversity because two separate carriers are used in FSK.
Better ways to detect and correct errors exist (and used by all modern modes
such as QPSK31, MFSK16, etc). All of them introduce structured redundancies
into the data stream. Repetition codes are actually extremely inefficient.
You don't find computer memory systems using repetition for the ECC mechanism
:-).
It is also known that you can introduce bad redundancies -- the MTBF case that
Jim mentioned is a prime example.
73
Chen, W7AY
_______________________________________________
RTTY mailing list
RTTY@contesting.com
http://lists.contesting.com/mailman/listinfo/rtty
|