On Feb 10, 2013, at 11:56 AM, Bill Turner wrote:
> What to do? The answer is REDUNDANCY. Send the exchange more than once, as
> most of us do. But for maximum redundancy, you should send not only the
> exchange more than once BUT ALSO the FIGS character in front of each "001".
Aha, I see you fell for that fallacy, Bill.
If you are repeating an exchange only once, it is only good as what is known as
an Error Detection code. I.e., it tells you something is wrong, it does not
tell you *how* it is wrong.
So, if you see "599 QWE 123" (and assuming one is ignorant of the QWERTY
mapping), that will still trigger a request for a repeat. The 123 might look
good to you because you know a-priori that you are trying to send a 123. The
other end (assuming a neophyte) would not know. How does he know that you mean
023 or 123? (0 and 1 in Baudot are Hamming Distance of 1 apart).
In either of the two cases, whether you see "599%QWEAQWE" (dashes), or if you
receive "599 %AQWE 123" (spaces with USOS), the neophyte will still ask for a
repeat. Real RTTY ops can read either of them fine, as many folks have already
stated. (the percent sign above indicates the location of the mis-received
character.)
I.e., as long as we are talking about redundancy that is only good for Error
Detection, that second space and second repeat is not going to keep you from
having to repeat an exchange. Both Space and Dash will let the other end know
that something is wrong, and not what is wrong, and both triggers a repeat.
If we are using primitive repeat codes it is only when you use three or more
repeats where a single error does not trigger a repeat.
In the hardware world, the repeat-twice technique to correct errors is known as
Triple Modular Redundancy and was used by Avizienis, as I have mentioned here
on the past:
http://lists.contesting.com/archives//html/RTTY/2006-01/msg00361.html
73
Chen, W7AY
_______________________________________________
RTTY mailing list
RTTY@contesting.com
http://lists.contesting.com/mailman/listinfo/rtty
|