On Jan 15, 2006, at 11:33 AM, Douglas Hall wrote:
> From my perspective, it's not so much odd numbers as (in this case) 3
> of them. Sending "IL" 3 times provides a way to do a 2/3 majority
> vote.
> If it comes out on my side as "IL IN IL" I can say with some certainty
> that it's "IL" whereas if it's only sent twice it could be either, and
> if only once it could be anything.
In the early days of spacecrafts, Avizienis at JPL had introduced the
concept of "Triple Modular Redundancy" into their hardware. Critical
hardware were replicated in three, together with a voting mechanism.
Error correcting bits of data were introduced way earlier than that,
the first formalization probably came from Hamming.
Hamming's original paper included descriptions of the ad-hoc error
correction/detection used up until then:
http://guest.engelschall.com/~sb/hamming/
When error rates are low enough, I prefer to send just two exchanges
(ditto call signs when calling in S&P mode). This provides
reasonable error detection, and a complete repeat when requested.
When error rates are low, the complete repeats are rare enough that
it is a decent tradeoff for me.
Does Cabrillo check that both stations have copied the exchanges
correctly before either one gets credit for the contact? If not, it
could explain the practice of not repeating an exchange :-).
73
Chen, W7AY
_______________________________________________
RTTY mailing list
RTTY@contesting.com
http://lists.contesting.com/mailman/listinfo/rtty
|