On 26/05/2020 10:33, Reino Talarmo wrote:
Just a single bit error in a proper place adds e.g. /R to the decoded message.
Hi Reino,
that is far from the actual position. A single bit error will never
result in an incorrect decode, the amount of FEC information incuded can
recover a lot more than single bit errors. On top of that the
FT4/T8/MSK144 message encoding includes a CRC value used to validate the
message, because of this the vast majority of incorrect message decodes
are rejected because the CRC does not match the message content.
All waveforms will decode to something, including noise. A false decode
is only printed when the noise happens to represent a message that
matches the CRC also represented by noise. What must be noted is that on
an empty band 100% of decodes are false, when listening to an empty band
hoping for a DX opening you must expect some false decodes. Much of the
prevalence of false decodes results from false sync detections, the
modes are designed with a carefully chosen balance of sync energy versus
information, FEC and CRC information. The optimum balance gives a very
good ratio of valid to false decodes at the signal levels targeted (i.e.
the weak ones), on an empty band or with all signals badly corrupted
that balance is not optimal and the false to valid decode ratio
increases. Nevertheless the false decode rate does not change more than
a tiny amount.
For those concerned about automatically answering false decodes on empty
bands then I suggest not checking the "Call 1st" option.
73
Bill
G4WJS.
_______________________________________________
wsjt-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/wsjt-devel
