|
With a much longer list, you would still get the
same sensitivity in the first place, but a higher probability of false
positives. To avoid these, you would then want to raise the threshold
(eg. from 15 to 16 dB), thus decreasing sensitivity indirectly. At
constant false alarm rate, the required threshold fortunately scales only
weakly (less than logarithmic) with the number of entries.
But processing time will be proportional to
the length of the list. For a "true decode" of the 28 bit compressed
callsign, one would need to go through 268 million
templates ;-)
73, Markus
Sent: Saturday, October 19, 2013 10:24 PM
Subject: Re: LF: T/A OPDS
Hello Markus,
Am
19.10.2013 21:36, schrieb Markus Vester:
BTW In my other posting I hadn't explained
the last dB number in the opds result line. This describes the quality of
the match to the callsign template, ie how many dBs the highest peak in the
cross-correlation function exceeds its average power,
which comes from noise and autocorrelation sidelobes. The range
is from about 22 dB for a perfect signal, down to a chosen threshold
of 15 dB where false positives start to appear
occasionally. OK,
understood.
Another question for a better OPDS understanding: Is the
decode performance of the program also a function of the number of call signs in
the list? Example: If i go to add any valid amateur radio callsign in the list,
will decodes (close to the limit) become more unlikely?
73,
Stefan/DK7FC
|
|
