複製下來當備份 XD
原文沒有斷行, 好難閱讀 QQ
from http://mailman.isi.edu/pipermail/ns-users/2004-June/043128.html
Well, it is actually the other way. RXthreshold and CSthreshold are inherent
properties of the card (that depends on modulation/demodulation, among other
things). So for usual commercial cards, CSThreshold is considered to be about
-90dBm (exact numbers are usually not released by the manufacturers).
RXThreshold depends on the packet size. In my experience, -65 dBm is sufficient
to decode very large packets (about 2000 byte) with very low probability of error;
for very small packets, even -80 dBm may be enough. Also remember that these
are not parameters of the card; meaning that you cannot control them.
It just so happens that the card can reliably detect carrier only when
the power is more than -90dBm; same with the packet. threshold.cc actually calculates what should the RXThreshold be if you "want"
the receive range to be 250m (or whatever is the distance). The same program
will tell you what the CSThreshold should be if you want the carrier sense
range to be 550m (which should be around -90 dBm). So it is simply provides
a way to "cheat" in the simulation. In other words, if you want to simulate
a situation where the RX-range is 400m and CSrange is 450m, using threshold.cc
you can set the RXthresh_ and CSThresh_ appropriately, but that does not
mean that there exists a physical card which will have those values of RX
and CS thresholds.