Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

SLP proposal


I was looking over your SLP proposal slides and there seem to be
some errors in the calculation of the probability of a false match.
The equation for the number of 96-bit patterns per year was

(10^10 bits/s) x (1/96 bits) x (3600*365 s/year) ~ 1 * 10^12 / year

Since Ethernet packets can be any number of bytes long within the
constraints of minimum and maximum packet size, one must check for 
a match every 8 bit rather than every 96 bits. Also, there are
(3600*24*365 s/year). When I multiply your existing equation out,
the result is 137 x 10^12 rather than 1 x 10^12.

The equation should be
(10^10 bits/s) x (1/8 bits) x (3600*24*365 s/year) ~ 40 * 10^15 / year

which results in a false match about every 14 million years which is 
still quite rare, but I hate to see an incorrect calculation propogated
as others may copy it without checking. 


PS. Please pardon the lack of your address in the To line, but I don't
have an address for you.