On Thu, Oct 18, 2001 at 09:11:05AM -0700, Greg Wiley wrote: > This is just a general idea: > > I'm assuming a 32-bit timestamp. > > You want to encode that value somewhat securely > with an invertable hash into 16 * 7 = 112 bits > (less if you can't use control chars). It has to be > somewhat resistant to tampering (you gave no > indication of the degree to which it must resist). > > So you can multiply your timestamp by > any number < 2^(112-32) = 2^80 = > 1208925819614629174706176 . > and still fit in 16 7-bit chars. > > I would pick a prime value < 2^80 and multiply the time > by it, then format it into 7-bit characters. > > To extract, pack it and divide. > > Does that work? > > Note that to break this, the attacker will have to > search a fairly large space but not an impossible > one. Some of the effectiveness of this will depend > on how often you change the prime factor. Also, > if it is fairly expensive for an attacker to check > a single value, the search could take a while. > This also doesn't take into account the value > of breaking it, i.e. how big is the "prize" for > finding your prime factor?
If the attacker knows the algorithm (although not the prime number) this is unfortunately trivial to crack: they just have to guess the time that is encoded by the timestamp. :( -- David Roundy http://civet.berkeley.edu/droundy/
