> Clocking forward 100 times opens up 12 alternative pathways on > average, so there's ways around this. In 2.7% of cases (see above) > you would hit on the exact state you came from. In other cases > there's a chance to get around the illegal state. So this chance is > small*small. > > With reference to the mail I sent yesterday, this is also a method > to get even more states which in the end generate the same > keystream: clock forward 101 times, clock back 1 time into another > path while checking if the same bit of keystream had been generated. > With roughly a 5% chance of finding more states leading to the same > keystream. (and do the same clocking forward 100+N times for N=2..16 > to get even more sibling states)
If you start from a random state and do the above in a loop with N->large, (i.e. as much keystream as we have in a burst) how many "connected" states are you going to find producing the same keystream? > > By the way: to verify if a state is in fact the state that led up to > the keystream, you would only have to generate the rest of the > keystream and match it against the actual keystream. The same holds > for discarding streams after 100+N clockings, where you would match > bits generated before the bits you've looked up. Considering the > growth of ancestor states (some 5% per clock max) and the > information in the keystream (50% per clock) you can be sure to have > the right Kc even before looking at another burst of the > conversation. > ...if the real state is indeed conntected to the one you looked up. _______________________________________________ A51 mailing list [email protected] http://lists.lists.reflextor.com/cgi-bin/mailman/listinfo/a51
