Andreas Jellinghaus wrote: > Am Montag 19 April 2010 16:57:35 schrieb Jan Just Keijser: >> Note: there is no absolutely secure method to establish a connection >> between a card on a remote machine and the CA . Period. > > I think that is wrong. some smart card protocols work like that: > step 1: ask card for some serial number (or given number) > step 2: start encrypted communication using a key known to you and > stored on the card when it was initialised. > > of course the protocol started with step 2 needs to be secure > against replay attacks and so on, but I guess there are well > known protocols that can be implemented right away. > > I'm no expert on this, but I see no reason why encrypted secure > communication with a card needs more than a one time minimal > initialisation (e.g. writing a number and a key to the card).
This is correct. The future solution is likely to be some kind of ECDH operation because symmetric keys is a PITA. You basically would like to know that this is card brand X serial number nnnnn, not keep track of a gazillion secret keys. The shared secret created by ECDH is then used to secure the rest using symmetric key operations for MACin, encryption etc. Regarding the the rest, there are two schools, one is really a tunnel (like TLS) and the other is using a semi-trusted proxy. I guess semi-trusted proxy sound pretty slimy but it actually works very well if you use "rolling" MACs etc which make the card and issuer able to detect any faults and case operation. I have gotten some indications that this is a minefield of patents but I don't care :-) Anders _______________________________________________ opensc-devel mailing list opensc-devel@lists.opensc-project.org http://www.opensc-project.org/mailman/listinfo/opensc-devel
