On Thu, Feb 21, 2002 at 10:11:31AM -0800, Crispin Cowan wrote: > >Note squib on City of Glendale going to wireless LAN. I'm not sure that 3DES > >encryption is all that secure--though it may be for Glendale... > > > 3DES is pleanty secure for most any application requiring symmetric > cryptography. 3DESs primary shortcoming is its speed: very, very, slow, in comparison to other symmetric ciphers. Hardware implementations don't have this weakness. > What makes the security of the Glendale application sketchy is that 3DES > (and single DES, and AES) are not very useful for authentication, and > the one-paragraph article does not discuss authentication. This is a > concern, because it is authentication problems that broke 802.11b and > 802.11a. Hmm; if I recall correctly, authentication was only one problem encountered with WEP/802.11; improper use of rc4 allowed keys to be recovered quickly. From a cursory glance at the Borisov, Goldberg, Wagner paper, the problem is liable to exist in a 3DES implementation as well. Furthermore, the packets are protected only with crc-32, woefully inadequate for the job. This problem is liable to exist in a 3DES implementation as well -- but should be possible to be replaced, should someone care enough. There are two authentication problems -- traditional 802.11 implementations use a single shared key amongst all workstations. (!) The 802.1x RSN 'fix' is actually not a fix, as pointed out by Arbaugh recently. (It lacks bidirectional authentication. Oops.) Maybe some of the 802.11 stuff is salvagable for secure use -- but I think it will require extensive re-engineering of the security protocols. Cheers :) -- "I'm not sure which upsets me more: that people are so unwilling to accept responsibility for their own actions, or that they are so eager to regulate everyone else's." -- Kee Hinckley
This archive was generated by hypermail 2b30 : Sun May 26 2002 - 11:39:05 PDT