Starium (was Re: article: german secure phone)

Peter Fairbrother peter.fairbrother at ntlworld.com
Wed Jun 6 20:30:09 EDT 2001


> Bram Cohen at bram at gawth.com wrote:
[..]
> I can't emphasize enough that it's very important that the form factor be
> a double-female phone jack and work when plugged in with *either*
> orientation - is this an easy thing to detect?

Surely a male-to-female jack. Plug it (male) into the wall socket and plug
the phone into it (female). You could put the electronics in a separate box
if the jack is too small. Or a box with a tail. If you have to have a double
female, like an external modem has, does it matter if, like a modem, it
doesn't work if you plug it in the wrong way round? That is easy to detect.

I don't see why you can't sell a handset though. People buy new ones when
they redecorate. And they're more resistant to Tempest-style attacks, as
unencrypted speech isn't transmitted along the cord.

A handset also allows a protocol to avoid some MITM without authentication:
Alice calls Bob, exchange DH, Bob replies with a few spoken digits hashed
from their shared secret, which he sees on a display on the handset. Alice
keys these digits into her phone, if they match she can speak (until then
her voice doesn't get transmitted). Not perfect, but it helps.

Too much user interaction? Users _need_ interaction to make them aware of
the security status of the line they are using. They've all seen the spy
films where people use STU's and say "go secure" and press a button and wait
for the bleeps, if it doesn't do that they won't believe it. If you have no
user interaction then what happens if eg the kids unplug the box and you
don't notice?

A handset should, of course, be compatible with a computer/headset and
encrypted-voice-over-internet software. Perhaps a pda/mobile combo too.


-- Peter




---------------------------------------------------------------------
The Cryptography Mailing List
Unsubscribe by sending "unsubscribe cryptography" to majordomo at wasabisystems.com




More information about the cryptography mailing list