NetNews Usenet Archive 1992 #26/NN_1992_26.iso/spool/sci/crypt/4489

Text File | 1992-11-09 | 1.5 KB | 39 lines

Newsgroups: sci.crypt
Path: sparky!uunet!mcsun!sun4nl!utrcu1!infnews!infnews!roana
From: [e-mail] (Room 101)
Subject: secure communication via third party
Message-ID: <[e-mail]>
Keywords: diffie-hellman
Sender: [e-mail]
Nntp-Posting-Host: buick
Organization: University of Twente, Dept. of Computer Science
Date: Mon, 9 Nov 1992 11:23:04 GMT
Lines: 26

Consider two parties X and Y who have never met and never exchanged
information. They wish to establish a confidential and authenticated
channel. Since they can't communicate directly, they need to send their
messages to Z, who (supposedly) routes them to the recipient.

X and Y can use Diffie-Hellman to establish a secret key using public
messages that Z can read. However, Z might intelligently change these
messages to create a transparent channel between X and Y. Z will be able
to read and change all messages sent over this supposedly secure channel.

Is it possible for X and Y to detect if Z has created a transparent
channel between them, for example by using the fact that both have
calculated different secret keys. I know authentication is possible
if all communication between the parties is conversational, since X
can send Y a hashed value of its next block and the key, wait for Ys reply
and send the real block. Z cannot fake this, since it doesn't know the
original block of data. However, this type of check is impractical
for everyday use on network layers.

Any suggestions are welcome.

Ronald
[e-mail]