stuart@cs.rochester.edu (01/21/88)
I am somewhat surprised that no-one has mentioned it (perhaps I
missed a relevant article or two) but one standard technique for
fooling traffic analysis (an element of cryptology but not
cryptography) is to insert enciphered "dummy" messages into the
traffic. On continuous data streams this can be taken to the limit
of sending traffic constantly whether you have "real" traffic or not.
A moment's reflection will show you that this technique would be
without value if a dummy message could be distinguised from a real
message in the enciphered traffic. (Note that on continuous streams
you are not going to go to very much trouble to produce dummy messages
with the same characteristics as real messages; you'll just send bits
and exploit the superiority of your cipher.)
Stu Friedberg {ames,cmcl2,rutgers}!rochester!stuart stuart@cs.rochester.eduhenry@utzoo.uucp (Henry Spencer) (01/28/88)
> ... (Note that on continuous streams > you are not going to go to very much trouble to produce dummy messages > with the same characteristics as real messages; you'll just send bits > and exploit the superiority of your cipher.) Ah, but you'd better exercise at least a little bit of cleverness. If the dummy-message content is too predictable, and the opposition finds out about it (remember the standard assumption that the keys can be kept secret but the machinery can't be), you've just handed them a dandy starting point for known-plaintext attacks on your cipher. -- Those who do not understand Unix are | Henry Spencer @ U of Toronto Zoology condemned to reinvent it, poorly. | {allegra,ihnp4,decvax,utai}!utzoo!henry