I got some e-mail back from Rivest; in fact he does forsee a key exchange method like Diffie-Hellman as a way to overcome the problem of exchanging keys. Agree on a large prime in the clear, n, and a number g such that g is primitive mod n (fuzzy on what that really means... g exponentiated to different powers, mod n will give at least n different results. Actually D-H needs "fairly primitive" g's.) Alice chooses a random large integer x and sends Bob ( X = (g exp x) mod n ). Bob chooses a random large integer y and sends Alice ( Y = (g exp y) mod n ). Alice finds k = (Y exp x) mod n, Bob finds k' = (X exp y) mod n. k = (Y exp x) mod n k = ( (g exp y) mod n ) exp x ) mod n k = (g exp xy) mod n k' = (X exp y) mod n k' = ( (g exp x) mod n ) exp y ) mod n k' = (g exp xy) mod n k = k' x and y are still secret to Alice and Bob, and no one can find k/k' without knowing x or y. Alice and Bob now both have the secret key k. Magic! -hedges- (Schneier, Applied Cryptography, 1996, p. 513)
This archive was generated by hypermail 2b30 : Fri Apr 13 2001 - 13:06:46 PDT