Notes for January 14, 1998

1. Greetings and felicitations!
   1. Reading: Pfleeger, pp. 21-46; Garfinkel & Spafford, pp. 139-159, 175-179
2. Puzzle of the day
   1. Just to get you thinking; I'll ask this one again later on to see how your thinking has changed.
3. Classical
   1. monoalphabetic (simple substitution): f(a) = a + k mod n
   2. example: Caesar with k = 3, RENAISSANCE -> UHQDLVVDQFH
   3. polyalphabetic: Vigenère, f_i(a) = (a + k_i) mod n
   4. cryptanalysis: first do index of coincidence to see if it's monoalphabetic or polyalphabetic, then Kasiski method.
   5. problem: eliminate periodicity of key

4. Long key generation
   1. Running-key cipher: M=THETREASUREISBURIED; K=THESECONDCIPHERISAN; C=MOILVGOFXTMXZFLZAEQ; wedge is that (plaintext,key) letter pairs are not random (T/T, H/H, E/E, T/S, R/E, A/O, S/N, etc.)
   2. Enigma/rotor systems; wheels, 3 rotors and a reflecting one. Go through it; UNIX uses this for crypt(1) command.
   3. Perfect secrecy: when the probability of computing the plaintext message is the same whether or not you have the ciphertext
   4. Only cipher with perfect secrecy: one-time pads; C=AZPR; is that DOIT or DONT?
5. DES
   1. Go through the algorithm
6. Breaking UNIX crypt(1)
   1. Purely statistical attack is possible (me) but it takes gobs of ciphertext
   2. Known plaintext attack: that's Reeds and Weinberger's attack, with a nice suggestion by Bob Morris