What are major advances in technology that effected cryptology?
Early 19th century: telegraph and Morse code
(telegraph operator sees what is being sent)
Later 19th century: Marconi & radio
long distance communication
send to anyone without land line - no wires!
anyone else can listen in!
(Others: Digital information, computers, networks, satellites, fiber
optics, etc.)
How did use of radio change face of cryptanalysis?
used to be too few samples, now almost too many
Intercepting radio codes (in Morse)
What are two aspects that assisted cryptanalysis?
"fist" - signature of telegraph operator (might be able to identify whose
communication)
direction finding - from which direction signal is strongest
example of intersecting two lines:
base1 ------X------------> (direction of signal)
/
base2/
What kind of cryptanalytic attack is this?
Traffic analysis (interception)
this is useful even without decryption
During WWI, why were Germans not as advanced in cryptanalysis?
(hypothesized, pp. 106-7):
Germans march on French, retreating French remove their landlines, but
can still communicate via landline; Germans must rely on radio, more
easily intercepted
Germans doing well, worry less about crypto, more about advance? maybe.
have less to work on to perform cryptanalysis.
Occam's razor: your cryptologic technology is as good as necessity dictates?
ideally... (of course, changes are not instant!)
What is the ADFGVX Cipher?
kind of like Playfair... but two letters encode one letter; numbers
included in grid.
Combines substitution and transposition.
Why ADFGVX: different Morse Code.
Not really new, but, combo makes difficult to use regular frequency
analysis; and still hard to unscramble. in partic: scrambling/breaking
of digrams effectively gives polyalphabetism.
keys: hard- a whole grid; easy - a simple word.
how many possibilities? 36!
36! = 371,993,326,789,901,217,467,999,448,150,835,200,000,000
(2139) (1042)
and then... # of ways to scramble columns so really many more.
When used? Germans, world war I.
Big cryptanalysis? French, Painvin.
Example:
ciphertext: XGFGDFFFDAGFDVXXDVAVGD
keyword: RIGHT
4 3 1 2 5
R I G H T
---------
V D X D A
X A G F V
X G F F G
D F G F D
V D -----
---/
decrypting (by intended):
1. Find shape of grid
count total cipher letters (n)
divide by k (length of keyword) get n = qk + r
r is number of columns (staring from left) with q+1 letters
others have q letters
2. Label grid with keyword in order, but numbered alphabetically
3. transcibe ciphertext down columns in numerical order
4. decrypt from key-grid
Cryptanalysis - nontrivial: what might work? suggestions...
The Zimmerman Telegram
Who was Arthur Zimmerman?
new German foreign minister in 1916
What was his plan?
Have Mexico invade US from South to divert US from sending troops to
Europe. Possible three front attack with Germany (E), Mexico (S), Japan
(W), if Mexico could mediate with Japanese.
kind of code? use diplomatic code - code books involved ... convoluted
some superencipherment, some nomenclator ascpect, some homophone aspect
Why did the British not immediately forward telegram to US?
did not want to Germans to know they could read their telegrams
only partial decrypt initially
eventually, telegram used as a "lever" to encourage US to enter war
At end of WWI, who's winning crypto game? Cryptanalysts.
Improving Vigenere
What was main weakness? Short keys - lead to Babbage/Kasiski
cryptanalysis - break into multiple shift ciphers
What was first idea? Use key as long as message itself - no repetition in
system.
Problem? Likely familiar phrases in key. Can decrypt by manufacturing a
"crib"
Example: decrypt: LAVDYGCFVYCIAQBT
try:
the
???
LAVDYGCFVYCIAQBT
---
STR...
One-Time Pad
How can we avoid this problem?
Use random keys. Do not reuse them. (Ah, but what is random?)
Mauborgne (circa 1918)
Perfect security: if keys are truly random, then any key is equally
likely. a ciphertext of length n could correspond to _any_ plaintext of
length n - there are not only zillions of possibilities, but numerous
possibilities that are grammatical, sensible, etc.
Example: six-letter word. There are 266 = 308,915,776 possibles combos
of letters. A lot by hand - but not by computer. But any six letters word
suffices! My computer dictionary has roughly 25,000 words - close to
4,000 are six letter words!
issues: difficulty in key distribution, random generation, and one-time
use requires synchronization of pads (i.e., if one side falls behind by
one - they never get a message at all, lets say)
So, who uses a one-time pad these days?
(We'll see they can make a comeback via RSA and quantum stuff.)
In the News
What is is meant by the term "biometrics"?
How might biometrics be employed to improve airport security?
How do some ferderal agents think terrorists employed steganogrpahy in
their emails?
Machines
What is a cipher disk?
ease process of generating shift ciphers and vigenere
ciphers
do not make better ciphers, but improve speed of encryption/decryption;
however- can also speed cryptanalysis!
Related idea: Jefferson wheel ... draw diagram
one of earliest cipher machines
What is a rotor?
(See p. 129, Singh)
main idea: subst cipher that changes after each letter is encoded
26 (or whatever) possible subst ciphers per rotor. For long messages
could just use Babbage/Kasiski style attack (although not shift ciphers,
still freq. analysis would work).
So, stack rotors together.
Odometer effect!
Three rotors: 263 = 17,576 possibilities
not that many unless in a rush,
BUT:
ENIGMA
keyboard -> scrambler -> lamp
scrambler: 3 rotors can be put in any order ( * 3! which is 6)
(later of 5 possible rotors: 5*4*3 which is * 60)
What does reflector do?
Makes it easy to decrypt!
What is the key? The initial setting of the rotors (which ones and what
pos they are at) and the plugboard...
What is the plugboard? simple mono susbt cipher (6 pairs of letters,
e.g. swapped), but makes for tremendous number of possible combos.
By itself easy to crack - freq analysis - since plugboard settings fixed
for a given message, but with rotors makes # of keys in thousands of
trillions.
Note: with today's computers it might be possible to brute force decrypt
an enigma.. BUT if the cryptographers had that power they would not be
constrained by the weight of the machine and could easily increase the
number of rotors, etc: that is basic idea for DES
What is the Black Chamber? nickname for Yardley et. al dept for code
breaking in US. Very effective so discouraged people to bother with
encryption...
Why did Enigma not first take off, but then later do so? And why in
Germany and not like items in US? (e.g. Hebern's invention)