计算机网络复习笔记第十部分--加密（英文版本）

Encryption and Decryption

encryption model

Encryption:
an encryption process E (with an encryption key k, E角标k) transforms a plaintext P to a ciphertext C: C=Ek ( P )
Decryption
a decryption process D (with a decryption key k’) transforms the ciphertext C to the original message P: P = Dk’ ( C )
– E and D are just mathematical algorithms
– often k = k’

traditional cryptography

substitution代替 cipher （Caesar cipher）

basic idea is replacement between bits

Plaintext: ‘the caesar cipher fooled the carthaginians’

• shifting each alphabet by k. 假如k=1，每个字母都往后挪一个。 a→ B, b→ C, . . ., z→ A, 码字变成‘ UIF DBFTBS DJQIFS GPPMFE UIF DBSUIBHJOJBOT ’
• mapping alphabets randomly to others, a→ Q, b→ W, c→ E, . . .码字变成‘ ZIT EQTLQK EOHITK YGGSTR ZIT EQKZIQUOFOQFL ’
• (a bit more complex) mapping all ASCII characters randomly to others.

Ciphertext: ‘ ZIT EQTLQK EOHITK YGGSTR ZIT EQKZIQUOFOQFL ’

– 26!∼4× 1026 possible keys exist for random alphabet mapping — at 1µ(sec) per key, 1000 machines will solve in 1010 years.

– in fact, not difficult to break by exploiting the statistical properties of natural language — e.g., for English,

are most common.

Breaking the subsititution cipher: ‘ ZIT EQTLQK EOHITK YGGSTR ZIT EQKZIQUOFOQFL ’

• frequent letters: Q (5 times), T (5), I (4), E (3), K (3), O (3), Z (3), .=⇒ they may be substitutions for ‘ e, t, o, a, n, i ’.
• a trigram三字母组: ZIT (twice) =⇒ possibly ‘ the ’.
• here is one guess: Z→ t, I→ h, T→ e, Q→ a, O→ i, ‘ the EaeLaK EiHheK YGGSeR the EaKthaUiFiaFL ’
• and this ‘guessing game’ continues for the rest, using further statistical clues.

transposition cipher 转移密码

basic idea is rearrangement of bits

how to do this cipher:

• choose a key k to number the columns
• write the plaintext in rows
• read the ciphertext by columns, starting with the column whose key letter is the lowest.

breaking the transposition cipher:

• find a cipher ⇐ test the statistical pattern
  if there are many e, t, o, a, n, i, it is more likely to be a transposition cipher than a substitution cipher.
• find # columns ⇐ guess for contents of message
  we suppose ‘swiss bank account’ to occur and this cipher to have nine columns, bigrams such as sk, ia, sc, bu, an, nt may appear.
• order the columns ⇐ test the statistical pattern
  – first we choose two columns that fit best to the bigram pattern of plaintext
  – then we find the third column that fits best to the bigram/trigram patterns (continue for all columns)

Product cipher

combination of elements to build complex ciphers

modern cryptography

compare modern and traditional cryptography:

• traditional cryptography: uses simple algorithms, but relies on long keys in order to make ciphertxt more cryptic
• modern cryptography: keeps the keys short, but uses complex procedures to encrypt data

DES(data encryption standard)

• key is 56 bits length
• plaintext is encrypted in blocks of 64-bits, yielding 64-bits of ciphertext
• 19 stages includes 16 functionally identical iterations迭代
• decryption is done with the same 56-bit key, but run in reverse order
  

breaking the DES— suppose a small piece of plaintext and matched ciphertext is given:

• can design a machine that does exhaustive search (256∼7.2× 1016 keys) in less than one day.
• national lottery attack
  – ten million people own a TV set, installed with a DES chip芯片 that does only one million encryptions per second (very cheap).
  – once a plaintext/ciphertext pair is broadcasted, each of ten million chips begins searching preassigned section of key space.
  – within two hours, one TV set hits a jackpot and the owner wins a fortune. ⇒ DES cannot be used for anything important.

Brute-force attack

• also known as an exhaustive详尽的 search for discovering keys
• try every combination until the correct key is found
• can be used against most of encrypted data
• often prohibitively过分地 inefficient
• national lottery attack is one example for a brute-force attack

Merkle’s puzzles and Shamir’s method

Merkle’s puzzles


Shamir’s method:

• each of k persons - neither of them knows the key --carries unique data point (xi,yi)
• k points uniquely determine the polynomial, whose coefficients系数 (ai’s) are then used to derive a key
• if less than k persons arrives the destination, a polynomial cannot be uniquely determined, but it is still possible to find a relation between coefficients

public key encryption

There is no reason encryption algorithm cannot be made public, suppose

• D(E§) =P
  where E/D: encryption / decryption algorithms, P: plaintext
• it is exceedingly difficult to deduce推论 D from E

Scenario:

• encryption algorithm E with key k is made public, but decryption algorithm D, whose key k ′ is different from k, is kept secret.
• anybody, wishing to send a message, encrypts with E.
• ciphertext cannot be read unless you know D.

RSA algorithm

• parameter selection:
  – choose two large primes质数, p and q. (typically greater than 10的100次方)
  – compute n = pq, and z = (p-1)(q-1)
  – choose a number relatively prime to z and call it d
  – find e such that $e\times d$ modz =1
• procedure:
  – the pair (e,n) is made public.
  
  – the rest (p,q,z,d) is kept private.
  

Why is the RSA algorithm secure?

• n is made public , n=pq
• difficult to fatoring large numbers 分解大的数字难
• Estimation — using a computer that makes one calculation per one micro second: factoring 200-digit number requires four billion years ; factoring 500-digit number requires 1025 years

看例子可以很清楚地看明白这些系数的关系。

digital signature

‘Signed message’ system签名系统 should satisfy

• authentication of the process: the receiver can verify the claimed声明 identity of the sender
  (e.g.) a bank needs a authorised signature for transaction.
• protection of message receiver: the sender cannot later change the contents of the message.
  (e.g.) a dishonest customer might withdraw money, then sue the bank that he never did such transaction.
• protection of the message sender: the receiver cannot forge伪造 the message
  (e.g.) a stock broker might buy a junk stock without a customer’s approval.

Secret key signature:

advantage:

• fast and efficient

disadvantage:

• the secret key is required to verify the authenticity of the message, leading to a key distribution problem;
• or find a third party that moderate协调 the communication.

Secret key system:

• needs a moderator that everyone trusts
• not that easy to find such moderator
  

Public/private key signature:

advantage:

• public keys can be freely and openly distributed分发

disadvantage:

• calculations for encryption/decryption are very costly

signing the message签署消息 using a public/private key system:

• authentication of message
• sender digitally signs a message using the private key
• suppose a public key encryption/decryption algorithm can do E(D§) = P
• of the recipient can decrypt this message using sender’s public key, it is the proof for authenticity真实性 of the sender
  

signing and encoding a message using the public/private key system:

• suppose D(E§) = P , E (D§) = P
  (e.g.) RSA algorithm has this property
• the recipient (‘B’) is sure that
  – ‘A’ is the actual sender of the message
  – only the recipient can read the message
  

message digest (MD)

• also known as a hash value散列值 (or simply a hash)
• a fixed size, quasi-unique fingerprint generated by a formula from an arbitrary任意值 sized message
• similar in principle to the computation of CRC
• typical digest size of 128 to 256 bits to represent 10的38次方 to 10的77次方 different values
• much less costly than encrypting the entire message
• detection of corrupted message
• protection against unauthorised modification, achieved by encryption. (note) MD alone does not offer any protection

Sending a message:

• compute MD from a plaintext P using a hash function
• encrypt MD using the sender’s private key
• send the plaintext P and the encrypted MD

Receiving a message:

• compute MD′ from P using the same hash function
• decrypt MD using the sender’s public key
• compare MD′ with MD

A good hash function:

• efficient to compute MDs
• virtually impossible to find a message having the
  identical MD
• dependent on every bit of the message
  (i.e., two nearly identical messages to have totally different hash values)