Advanced Encryption Standard (AES) - SP Network Cipher #

AES is a symmetric block cipher. It uses a 128-bit (16 bytes) block size and a 128, 192, or 256-bit key size to encrypt the block of plain text.

AES utilizes both Substitution and Transposition.

Key Size	Rounds
128-bit	10
192-bit	12
256-bit	14

Encryption #

Key Expansion #

Generate keys for each round from the original key. Number of keys generated is equal to the number of rounds + 1.

Initial Round #

Add the first round key to the plain text to create the cipher text.

Rounds #

Repeat the following steps for the number of rounds - 1.

Substitute Bytes: Substitute each byte of the state matrix with a different byte from the S-Box. S-Box is predefined and is used to substitute the bytes.
Shift Rows: Shift the rows of the state matrix by different offsets. Row 0 is shifted by 0 bytes, Row 1 is shifted by 1 byte, Row 2 is shifted by 2 bytes and so on.
Mix Columns: Mix the columns of the state matrix by multiplying the state with a fixed matrix.
Add Round Key: Combines the state matrix with the current round key through bitwise XOR operation.

Key Space #

The key space depends on the size of the key.

\[ Key \ Space = 2^{key \ size} \]

Attacks #

Exhaustive Key Search
Side Channel Attack

Also See #

Encryption Modes #

Electronic Code Block (ECB) Mode #

Each Block is encrypted independently.

\[ C_i = E_k(B_i) \] \[ B_i = D_k(C_i) \]

Pros #

Simplicity
Tolerate Losses - a corrupted block will not affect other blocks as every block is encrypted independently

Cons #

Same plaintext is encrypted to the same ciphertext and can thus reveal patterns in the plaintext. (ECB Penguin)

Cipher Block Chaining (CBC) Mode #

Use previous cipher to calculate the cipher for the current block. When encrypting the first block use an initialization vector (IV) as the previous cipher.

\[ C_0 : Initialization \ Vector \] \[ C_i = E_k(B_i \oplus C_{i-1}) \] \[ B_i = D_k(C_i \oplus C_{i-1}) \]

Pros #

Fixes ECB Penguin
Tolerate losses - if a block is lost, at the maximum two block will be affected (the lost block and the next block)

Cons #

Encryption happens sequentially, not parallelizable.

Cipher Feedback (CFB) Mode #

Encrypt the previous block and XOR it with the current block.

\[ C_i = E_k(C_{i-1}) \oplus B_i \] \[ B_i = E_k(C_{i-1}) \oplus C_i \]

Pros #

Faster - Decryption is not applied to the whole block, only the last block.

Output Feedback (OFB) Mode #

OFB follows the principle of one-time pad by using the provided key and the initialization vector (IV) to generate a stream of bits(expanded).

The stream is then XORed with the plaintext to produce the ciphertext.

\[ V_0 : Initialization \ Vector \] \[ V_i = E_k(V_{i-1}) \] \[ C_i = B_i \oplus V_i \] \[ B_i = C_i \oplus V_i \]

Pros #

Parallelizable - After the Key Blocks (keys stream from initialization vector) is generated, the XOR operation on the blocks can be applied in parallel.
Tolerate losses - Each block is encrypted independently.

Counter (CTR) Mode #

Builds upon the idea of OFB.

Use a random seed and counter to generate a stream of bits(expanded).

\[ S : Random \ Seed \] \[ i : Counter \] \[ V_i = E_k(S + i - 1) \] \[ C_i = B_i \oplus V_i \] \[ B_i = C_i \oplus V_i \]

Pros #

Completely parallelizable
Tolerate Losses

Summary #

Mode	Pros	Cons
Electronic Codebook (ECB)	Simple, tolerant to losses	Same plaintext encrypted to same ciphertext, revealing patterns in plaintext
Cipher Block Chaining (CBC)	Fixes ECB Penguin, tolerant to losses	Encryption happens sequentially, not parallelizable
Cipher Feedback (CFB)	Faster, tolerant to losses	-
Output Feedback (OFB)	Parallelizable, tolerant to losses	-
Counter (CTR)	Completely parallelizable, tolerant to losses	-