High throughput AES architecture
View Patent ↗An advanced encryption system (AES) architecture includes a maximum parallel encryption module which implements one round of the AES algorithm in one clock cycle, and a maximum parallel key scheduling module which generates sub-keys in one clock cycle in parallel with the encryption module, thereby permitting feedback modes of operation to be used without adversely affecting AES throughput. A controller controls the operation of the encryption and key scheduling modules such that one round is completed per clock cycle. The controller is preferably part of a hierarchical distributed control scheme comprising communicating finite state machines (FSMs). The architecture also preferably includes asynchronous input and output buffers.
1. An advanced encryption standard (AES) architecture which provides high throughput and low latency, comprising:
a maximum parallel encryption circuit that receives a plurality of data bytes to be encrypted and implements one round of the AES algorithm in one clock cycle;
a maximum parallel key scheduling circuit that generates sub-keys in one clock cycle in parallel with said maximum parallel encryption circuit, said sub-keys provided to said maximum parallel encryption circuit; and
a controller that controls the operation of said maximum parallel encryption and key scheduling circuits such that said AES architecture performs one round of the AES algorithm in one clock cycle;
wherein said AES architecture employs a chosen key-length k and a data-block length m, said maximum parallel key scheduling circuit comprising:
a first key scheduling sub-circuit that receives a previously-generated sub-key and generates a current sub-key;
a second key scheduling sub-circuit that receives the current sub-key and generates a next sub-key; and
a roundkey assembly sub-circuit which receives said previously-generated sub-key, said current sub-key, said next sub-key and which assembles an m-bit roundkey from said previously-generated sub-key, said current sub-key, and said next sub-key, and provides said roundkey to said maximum parallel encryption circuit.
2. An advanced encryption standard (AES) architecture which provides high throughput and low latency, comprising:
a maximum parallel encryption circuit that receives a plurality of data bytes to be encrypted and implements one round of the AES algorithm in one clock cycle;
a maximum parallel key scheduling circuit that generates sub-keys in one clock cycle in parallel with said maximum parallel encryption circuit, said sub-keys provided to said maximum parallel encryption circuit; and
a controller that controls the operation of said maximum parallel encryption and key scheduling circuits such that said AES architecture performs one round of the AES algorithm in one clock cycle;
wherein said AES architecture implements the AES standard with at least 128 bit data-blocks and a chosen key-length k, said maximum parallel key scheduling circuit comprising:
a key scheduling sub-module arranged to receive a previously-generated sub-key and to generate a current sub-key; and
a roundkey assembly sub-circuit which receives said previously-generated sub-key and said current sub-key and assembles an at least 128-bit roundkey from said previously-generated sub-key and said current sub-key, said at least 128-bit roundkey provided to said maximum parallel encryption circuit.
3. The AES architecture of claim 2 , further comprising:
an asynchronous input buffer that receives data bytes to be encrypted, to buffers a plurality of said data bytes in parallel, and provides parallel data bytes to said maximum parallel encryption circuit; and
an asynchronous output buffer that receives an output of said maximum parallel encryption circuit and outputs encrypted data bytes to an output bus.
4. The AES architecture of claim 3 , wherein said maximum parallel encryption circuit comprises:
a substitution sub-circuit comprising substitution blocks which are replicated as needed to receive all of said parallel data bytes from said asynchronous input buffer simultaneously;
a shift row sub-circuit which receives the outputs of said substitution sub-circuit;
a mix column sub-circuit which receives the outputs of said shift row sub-circuit; and
a key addition sub-circuit that receives and combines the outputs of said mix column sub-circuit and said sub-keys from said maximum parallel key scheduling circuit, and provides the results at an output, said output being the output of said maximum parallel encryption circuit.
5. The AES architecture of claim 4 , wherein said maximum parallel encryption and key scheduling circuits are implemented exclusively with combinatorial logic.
6. The ABS architecture of claim 3 , wherein said controller is implemented with a hierarchical distributed control scheme comprising communicating finite state machines (FSMs), comprising:
a main FSM; and
local FSMs which are controlled by said main FSM, said local FSMs comprising:
a maximum parallel encryption circuit FSM which controls said maximum parallel encryption circuit;
a key scheduling circuit FSM which controls said key scheduling circuit;
an input buffer FSM which controls said asynchronous input buffer; and
an output buffer FSM which controls said asynchronous output buffer.
7. The AES architecture of claim 2 , wherein said controller is implemented with a hierarchical distributed control scheme comprising communicating finite state machines (FSMs).
8. The AES architecture of claim 2 , wherein said AES architecture implements a Rijndael algorithm with a data-blocks length of 128, 192 or 256 bits and a key-length of 128, 192 or 256 bits.
9. The AES architecture of claim 2 , wherein said AES architecture implements the AES standard with a data-block length of 128 bits and a key-length of 128, 192 or 256 bits.
10. The AES architecture of claim 2 , wherein said AES architecture implements the electronic code book (ECB) mode of operation.
11. The AES architecture of claim 2 , wherein said AES architecture implements a feedback mode of operation.
12. The architecture of claim 2 , wherein:
a said maximum parallel encryption circuit receives the output of an asynchronous input buffer and implements one round of the AES algorithm in one clock cycle;
wherein said controller is a hierarchical distributed control scheme comprising communicating finite state machines (FSMs).
13. The AES architecture of claim 12 , wherein said maximum parallel encryption circuit comprises:
a substitution sub-circuit comprising substitution blocks which are replicated as needed to receive all of said parallel data bytes from said asynchronous input buffer simultaneously;
a shift row sub-circuit which receives the outputs of said substitution sub-circuit;
a mix column sub-circuit which receives the 10 outputs of said shift row sub-circuit; and
a key addition sub-circuit that receives and combines the outputs of said mix column sub-circuit and said sub-keys from said maximum parallel key scheduling circuit, and provides the results at an output, said output being the output of said maximum parallel encryption circuit;
each of said maximum parallel encryption module sub-circuits implemented exclusively with combinatorial logic.
14. The AES architecture of claim 12 , wherein said communicating FSMs comprise:
a main FSM; and
local FSMs which are controlled by said main FSM, 5 said local FSMs comprising:
a maximum parallel encryption circuit FSM which controls said maximum parallel circuit module;
a key scheduling circuit FSM which controls said key scheduling circuit;
an input buffer FSM which controls said asynchronous input buffer; and
an output buffer FSM which controls said asynchronous output buffer.
15. An advanced encryption system (AES) architecture which provides high throughput and low latency, comprising:
an asynchronous input buffer that receives data bytes to be encrypted, buffers a plurality of said data bytes in parallel, and provides said parallel data bytes at an output;
a maximum parallel encryption circuit that receives the output of said asynchronous input buffer and implements one round of the AES algorithm in one clock cycle;
a maximum parallel key scheduling circuit that generates sub-keys in one clock cycle in parallel with said maximum parallel encryption circuit;
an asynchronous output buffer that receives the output of said maximum parallel encryption circuit and outputs said encrypted data bytes to an output bus; and
a controller that controls the operation of said maximum parallel encryption and key scheduling circuits such that said AES architecture performs one round of the AES algorithm in one clock cycle;
wherein said controller is a hierarchical distributed control scheme comprising communicate finite state machines (FSMs);
wherein said AES architecture employs a chosen key-length k and a data-block length m, said maximum parallel key scheduling circuit comprising:
a first key scheduling sub-circuit that receives a previously-generated sub-key and generates a current sub-key;
a second key scheduling sub-circuit that receives the current sub-key and generates a next sub-key; and
a roundkey assembly sub-circuit which receives said previously-generated sub-key, said current sub-key, said next sub-key and which assembles a m-bit roundkey from said previously generated sub-key, said current sub-key, and said next sub-key, and provides said roundkey to said maximum parallel encryption circuit.
16. An advanced encryption system (AES) architecture which provides high throughput and low latency, comprising:
an asynchronous input buffer that receives data bytes to be encrypted, buffers a plurality of said data bytes in parallel, and provides said parallel data bytes at an output;
a maximum parallel encryption circuit that receives the output of said asynchronous input buffer and implements one round of the AES algorithm in one clock cycle;
a maximum parallel key scheduling circuit that generates sub-keys in one clock cycle in parallel with said maximum parallel encryption circuit;
an asynchronous output buffer that receives the output of said maximum parallel encryption circuit and outputs said encrypted data bytes to an output bus; and
a controller that controls the operation of said maximum parallel encryption and key scheduling circuits such that said AES architecture performs one round of the AES algorithm in one clock cycle;
wherein said controller is a hierarchical distributed control scheme comprising communicating finite state machines (FSMs);
wherein said AES architecture implements the AES standard with at least 128 bit data-blocks and a chosen key-length k, said key scheduling module comprising:
a key scheduling sub-circuit that receives the previously-generated sub-key and generates the current sub-key; and
a roundkey assembly sub-circuit which receives said previously-generated sub-key and said current sub-key and assembles an at least 128-bit roundkey from said previously-generated sub-key and said current sub-key, said at least 128-bit roundkey provided to said maximum parallel encryption circuit.
17. The AES architecture of claim 16 , wherein said AES architecture implements the a Rijndael algorithm with a data-blocks length of 128, 192 or 256 bits and a key-length of 128, 192 or 256 bits.
18. The AES architecture of claim 16 , wherein said AES architecture implements the AES standard with a data-block length of 128 bits and a key-length of 128, 192 or 256 bits.
19. The AES architecture of claim 16 , wherein said AES architecture implements the electronic code book (ECB) mode of operation.
20. The AES architecture of claim 16 , wherein said AES architecture implements a feedback mode of operation.
21. The AES architecture of claim 20 , wherein said AES architecture implements the Cipher Block Chaining (CBC) feedback mode of operation.
22. The AES architecture of claim 20 , wherein said AES architecture implements the Cipher Feedback (CFB) feedback mode of operation.
23. The AES architecture of claim 20 , wherein said AES architecture implements the Output Feedback (OFB) feedback mode of operation.