U.S. patent application number 14/279856 was filed with the patent office on 2015-02-26 for method and system for high throughput blockwise independent encryption/decryption.
This patent application is currently assigned to IP Reservoir, LLC. The applicant listed for this patent is IP Reservoir, LLC. Invention is credited to Roger D. Chamberlain, Ronald S. Indeck, David E. Taylor, Jason R. White.
Application Number | 20150055776 14/279856 |
Document ID | / |
Family ID | 38610289 |
Filed Date | 2015-02-26 |
United States Patent
Application |
20150055776 |
Kind Code |
A1 |
Taylor; David E. ; et
al. |
February 26, 2015 |
Method and System for High Throughput Blockwise Independent
Encryption/Decryption
Abstract
An encryption technique is disclosed for encrypting a plurality
of data blocks of a data segment where the encryption selectively
switches between a blockwise independent randomized (BIR)
encryption mode and a cipher block chaining (CBC) encryption mode
based on a configurable feedback stride. A corresponding decryption
technique is also disclosed.
Inventors: |
Taylor; David E.; (St.
Louis, MO) ; Indeck; Ronald S.; (St. Louis, MO)
; White; Jason R.; (Manchester, MO) ; Chamberlain;
Roger D.; (St. Louis, MO) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
IP Reservoir, LLC |
St. Louis |
MO |
US |
|
|
Assignee: |
IP Reservoir, LLC
St. Louis
MO
|
Family ID: |
38610289 |
Appl. No.: |
14/279856 |
Filed: |
May 16, 2014 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
13759227 |
Feb 5, 2013 |
8737606 |
|
|
14279856 |
|
|
|
|
11690034 |
Mar 22, 2007 |
8379841 |
|
|
13759227 |
|
|
|
|
60785821 |
Mar 23, 2006 |
|
|
|
Current U.S.
Class: |
380/28 |
Current CPC
Class: |
H04L 2209/125 20130101;
H04L 9/14 20130101; H04L 9/0662 20130101; H04L 2209/24 20130101;
H04L 9/0637 20130101 |
Class at
Publication: |
380/28 |
International
Class: |
H04L 9/06 20060101
H04L009/06; H04L 9/14 20060101 H04L009/14 |
Claims
1. An apparatus comprising: an encryption circuit for encrypting a
plurality of data blocks of a data segment, the encryption circuit
configured to selectively switch between a blockwise independent
randomized (BIR) encryption mode and a cipher block chaining (CBC)
encryption mode based on a configurable feedback stride; wherein
the encryption circuit comprises a combiner circuit, a block cipher
circuit, and a sequence generator circuit; wherein the combiner
circuit is configured to combine a plurality of bit vectors with a
plurality of data blocks of the data segment to generate a
plurality of bit vector data block combinations for encryption by
the block cipher circuit; the block cipher circuit comprising a
plurality m of pipelined stages for simultaneously processing, in a
pipelined fashion, a plurality m of different bit vector-data block
combinations corresponding to m data blocks of the data segment to
thereby generate a plurality of encrypted data blocks, the
encrypted data blocks comprising a first plurality of encrypted
data blocks for the data segment that were encrypted according to
the BIR encryption mode and a second plurality of encrypted data
blocks for the same data segment that were encrypted according to
the CBC encryption mode, the value for m being configurable to
define the feedback stride; and the sequence generator configured
to selectively switch between the BIR encryption mode and the CBC
encryption mode based on the feedback stride, wherein the sequence
generator is further configured to (1) generate and output a
plurality of randomized blockwise independent bit vectors for
delivery to the combiner circuit while in the BIR encryption mode,
and (2) generate and output a plurality of blockwise dependent bit
vectors for delivery to the combiner circuit while in the CBC
encryption mode, the blockwise dependent bit vectors being based on
a previously encrypted bit vector-data block combination fed back
from the block cipher circuit.
2. The apparatus of claim 1 wherein the combiner circuit comprises
XOR logic.
3. The apparatus of claim 1 wherein the encryption circuit is
resident on a reconfigurable logic device.
4. The apparatus of claim 1 wherein the encryption circuit is
resident on a field programmable gate array (FPGA).
5. The apparatus of claim 1 wherein the encryption circuit is
resident on a chip multi-processor (CMP).
6. The apparatus of claim 1 wherein the encryption circuit is
resident on an application-specific integrated circuit (ASIC).
7. A method for encrypting a plurality of data blocks of a data
segment, the method comprising: processing a first plurality j of
data blocks of a data segment in a blockwise independent randomized
(BIR) mode by (1) combining the first plurality j of the data
blocks with a plurality of blockwise independent bit vectors to
generate a first plurality j of data block-bit vector combinations,
and (2) encrypting the j data block-bit vector combinations through
a plurality of pipelined block cipher stages; processing a second
plurality k of data blocks of the same data segment in a cipher
block chaining (CBC) mode by (1) combining the second plurality k
of the data blocks with a plurality of blockwise dependent bit
vectors to generate a second plurality k of data block-bit vector
combinations, and (2) encrypting the k data block-bit vector
combinations through the pipelined block cipher stages; and
selectively switching between the BIR mode and the CBC mode by
generating the blockwise independent bit vectors and the blockwise
dependent bit vectors under control of a feedback stride that
defines values for j and k; and wherein the method steps are
performed by an encryption circuit, the encryption circuit thereby
generating the plurality of encrypted data blocks such that the
first j encrypted data blocks of the same data segment are
encrypted based on blockwise independent bit vectors and wherein k
of the other encrypted data blocks of the same data segment are
encrypted based on blockwise dependent bit vectors.
8. The method of claim 7 wherein the number of pipelined block
cipher stages is m, and wherein the value of j is greater than or
equal to m.
9. The method of claim 7 wherein the encryption circuit comprises a
combiner circuit, a block cipher circuit, and a sequence generator
circuit, wherein the combiner circuit performs the combining steps,
wherein the block cipher circuit performs the encrypting steps, and
wherein the sequence generator circuit performs the selectively
switching and bit vector generating steps.
10. The method of claim 7 further comprising the encryption circuit
processing the data blocks of the data segment in series such that
the first plurality j of data blocks are encrypted prior to
encrypting the second plurality k of data blocks.
11. The method of claim 10 wherein the generating step comprises:
for the first of the k data blocks, (1) feeding back the first of
the j encrypted data block bit vector combinations, and (2)
generating the blockwise dependent bit vector for combination with
the first of the k data blocks based on the fed back first of the j
encrypted data block bit vector combinations.
12. The method of claim 7 wherein the encryption circuit comprises
a reconfigurable logic device, the reconfigurable logic device
performing the method steps.
13. The method of claim 7 wherein the encryption circuit comprises
a field programmable gate array (FPGA), the FPGA performing the
method steps.
14. The method of claim 13 further comprising the FPGA repeating
the method steps for a plurality of different data segments that
are streaming through the FPGA.
15. The method of claim 7 wherein the encryption circuit is
resident on a chip multi-processor (CMP), the CMP performing the
method steps.
16. The method of claim 7 wherein the encryption circuit is
resident on an application-specific integrated circuit (ASIC), the
ASIC performing the method steps.
17. An apparatus comprising: an decryption circuit for decrypting a
plurality of encrypted data blocks of a data segment, the
decryption circuit configured to selectively switch between a
blockwise independent randomized (BIR) decryption mode and a cipher
block chaining (CBC) decryption mode based on a configurable
feedback stride; wherein the decryption circuit comprises a
combiner circuit, a block cipher circuit, and a sequence generator
circuit; wherein the combiner circuit is configured to combine a
plurality of bit vectors with a plurality of encrypted data blocks
of the data segment to generate a plurality of bit vector data
block combinations for decryption by the block cipher circuit; the
block cipher circuit comprising a plurality m of pipelined stages
for simultaneously processing, in a pipelined fashion, a plurality
m of different bit vector-data block combinations corresponding to
m data blocks of the data segment to thereby generate a plurality
of decrypted data blocks, the decrypted data blocks comprising a
first plurality of decrypted data blocks for the data segment that
were decrypted according to the BIR decryption mode and a second
plurality of decrypted data blocks for the same data segment that
were decrypted according to the CBC decryption mode, the value for
m being configurable to define the feedback stride; and the
sequence generator configured to selectively switch between the BIR
decryption mode and the CBC decryption mode based on the feedback
stride, wherein the sequence generator is further configured to (1)
generate and output a plurality of randomized blockwise independent
bit vectors for delivery to the combiner circuit while in the BIR
decryption mode, and (2) generate and output a plurality of
blockwise dependent bit vectors for delivery to the combiner
circuit while in the CBC decryption mode, the blockwise dependent
bit vectors being based on a previously decrypted bit vector-data
block combination fed back from the block cipher circuit.
18. A method for decrypting a plurality of encrypted data blocks of
a data segment, the method comprising: processing a first plurality
j of encrypted data blocks of a data segment in a blockwise
independent randomized (BIR) mode by (1) combining the first
plurality j of the encrypted data blocks with a plurality of
blockwise independent bit vectors to generate a first plurality j
of encrypted data block-bit vector combinations, and (2) decrypting
the j encrypted data block-bit vector combinations through a
plurality of pipelined block cipher stages; processing a second
plurality k of encrypted data blocks of the same data segment in a
cipher block chaining (CBC) mode by (1) combining the second
plurality k of the encrypted data blocks with a plurality of
blockwise dependent bit vectors to generate a second plurality k of
encrypted data block-bit vector combinations, and (2) decrypting
the k data block-bit vector combinations through the pipelined
block cipher stages; and selectively switching between the BIR mode
and the CBC mode by generating the blockwise independent bit
vectors and the blockwise dependent bit vectors under control of a
feedback stride that defines values for j and k; and wherein the
method steps are performed by a decryption circuit, the decryption
circuit thereby generating the plurality of decrypted data blocks
such that the first i decrypted data blocks of the same data
segment are decrypted based on blockwise independent bit vectors
and wherein k of the other decrypted data blocks of the same data
segment are decrypted based on blockwise dependent bit vectors.
19. The method of claim 18 wherein the number of pipelined block
cipher stages is m, and wherein the value of j is greater than or
equal to m.
20. The method of claim 18 further comprising: the decryption
circuit processing the encrypted data blocks of the data segment in
series such that the first plurality j of encrypted data blocks are
decrypted prior to decrypting the second plurality k of encrypted
data blocks; wherein the decryption circuit comprises a
reconfigurable logic device, the reconfigurable logic device
performing the method steps; and wherein the decryption circuit
comprises a combiner circuit, a block cipher circuit, and a
sequence generator circuit, each deployed on the reconfigurable
logic device, wherein the combiner circuit performs the combining
steps, wherein the block cipher circuit performs the decrypting
steps, and wherein the sequence generator circuit performs the
selectively switching and generating steps.
Description
CROSS-REFERENCE AND PRIORITY CLAIM TO RELATED PATENT
APPLICATIONS
[0001] This application is a continuation of pending U.S. patent
application Ser. No. 13/759,227, filed Feb. 5, 2013, and entitled
"Method and System for High Throughout Blockwise Independent
Encryption/Decryption", now U.S. Pat. No. ______, which is a
continuation of pending U.S. patent application Ser. No.
11/690,034, filed Mar. 22, 2007, and entitled "Method and System
for High Throughout Blockwise Independent Encryption/Decryption",
now U.S. Pat. No. 8,379,841, which claims priority to provisional
patent application 60/785,821, filed Mar. 23, 2006, and entitled
"Method and System for High Throughput Blockwise Independent
Encryption/Decryption", the entire disclosures of each of which are
incorporated herein by reference.
FIELD OF THE INVENTION
[0002] The present invention relates to improved techniques for
encrypting and decrypting data.
BACKGROUND AND SUMMARY OF THE INVENTION
[0003] The need for effective and efficient data
encryption/decryption is widespread throughout today's world.
Whether it be data maintained by a governmental agency that
pertains to national security or data maintained by a private
company that pertains to the company's trade secrets and/or
confidential information, the importance of effective and efficient
encryption/decryption cannot be understated.
[0004] Effective encryption/decryption is needed to preserve the
integrity of the subject data. Efficient encryption/decryption is
needed to prevent the act of encrypting/decrypting the subject data
from becoming an overwhelming burden on the party that maintains
the subject data. These needs exist in connection with both "data
at rest" (e.g., data stored in nonvolatile memory) and "data in
flight" (e.g., data in transit from one point to another such as
packet data transmitted over the Internet).
[0005] A number of data encryption/decryption techniques are known
in the art. Many of these encryption techniques utilize a block
cipher (see, e.g., block cipher 100 in FIG. 1). A block cipher is a
cryptographic mechanism that operates on fixed length blocks of
plaintext and produces fixed length blocks of ciphertext (see,
e.g., blocks 108, 110 and 112 in FIG. 1). Plaintext refers to data
needing encryption and ciphertext refers to data that has been
encrypted. A block cipher encrypts each plaintext block using a key
as per well-known key-based encryption algorithms (see, e.g., key
114 in FIG. 1). The key is typically (but need not be) the same
size as the plaintext block. Using different keys to encrypt the
same block of plaintext typically (but need not) produces different
blocks of ciphertext. Block ciphers 100 can operate on data blocks
of varying sizes, with typical data block sizes ranging between 64
bits and 512 bits. For example, the Advanced Encryption Standard
(AES) block cipher operates on blocks of 128 bits (16 bytes).
Encrypting large segments of plaintext requires a mode of
encryption operation that defines the flow of a sequence of
plaintext data blocks through one or more block ciphers. Likewise,
decrypting large segments of ciphertext requires a mode of
decryption operation that defines the flow of a sequence of
ciphertext data blocks through one or more block ciphers.
[0006] As an example of one such known mode of
encryption/decryption, the electronic codebook (ECB) mode of
encryption/decryption is commonly used due to its simplicity and
high data throughput. Examples of the ECB mode of
encryption/decryption are shown in FIG. 1. With the ECB mode, a
data segment needing encryption is divided into a plurality of data
blocks, each data block comprising a plurality of data bits (see
data blocks 102, 104 and 106 in FIG. 1). Each block cipher 100 then
encrypts each data block independently using key 114. At time
t=t.sub.0, plaintext data block 102 is encrypted by the block
cipher 100 using key 114 to produce ciphertext data block 108.
Subsequently, at time t=t.sub.1, plaintext data block 104 is
encrypted by the block cipher 100 using key 114 to produce
ciphertext data block 110. Then, at time t=t.sub.2, plaintext data
block 106 is encrypted by the block cipher 100 using key 114 to
produce ciphertext data block 112. To later decrypt the ciphertext
data blocks 108, 110 and 112, these steps can then be repeated to
reconstruct the original plaintext data blocks 102, 104, and 106.
It is worth noting that the same block cipher 100 can be used to
both encrypt and decrypt data using a key.
[0007] With ECB, the lack of sequential blockwise dependency in the
encryption/decryption (i.e., feedback loops where the encryption of
a given plaintext block depends on the result of encryption of a
previous plaintext data block) allows implementations of the ECB
mode to achieve high data throughput via pipelining and parallel
processing techniques. While ECB exhibits these favorable
performance characteristics, the security of ECB's encryption is
susceptible to penetration because of the propagation of
inter-segment and intra-segment uniformity in the plaintext to the
ciphertext blocks.
[0008] For example, a 256 bit segment of plaintext containing all
zeros that is to be encrypted with a 64 bit block cipher using ECB
will be broken down into 4 64-bit blocks of plaintext, each 64-bit
plaintext block containing all zeros. When operating on these
plaintext blocks, ECB will produce a segment of ciphertext
containing four identical blocks. This is an example of
intra-segment uniformity. Furthermore, if another such 256-bit all
zero segment is encrypted by ECB using the same key, then both of
the resulting ciphertext segments will be identical. This is an
example of inter-segment uniformity. In instances where
intra-segment and/or inter-segment uniformity is propagated through
to ciphertext, the security of the ciphertext can be compromised
because the ciphertext will still preserve some aspects of the
plaintext's structure. This can be a particularly acute problem for
applications such as image encryption.
[0009] To address intra-segment and inter-segment uniformity
issues, there are two commonly-used approaches. One approach is
known as cipher block chaining (CBC). An example of the CBC mode of
encryption/decryption is shown in FIG. 2. The CBC mode combines the
most recent ciphertext output from the block cipher with the next
input block of plaintext. The first plaintext block to be encrypted
is combined with an initialization vector that is a bit string
whose bits have random values, thereby providing the CBC mode with
inter-segment randomness.
[0010] As shown in FIG. 2, At time t=t.sub.0, the first plaintext
data block 102 is combined with a random initialization vector (IV)
200 using a reversible combinatorial operation 210, to thereby
create a block-vector combination. This block-vector combination is
then encrypted by block cipher 100 using key 114 to thereby
generate ciphertext block 202. Next, at time t=t.sub.1, the
ciphertext block 202 is fed back to be combined with the second
plaintext block 104 via XOR operation 210. The resultant
block-vector combination is key encrypted by block cipher 100 to
produce ciphertext block 204, which is in turn fed back for
combination with the next plaintext block at time t=t.sub.2 to
eventually produce ciphertext block 206. Thus, as can be seen, when
the CBC mode is used to encrypt a data segment comprising a
plurality of data blocks, the bit vectors that are used for the
reversible combinatorial operations with the plaintext data blocks
that follow the first plaintext data block are bit vectors that are
dependent upon the encryption operation(s) performed on each
previously encrypted plaintext data block.
[0011] Preferably, the reversible combinatorial operation 210 is an
XOR operation performed between the bits of the vector 200 and the
block 102. The truth table for an XOR operation between bits X and
Y to produce output Z is as follows:
##STR00001##
As is well known, the XOR operation is reversible in that either of
the inputs X or Y can be reconstructed by performing an XOR
operation between Z and the other of the inputs X or Y. That is, if
one XORs X with Y, the result will be Z. If one thereafter XORs Z
with Y, then X will be reconstructed. Similarly, if one thereafter
XORs Z with X, then Y will be reconstructed.
[0012] Thus, on the decryption side, the CBC mode operates to
decrypt ciphertext block 202 with the cipher block 100 using key
114 to thereby reconstruct the XOR combination of plaintext data
block 102 and the initialization vector 200. Thereafter, this
reconstructed combination can be XORed with the initialization
vector 200 to reconstruct plaintext block 102. Next, at time
t=t.sub.1, the process is repeated for the next ciphertext block
204, although this time the XOR operation will be performed using
ciphertext block 202 (rather than initialization vector 200) to
reconstruct plaintext data block 104.
[0013] Ciphertext block 202 is used in this XOR operation because
it was ciphertext block 202 that was used in the XOR operation when
plaintext block 104 was encrypted. Then, once again this process is
repeated at time t=t.sub.2, albeit with ciphertext block 204 being
used for the XOR combination operation with the output from cipher
block 100.
[0014] While the use of feedback by the CBC mode addresses the
issue of inter-segment and intra-segment uniformity, such feedback
imposes a sequential processing flow on the encryption that
significantly limits the achievable throughput of the encryption
engine. As such, the CBC mode cannot make ready use of pipelining
because one of the inputs for the reversible combinatorial
operation stage 210 of the encryption for a given data block
depends upon the output of the cipher block stage 100 of the
encryption performed on the previous data block. That is, because
of the feedback, the reversible combinatorial operation stage in a
CBC encryption engine must wait for the block cipher to complete
its encryption of a given data block-bit vector combination before
it can begin to process the next data block.
[0015] Furthermore, on the decryption side, the CBC mode's
dependence on the sequential order of data block encryption can
raise problems when one wants to retrieve only a portion of the
encrypted data segment. For example, for a data segment that
comprises data blocks DB.sub.1 through DB.sub.20, when that data
segment is encrypted and stored for subsequent retrieval in its
encrypted form, an instance may arise where there is a need to
retrieve data blocks DB.sub.6 through DB.sub.10, wherein the other
data blocks of the data segment are not needed. However, to be able
to successfully decrypt data blocks DB.sub.6 through DB.sub.10, the
retrieval operation and decryption operation will nevertheless need
to operate on data blocks DB.sub.1 through DB.sub.5 so that
decryption can be performed for data blocks DB.sub.6 through
DB.sub.10.
[0016] Furthermore, when used for disk encryption, the CBC mode may
be vulnerable to a "watermark attack" if the initialization vector
200 is not kept secret (such as may be the case when the
initialization vector is derived from a quantity such as a disk
volume number). With such an attack, an adversary can determine
from the output ciphertext whether or not a specially crafted file
is stored. While there are solutions to such an attack (such as
using hashing to derive the initialization vector from the data
blocks in the sector), these solutions add to the computational
complexity of the encryption operation and thus further degrade the
throughput and/or increase the computational resources required for
the encryption.
[0017] A second approach is known as the Segmented Integer Counter
(SIC) mode, or more succinctly the counter (CTR) mode. FIG. 3
depicts an example of the SIC/CTR mode of encryption/decryption.
The SIC/CTR mode key encrypts a block comprising a combination of a
random value (or nonce) and a counter value. This random
value-counter combination can be achieved in any of a variety of
ways (e.g., concatenation, XOR, etc.) The counter values may be any
sequence of values that do not repeat over a long duration, but a
simple incremental counter is believed to be the most commonly-used
approach. The output of the block cipher 100 is then combined with
the plaintext block using a reversible combinatorial operation 210
(e.g., XOR), with the output of the operation 210 being the
ciphertext block. The SIC/CTR mode belongs to the general class of
encryption modes known as a stream cipher.
[0018] As shown in FIG. 3, at time t=t.sub.0, the random value 300
is combined with a counter value 308 in some manner to create a
random value-counter combination block 302. This block 302 is then
encrypted by block cipher 100 using key 114, and the output
therefrom is then XORed with plaintext block 102 to generate
ciphertext block 322. Next, at time t=t.sub.1, the random value 300
is combined with a next counter value 310 in some manner to create
the random value-counter combination block 304. This block 304 is
then encrypted by block cipher 100 using key 114, and the output
therefrom is then XORed with plaintext block 104 to generate
ciphertext block 324. Finally, at time t=t.sub.2, the random value
300 is combined with a next counter value 312 in some manner to
create the random value-counter combination block 306. This block
306 is then encrypted by block cipher 100 using key 114, and the
output therefrom is then XORed with plaintext block 106 to generate
ciphertext block 326.
[0019] On the decryption side, this process can then be reversed
where the combination blocks 302, 304 and 306 are decrypted by
block cipher 100 using key 114, with the respective outputs
therefrom being XORed with the ciphertext blocks 322, 324 and 326
respectively to reconstruct plaintext blocks 102, 104 and 106.
[0020] The SIC/CTR mode of encryption/decryption also suffers from
a security issue if data segments are always encrypted with the
same random value 300. If an adversary is able to gather several
versions of the encrypted data segment, it would be possible to
derive information about the plaintext because the cipher text (C)
is simply the XOR of the variable (V) based on the random number
and the plaintext (P), e.g., C=P.sym.V, thus C.sym.C'=P.sym.P'.
[0021] Therefore, the inventors herein believe that a need exists
in the art for a robust encryption/decryption technique that is
capable of reducing both inter-segment and intra-segment uniformity
while still retaining high throughput and exhibiting blockwise
independence. As used herein, an encryption operation for a data
segment is said to be "blockwise independent" when the encryption
operations for each data block of that data segment do not rely on
the encryption operation for any of the other data blocks in that
data segment. Likewise, a decryption operation for a data segment
is said to be "blockwise independent" when the decryption
operations for each encrypted data block of that data segment do
not rely on the decryption operation for any of the other data
blocks in that data segment.
[0022] Toward this end, in one embodiment, the inventors herein
disclose a technique for encryption wherein prior to key
encryption, the plaintext data block is combined with a blockwise
independent bit vector using a reversible combinatorial operation
to thereby create a plaintext block-vector combination. This
plaintext block-vector combination is then key encrypted to
generate a ciphertext block. This process is repeated for all data
blocks of a data segment needing encryption. For decryption of the
cipher text blocks produced by such encryption, the inventors
herein further disclose an embodiment wherein each ciphertext data
block is key decrypted to reconstruct each plaintext block-vector
combination. These reconstructed plaintext block-vector
combinations can then be combined (using the reversible
combinatorial operation) with the corresponding randomized bit
vectors that were used for encryption to thereby reconstruct the
plaintext blocks.
[0023] As an improvement relative to the CBC mode of
encryption/decryption, each bit vector is blockwise independent. A
bit vector is said to be blockwise independent when the value of
that bit vector does not depend on any results of an
encryption/decryption operation that was performed on a different
data block of the data segment. Because of this blockwise
independence, this embodiment is amenable to implementations that
take advantage of the power of pipelined processing and/or parallel
processing.
[0024] Moreover, because of the blockwise independent nature of the
encryption performed by the present invention, a subset of the
encrypted data segment can be decrypted without requiring
decryption of the entire data segment (or at least without
requiring decryption of the encrypted data blocks of the data
segment that were encrypted prior to the encrypted data blocks
within the subset). Thus, for a data segment that comprises data
blocks DB.sub.1 through DB.sub.20, when that data segment is
encrypted and stored for subsequent retrieval in its encrypted form
using the present invention, a need may arise to retrieve plaintext
versions of encrypted data blocks DB.sub.6 through DB.sub.10 and
DB.sub.15, wherein the other data blocks of the data segment are
not needed in their plaintext forms. A preferred embodiment of the
present invention supports successful decryption of a subset of
data blocks within the encrypted data segment (e.g., data blocks
DB.sub.6 through DB.sub.10 and DB.sub.15) without requiring the
decryption of the data segment's data blocks that are not members
of the subset (e.g., data blocks DB.sub.1 through DB.sub.6, data
blocks DB.sub.11 through DB.sub.14 and data blocks DB.sub.16
through DB.sub.20). Accordingly, the present invention supports the
decryption of any arbitrary subset of the encrypted data blocks of
a data segment without requiring decryption of any data blocks that
are non-members of the arbitrary subset even if those non-member
data blocks were encrypted prior to the encryption of the data
blocks within the arbitrary subset.
[0025] Similarly, even if an entire encrypted data segment is to be
decrypted, the present invention supports the decryption of the
encrypted data blocks in a block order independent manner. Further
still, the present invention supports the encryption of data blocks
in a block order independent manner as well as supports limiting
the encryption to only a defined subset of a data segment's data
blocks (wherein such a subset can be any arbitrary subset of the
data segment's data blocks).
[0026] Furthermore, as an improvement relative to the SIC/CTR mode
of encryption/decryption, a greater degree of security is provided
by this embodiment because the data that is subjected to key
encryption includes the plaintext data (whereas the SIC/CTR mode
does not subject the plaintext data to key encryption and instead
subjects only its randomized bit vector to key encryption).
[0027] Preferably, the blockwise independent bit vector is a
blockwise independent randomized (BIR) bit vector. As is understood
by those having ordinary skill in the art, randomization in this
context refers to reproducible randomization in that the same
randomized bit vectors can be reproduced by a bit vector sequence
generator given the same inputs. Further still, the blockwise
independent randomized bit vector is preferably generated from a
data tag that is associated with the data segment needing
encryption/decryption. Preferably, this data tag uniquely
identifies the data segment. In a disk encryption/decryption
embodiment, this data tag is preferably the logical block address
(LBA) for the data segment. However, it should be noted that
virtually any unique identifier that can be associated with a data
segment can be used as the data tag for that data segment. It
should also be noted that rather than using a single data tag
associated with the data segment, it is also possible to use a
plurality of data tags that are associated with the data segment,
wherein each data tag uniquely identifies a different one of the
data segment's constituent data blocks
[0028] A bit vector generation operation preferably operates on a
data tag to generate a sequence of blockwise independent bit
vectors, each blockwise independent bit vector for reversible
combination with a corresponding data block. Disclosed herein are a
plurality of embodiments for such a bit vector generation
operation. As examples, bit vectors can be derived from the
pseudo-random outputs of a pseudo-random number generator that has
been seeded with the data tag; including derivations that employ
some form of feedback to enhance the randomness of the bit vectors.
Also, linear feedback shift registers and adders can be employed to
derive the bit vectors from the data tag in a blockwise independent
manner.
[0029] The inventors also disclose a symmetrical embodiment of the
invention wherein the same sequence of operations are performed on
data in both encryption and decryption modes.
[0030] One exemplary application for the present invention is to
secure data at rest in non-volatile storage; including the storage
of data placed on tape, magnetic and optical disks, and redundant
array of independent disks (RAID) systems. However, it should be
noted that the present invention can also be applied to data in
flight such as network data traffic.
[0031] These and other features and advantages of the present
invention will be apparent to those having ordinary skill in the
art upon review of the following description and figures.
BRIEF DESCRIPTION OF THE DRAWINGS
[0032] FIG. 1 depicts an example of a known ECB mode of
encryption/decryption;
[0033] FIG. 2 depicts an example of a known CBC mode of
encryption/decryption;
[0034] FIG. 3 depicts an example of a known SIC/CTR mode of
encryption/decryption;
[0035] FIG. 4 depicts an exemplary data segment;
[0036] FIGS. 5(a) and (b) depict an embodiment of the present
invention in both encryption and decryption modes;
[0037] FIG. 6 depicts an exemplary bit vector sequence
generator;
[0038] FIGS. 7(a) and (b) depict exemplary encryption and
decryption embodiments of the present invention;
[0039] FIGS. 8(a) and (b) depict exemplary encryption and
decryption embodiments of the present invention showing their
operations over time;
[0040] FIG. 9 depicts an exemplary embodiment of a bit vector
sequence generator;
[0041] FIGS. 10(a)-(c) depict three additional exemplary
embodiments of a bit vector sequence generator;
[0042] FIG. 11 depicts an exemplary embodiment of the present
invention where multiple block ciphers are chained together;
[0043] FIGS. 12(a) and (b) depict exemplary encryption and
decryption embodiments of the present invention that are hybrids of
the embodiments of FIGS. 8(a) and (b) and the CBC mode of
encryption/decryption;
[0044] FIGS. 12(c) and (d) depict exemplary embodiments of the bit
vector sequence generator for use with the hybrid embodiments of
FIGS. 12(a) and (b);
[0045] FIGS. 13(a) and (b) depict an exemplary embodiment for
symmetrical encryption/decryption in accordance with the present
invention;
[0046] FIGS. 14(a) and (b) depict an exemplary embodiment for
symmetrical encryption/decryption in accordance with the present
invention wherein the blockwise independent bit vectors are derived
from the data segment's LBA;
[0047] FIGS. 15(a) and (b) depict the embodiment of FIGS. 14(a) and
(b) showing its operation over time;
[0048] FIGS. 15(c) and (d) depict a symmetrical
encryption/decryption counterpart to the embodiments of FIGS. 12(a)
and (b);
[0049] FIG. 16 depicts a parallel architecture for encrypting or
decrypting data blocks;
[0050] FIGS. 17(a) and (b) depict exemplary hardware environments
for the present invention; and
[0051] FIGS. 18(a)-(c) depict exemplary printed circuit boards on
which the encryption/decryption embodiments of the present
invention can be deployed.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0052] FIG. 4 illustrates an exemplary data segment 400 on which
the encryption/decryption technique of the present invention can be
performed. The data segment 400 comprises a plurality of data
blocks 102, 104, 106, . . . . Each data block comprises a plurality
of data bits and preferably has a fixed length (e.g., 64 bits, 256
bits, etc.). In an exemplary embodiment, wherein AES block ciphers
are used, which as explained above operate on 16-byte data blocks,
it is preferred that the data blocks 102, 104, 106 possess a length
of 16 bytes. It should also be noted that the size of the data
segment 400 is typically much larger than the size of an individual
data block. For example, a data storage system may operate on
"logical blocks" of data having a size of 512 bytes. In such a
case, the "logical block", which can serve as the data segment 400,
will be a collection of 32 16-byte data blocks.
[0053] FIG. 5(a) illustrates an embodiment of the present invention
wherein the encryption operation is segmented into a plurality of
stages. At stage 504, the blockwise independent bit vector 506 is
generated, preferably from a data tag 502 that is associated with
the data segment 400. Preferably, the bit vector 506 has a length
that is the same as the data blocks of the data segment, although
this need not be the case. Further still, it is preferred that the
blockwise independent bit vector 506 have a randomized value to
thereby enhance the security of the encryption. Also, it is
preferred that a different bit vector 506 be generated for each
data block of a data segment that is encrypted, although this need
not be the case. The bit vectors that are used in the encryption of
a data segment's data blocks should be either stored for subsequent
use when it is time to decrypt one or more of the data segment's
data blocks or should be reproducible from a known quantity (such
as the data tag) when it is time to decrypt one or more of the data
segment's data blocks.
[0054] At stage 210, a reversible combinatorial operation such as a
bitwise XOR operation is performed on the blockwise independent bit
vector 506 and plaintext data block. This reversible combinatorial
operation preferably produces a data block-bit vector combination
508.
[0055] At stage 100, a block cipher performs an encryption
operation on the data block-bit vector combination 508 using key
114 as per well-known key encryption techniques (e.g., AES, the
Data Encryption Standard (DES), the triple DES (3DES), etc.). The
output of the block cipher stage 100 is thus a ciphertext data
block that serves as the encrypted counterpart to the plaintext
data block that was fed into stage 210. It should be noted that any
of several well-known key management techniques can be used in
connection with managing the key(s) 114 used by the block cipher(s)
100. As such, the inventors do not consider the key management for
the block cipher(s) 100 to be any limitation on the present
invention. It should also be noted that "keyless" encryption
techniques may also be used in the practice of the present
invention (e.g., substitution ciphers that do not require a
key).
[0056] FIG. 5(b) depicts the decryption counterpart to FIG. 5(a).
In FIG. 5(a), the flow of data blocks and stages is reversed such
that the ciphertext data block is first key decrypted by stage 100
to reconstruct combination 508. Combination 508 is in turn combined
with the same bit vector 506 that was used when creating that
ciphertext data block and using the same reversible combinatorial
operation 210 that was used when creating that ciphertext data
block, to thereby reconstruct the plaintext data block.
[0057] As can be seen in FIGS. 5(a) and (b), no feedback is
required between stages, thus allowing this encryption/decryption
technique to be implemented in a pipelined architecture and/or a
parallel processing architecture for the achievement of a high
throughput when performing encryption/decryption. Thus, as a stream
of data blocks are sequentially processed through the
encryption/decryption stages, a high throughput can be maintained
because the reversible combinatorial stage 210 can operate on a
given data block while the block cipher stage 100 simultaneously
operates on a different data block because the reversible
combinatorial operation stage 210 does not require feedback from
the block cipher stage 100 to operate.
[0058] The data tag 502 may be any data value(s) that can be
associated with the data segment 400. Preferably, the data tag 502
serves as a unique identifier for the data segment 400, although
this need not be the case. A preferred data tag 502 is the logical
block address (LBA) for the data segment to be encrypted. An LBA
for a data segment is the logical memory address for the data
segment that is typically assigned by an Operating System (OS) or
memory management system. However, other data tags may be used in
the practice of the present invention; examples of which include
file identifiers, physical memory addresses, and packet sequence
numbers. The source of the data tag can be any of a variety of
sources, including but not limited to communication protocol,
storage subsystem, and file management systems.
[0059] FIG. 6 illustrates how a sequence of bit vectors 506 can be
generated from a data tag 502. As an exemplary embodiment of bit
vector generation stage 504, bit vector sequence generator 600
preferably operates to produce a plurality of blockwise independent
randomized bit vectors 506.sub.i from an input comprising data tag
502. FIGS. 9 and 10, to be described hereinafter, illustrate
various exemplary embodiments for the bit vector sequence generator
600.
[0060] FIG. 7(a) and (b) illustrate embodiments of the invention
where the data segment's LBA is used as the data tag 502 for the
encryption/decryption operations. Sequence generator 600 processes
the LBA to produce a different blockwise independent randomized bit
vector 506 for XOR combination (210) with each plaintext data
block. On decryption (shown in FIG. 7(b)), the sequence generator
600 operates to produce the same plurality of different bit vectors
506 from the data segment's LBA as were produced by the sequence
generator 600 for encryption (see FIG. 7(a)) given the same LBA
input. Thus, as shown in FIG. 7(b), each bit vector 506 is then
used for XOR combination (210) with each decrypted ciphertext
block.
[0061] FIG. 8(a) illustrates the embodiment of FIG. 7(a) (wherein
the LBA is labeled as an initialization vector), but depicting how
the encryption operation can proceed over time. Thus, at time
t=t.sub.0, plaintext data block 102 is reversibly combined with bit
vector 506.sub.1 produced by sequence generator 600 to generate a
data block-bit vector combination that is key encrypted by a block
cipher 100 to thereby produce an encrypted data block-bit vector
combination 802 which serves as the ciphertext block. Subsequently,
at time t=t.sub.1, the sequence generator produces another bit
vector 506.sub.2 for reversible combination with plaintext data
block 104. The resultant data block-bit vector combination is then
key encrypted by the block cipher 100 to thereby produce an
encrypted data block-bit vector combination 804 which serves as the
next ciphertext block. This process then continues for subsequent
clock cycles as additional data blocks of the data segment 400 are
encrypted.
[0062] FIG. 8(b) depicts the decryption counterpart to FIG. 8(a),
wherein ciphertext blocks 802, 804 and 806 are decrypted in
accordance with the embodiment of FIG. 7(b) to reproduce plaintext
data blocks 102, 104 and 106.
[0063] FIG. 9 depicts an embodiment of the sequence generator 600
wherein a data tag 502 such as the LBA is used to seed a
pseudo-random number generator (PRNG) 900. When encrypting a first
data block, the bit vector 506 is initialized to be the LBA itself.
Then, when encrypting subsequent data blocks, the bit vector 506 is
incremented through adder 902 by the pseudo-random output from the
PRNG 900. Preferably, a new pseudo-random number is generated by
the PRNG 900 for each new data block of the data segment needing
encryption. By using a PRNG 900 to generate counter increments for
the bit vector 506, the sequence of bit vectors 506 used for
encrypting different data segments (identified by their LBA) will
be difficult to predict and provide more security than a simple
counter. For decryption, it should be noted that the PRNG 900
should operate to produce the same sequence of pseudo-random
outputs given the same data tag input, to thereby enable the
generation of the same set of bit vectors 506 when decrypting the
encrypted data segment (or a subset of the encrypted data
segment).
[0064] As can be seen, the sequence of bit vectors 506.sub.1,
506.sub.2, . . . 506.sub.n produced by the sequence generator 600
of FIG. 9 will be sequentially dependent in that each successive
bit vector 506.sub.i will be a function of the previous bit vector
506.sub.i-1 (via feedback to adder 902). This sequential nature of
the bit vectors does not preclude their use in a blockwise
independent encryption/decryption scheme. For example, consider a
case where a data tag (such as an LBA) for a data segment
comprising twenty data blocks is used as the basis for the
blockwise independent bit vectors, but it is only desired to
encrypt/decrypt data blocks DB.sub.6 through DB.sub.10. In such a
case, the sequence generator 600 is preferably initialized with the
data tag and the bit vectors for data blocks DB.sub.1 through
DB.sub.5 are generated but discarded by the sequence generator 600.
Such a configuration will require the reversible combinatorial
stage 210 and the downstream encryption stage 100 to pause until
the bit vector 506.sub.6 for data block DB.sub.6 is generated.
While this pause produces a delay and degradation in throughput for
the encryption/decryption technique, relative to the multiple
iterations through a block cipher as required in the conventional
CBC mode of encryption, the inventors herein believe that this
delay and throughput degradation is relatively minor. For example,
this pause will not need to wait for data blocks DB.sub.1 through
DB.sub.5 to be encrypted/decrypted via block cipher 100 before
being able to process data block DB.sub.6.
[0065] It should also be noted that if the encryption/decryption
technique involves using a data tag that is unique to each data
block to generate each data block's corresponding blockwise
independent bit vector 506, the need to pause operations while
cycling through unneeded bit vectors can be eliminated.
[0066] FIGS. 10(a)-(c) depict other examples of sequence generator
embodiments. FIG. 10(a) discloses a sequence generator 600 that
uses the LBA 502 to seed a PRNG 900 whose pseudo-random outputs
then serve as the bit vectors 506 for combination with the data
segment's data blocks. As with the embodiment of FIG. 9, preferably
the LBA itself is used as the bit vector 506 for reversible
combination with a first data block to be encrypted/decrypted.
[0067] FIG. 10(b) discloses a sequence generator 600 that uses the
LBA 502 to seed a linear feedback shift register (LFSR) 1000 whose
outputs then serve as the bit vectors 506 for combination with the
data segment's data blocks.
[0068] FIG. 10(c) discloses a sequence generator 600 that uses the
LBA 502 to seed a feedback counter 1002, wherein the feedback
counter 1002 has a constant increment 1004, and wherein the
counter's outputs then serve as the bit vectors 506 for combination
with the data segment's data blocks. As with the embodiments of
FIG. 9 and FIG. 10(a), preferably the LBA itself is used as the bit
vector 506 for reversible combination with a first data block to be
encrypted/decrypted. It should be noted that the sequence generator
embodiment of FIG. 10(c) can be configured to accommodate
encryption/decryption of arbitrary subsets of data blocks within a
data segment without requiring a pause while the sequence generator
cycles through unneeded bit vectors. If an encryption/decryption is
to begin at a data block within a data segment that is not the
first data block of the data segment (e.g., data block DB.sub.k of
a data segment, wherein k>1), the data tag 502 (such as an LBA)
that is passed to the sequence generator 600 can be computed
as:
Data Tag'=Data Tag+k*Constant
wherein Data Tag' represents the value of the data tag 502 that is
fed into the sequence generator 600, wherein Data Tag represents
the value of the data tag that is associated with the data segment,
wherein k represents the block number within the data segment of
the data block to be encrypted/decrypted, and wherein Constant
represents the value of the incremental constant 1004 for adder
1002. This computation can be performed either within the sequence
generator (in which case it will be the value Data Tag that is fed
into the sequence generator 600) or in a module upstream from the
sequence generator. Appropriate control logic is preferably used to
control whether the multiplexer passes the data tag value 502 or
the output of adder 1002 on to the reversible combinatorial stage
210.
[0069] It should also be noted that the present invention need not
be limited to a single combination of a blockwise independent bit
vector randomizer and a block cipher. Pairs of sequence generators
600, reversible combinatorial operations 210, and block ciphers 100
can be sequentially chained as shown in FIG. 11. Thus, a first
sequence generator 600.sub.1, a first reversible combinatorial
operator 210.sub.1 and a first block cipher 100.sub.1 can operate
to produce an encrypted data block-bit vector combination that is
fed into a second reversible combinatorial operator 210.sub.2 for
reversible combination with a bit vector produced by a second
sequence generator 600.sub.2. The resultant encrypted data
block-bit vector-bit vector combination produced by reversible
combinatorial operator 210.sub.2 can then be key encrypted by block
cipher 100.sub.2. The inventors herein believe that such chaining
can enhance the security of an encryption system. Moreover, the
inventors note that still greater numbers of sequence generators
600, reversible combinatorial operations 210, and block ciphers 100
can be sequentially chained to enhance security if desired by a
practitioner of this embodiment of the invention. It should also
readily be understood that corresponding sequential decryption
chains can be used. Preferably, in such a sequential chaining
embodiment, each different sequence generator 600.sub.i will
operate to produce different set of bit vectors given the same
input.
[0070] Further still, the inventors herein disclose an embodiment
that hybridizes the present invention and the CBC mode of
encryption/decryption. FIG. 12(a) illustrates an example of such an
embodiment to perform encryption. This configuration provides the
flexibility to include some feedback for higher security. Note that
the first output of ciphertext 1200 is not used as feedback to the
second encryption operation 1202, rather it is used as feedback for
encryption performed by subsequent block i where i is a feedback
stride. The feedback stride can then be chosen to provide a
favorable balance among security and throughput. If the feedback
stride i is greater than or equal to the number of pipeline stages
in the block cipher, then there is no performance penalty because
there will need not be a delay in the insertion of a block into a
block cipher. Furthermore, if one does choose a lower feedback
stride value that would require a delay, one can introduce stall
cycles in the processing. The added security provided by the
technique of FIG. 12(a) is that the encryption technique of FIG.
12(a) does not exclusively rely on the sequence generator 600 (or
the PRNG 900 in the sequence generator 600) to generate long,
difficult to predict initialization sequences. Once the system
begins feeding back ciphertext from previous blocks via feedback
link 1206, the system gains the strength of the block cipher in
producing more random initialization bit vectors. This technique
essentially narrows the visibility of an observer into the "window"
of the random increments produced by the PRNG 900. Thus, it is more
difficult for observers to reconstruct the entire random sequence
generated by the PRNG 900 (thereby making it more difficult for one
to crack the encryption scheme). FIG. 12(b) depicts a decryption
counterpart to FIG. 12(a).
[0071] FIGS. 12(c) and (d) depict exemplary embodiments of a
sequence generator 600' that could be used to generate bit vectors
for the embodiments of FIGS. 12(a) and (b). In the example of FIG.
12(c), the sequence generator 600' comprises any of the embodiments
for sequence generator 600 as described in connection with FIGS. 9
and 10(a)-(c). The bit vector 506 that is output by the sequence
generator 600 is preferably reversibly combined with the feedback
ciphertext i from link 1206 via reversible combinatorial operator
1250 to produce bit vector 506' (which is in turn provided to the
reversible combinatorial operator 210) when the conditions for the
feedback stride i are met. Sequence generator 600' also preferably
includes appropriate control logic to ensure that the feedback
stride defined for the hybrid embodiment is obeyed. As an example,
such control can be achieved with a multiplexer 1262 whose inputs
are either null value or the feedback ciphertext i. A counter-based
control circuit 1260 can define which of the inputs to multiplexer
1262 are passed to the reversible combinatorial operator 1250 such
that the feedback ciphertext i is only passed on when it is time to
use the ciphertext to further randomize the bit vectors.
[0072] FIG. 12(d) depicts another exemplary embodiment for the
sequence generator 600'. In the example of FIG. 12(d), the sequence
generator 600' comprises any of the embodiments for sequence
generator 600 as described in connection with FIGS. 9 and
10(a)-(c). The sequence generator 600 will receive as an input
either the data tag 502 or the feedback ciphertext i, as defined by
control logic. The control logic is preferably configured to pass
on the feedback ciphertext to seed the sequence generator 600 only
when the conditions for the feedback stride i are met. As an
example, such control can be achieved with a multiplexer 1262 whose
inputs are either the data tag 502 or the feedback ciphertext i. A
counter-based control circuit 1260 can define which of the inputs
to multiplexer 1262 are passed to the sequence generator 600 such
that the feedback ciphertext i is only passed on when it is time to
use the ciphertext to further randomize the bit vectors.
[0073] As another embodiment of the present invention, the
inventors disclose a symmetrical embodiment for
encryption/decryption. With "symmetrical" encryption/decryption,
the same order of operations can be performed on data blocks to
both encrypt and decrypt those data blocks. Thus, with a
symmetrical embodiment, the same module that is used to encrypt
data can be used to decrypt encrypted data. FIGS. 13(a) and (b)
illustrate a symmetrical embodiment of the present invention. As
can be seen, the same order of operations is used by FIG. 13(a) to
encrypt a data block as is used by FIG. 13(b) to decrypt a
ciphertext data block. The symmetrical encryption/decryption engine
1300 comprises a first reversible combinatorial stage 210, a block
cipher operation stage 100, and a second reversible combinatorial
stage 1302. A bit vector generation stage 504 (such as the sequence
generators 600 shown in FIG. 9 and FIGS. 10(a)-(c)) operates to
produce blockwise independent bit vectors 506 that are fed to both
the first reversible combinatorial stage 210 and the second
reversible combinatorial stage 1302.
[0074] As shown in FIG. 13(a), for encryption, a plaintext data
block is reversibly combined with a blockwise independent bit
vector 506 by first reversible combinatorial operation stage 210
(preferably XOR logic), to thereby generate a data block-bit vector
combination 508. Block cipher 100 then performs a block cipher
operation on this data block-bit vector combination 508 using a
key. The resultant block ciphered data block-bit vector combination
1304 is then reversibly combined with a blockwise independent bit
vector 506 by second reversible combinatorial operation stage 1302
(preferably XOR logic), to thereby generate a block ciphered data
block-bit vector-bit vector combination 1306, which can serve as
the ciphertext for the plaintext data block.
[0075] For decryption, as shown in FIG. 13(b), the same order of
operations is used, albeit starting from a ciphertext data block
rather than a plaintext data block. The ciphertext data block used
for decryption will be a block ciphered data block-bit vector-bit
vector combination 1306 that was produced during the encryption
operation. First reversible combinatorial operation stage 210
operates to reversibly combine such a ciphertext data block with
the same bit vector 506 that was used by the second reversible
combinatorial operation stage 1302 when encrypting that ciphertext
data block. The result of this reversible combination will be a
reconstruction of the block ciphered data block-bit vector
combination 1304. Block cipher 100 then performs a block cipher
operation (decryption in this example) using the key to reconstruct
the data block-bit vector combination 508. Second reversible
combinatorial operation stage 210 then operates to reversibly
combine the reconstructed data block-bit vector combination 508
with the same bit vector 506 that was used by the first reversible
combinatorial operation stage 210 when encrypting that ciphertext
data block. The output of the second reversible combinatorial
operation stage 1302 then serves as a reconstruction of the
plaintext data block.
[0076] Timing logic (not shown) can be employed to synchronize the
outputs of bit vectors 506 from the bit vector generation stage 504
such that the appropriate bit vector 506 is fed to the second
reversible combinatorial stage 1302 for each block ciphered data
block-bit vector combination 1304 (or reconstructed data block-bit
vector combination 508 for the decryption mode) that is processed
thereby. Such synchronization could be designed to accommodate the
latency within the block cipher 100 to thereby allow the same bit
vector 506 to be used for reversible combination with a given data
block by first reversible combinatorial operation stage 210 as is
used for later reversible combination with the block ciphered data
block-bit vector combination 1304 derived from that given data
block by the second reversible combinatorial operation stage
1302.
[0077] FIG. 14(a) (for encryption mode) and FIG. 14(b) (for
decryption mode) depict an example of the symmetrical embodiment of
FIGS. 13(a) and (b), wherein the bit vectors 506 are derived from
the LBA for the data segment 400.
[0078] FIG. 15(a) (for encryption mode) and FIG. 15(b) (for
decryption mode) depict the operation of the embodiment of FIGS.
14(a) and (b) over time.
[0079] It should also be noted that the symmetrical
encryption/decryption embodiments described herein can also be used
in a hybrid CBC embodiment like the ones shown in FIGS. 12(a) and
(b). An example of such a symmetrical hybrid embodiment is shown in
FIGS. 15(c) and (d), wherein the feedback link 1502 carries the
block ciphered data block-bit vector-bit vector output 1306 of the
second reversible combinatorial operation stage 1302 performed for
the first data block. The sequence generators 600' as shown in
FIGS. 12(c) and (d) can be employed, although the feedback
ciphertext will preferably emanate from the output of the second
reversible combinatorial operator 1302 rather than the output of
the block cipher 100.
[0080] As a further embodiment of the present invention, the
inventors note that a parallel architecture 1600 such as the one
shown in FIG. 16 can be employed. With this parallel architecture,
a stream of incoming data blocks 1604 (which can be either
plaintext data blocks or ciphertext data blocks) are separated into
a plurality of parallel streams for processing by parallel
encryption/decryption engines 1602. Such encryption/decryption
engines can take the form of any of the embodiments of the
invention described herein such as those shown in connection with
FIGS. 5(a) and (b), 7(a) and (b), 11, 12(a) and (b), 13(a) and (b),
and 14(a) and (b). The resultant data streams produced by each
parallel encryption/decryption engine 1602 can then be brought
together to form the outgoing data stream 1606 (which may be either
plaintext data blocks or ciphertext data blocks depending on
whether the encryption/decryption engines 1602 performed encryption
or decryption). It is also worth noting that each parallel engine
1602 can employ its own bit vector generation stage 504, or the
same bit vector generation stage 504 can be shared by multiple (or
all) of the parallel encryption engines 1602.
[0081] The encryption/decryption techniques of the present
invention can be implemented in a variety of ways including but not
limited to a software implementation on any programmable processor
(such as general purpose processors, embedded processors, network
processors, etc.), a hardware implementation on devices such as
programmable logic devices (e.g., field programmable gate arrays
(FPGAs)), ASICs, and a hardware and/or software implementation on
devices such as chip multi-processors (CMPs), etc. For example,
some CMPs include built-in hardware for encryption ciphers, in
which case software on parallel processors systems for the CMPs
could perform the bit vector generation and reversible
combinatorial tasks while offloading the block cipher operations to
the dedicated hardware.
[0082] However, the inventors herein particularly note that the
present invention is highly amenable to implementation in
reconfigurable logic such as an FPGA. Examples of suitable FPGA
platforms for the present invention are those described in the
following: U.S. patent application Ser. No. 11/339,892 (filed Jan.
26, 2006, entitled "Firmware Socket Module for FPGA-Based Pipeline
Processing" and published as 2007/0174841), published PCT
applications WO 05/048134 and WO 05/026925 (both filed May 21, 2004
and entitled "Intelligent Data Storage and Processing Using FPGA
Devices"), pending U.S. patent application Ser. No. 10/153,151
(filed May 21, 2002 entitled "Associative Database Scanning and
Information Retrieval using FPGA Devices", published as
2003/0018630, now U.S. Pat. No. 7,139,743), and U.S. Pat. No.
6,711,558 (entitled "Associative Database Scanning and Information
Retrieval"), the entire disclosures of each of which are
incorporated by reference herein.
[0083] FIG. 17(a) depicts an example of an implementation
environment for the present invention. FIG. 17(a) depicts a system
1700 comprising a host processor 1708 and host RAM 1710 in
communication with a disk controller 1706 via bus 1712. Disk
controller 1706 governs access to data store 1704 which may be any
device capable of storing data. In an exemplary embodiment, data
store 1704 is a mass storage medium such as a RAID system or
subsystem. In such an instance, disk controller 1706 is a RAID
controller.
[0084] Data flowing to or from data store 1704 can be routed
through reconfigurable logic device 1702 (which may be embodied by
an FPGA). One or more firmware application modules (FAMs) 1730 are
deployed on the reconfigurable logic using the techniques described
in the above-incorporated references. The different stages of the
encryption/decryption engine of the present invention can be
implemented on the reconfigurable logic device 1702 as a processing
pipeline deployed on one or more of these FAMs 1730. Firmware
socket module 1720 can be implemented as described in the
incorporated Ser. No. 11/339,892 patent application to control the
flow of data to and from the encryption/decryption engine(s)
deployed on the reconfigurable logic device 1702 via communication
paths 1732 and 1734. Data to be encrypted and stored in the data
store can be routed through the reconfigurable logic device 1702
along with appropriate control instructions for the encryption.
Such control information can include the data tag used to generate
the blockwise independent bit vectors. Moreover, these control
instructions can emanate from any source with access to system bus
1712 including sources that connect to the system bus 1712 over a
network. For example, in an embodiment wherein the data segment's
LBA is used as the data tag from which the bit vectors are
generated, the LBA can be passed to the FAM pipeline 1730 with the
data from the data store 1704 or it can be passed to the FAM
pipeline 1730 from processor 1708. Moreover, the data segments to
be encrypted can emanate from any source with access to the
reconfigurable logic device 1702. Encrypted data to be decrypted
can also be routed through the reconfigurable logic device 1702
along with appropriate control instructions for the decryption.
[0085] Thus, when encrypting a data segment to be stored at an LBA
of the data store 1704, the data blocks of the data segment can be
streamed through a FAM 1730 on reconfigurable logic device 1702
that is configured to perform encryption in accordance with the
teachings of the present invention (with the encryption FAM 1730
preferably deriving the blockwise independent bit vectors 506 from
the LBA). The resultant ciphertext produced by the encryption FAM
1730 can then be stored in data store 1704 starting at the LBA. On
decryption, the ciphertext data blocks of the encrypted data
segment (or a subset thereof) can be streamed through a decryption
FAM 1730 (or a symmetrical encryption/decryption FAM 1730) to
reconstruct the plaintext data segment (or subset thereof). Once
again, in an embodiment wherein the blockwise independent bit
vectors are derived form the data segment's LBA, the LBA can also
be used as the source of the bit vectors used during the decryption
process.
[0086] It should also be noted that for disk or file encryption
operations, it may be desirable to include the platform (e.g., FPGA
or ASIC) on which the encryption/decryption engine of the present
invention is deployed (or the encryption/decryption engine itself)
on-board the disk controller 1706. It may also be desirable for the
encryption/decryption engine to receive all data streaming to/from
the disk(s), in which case control information could be added to
the data streams to inform the encryption/decryption engine of
which data is to be encrypted/decrypted and which data is to be
passed through without modification. For example, such control
information can take the form of a flag within a data set's SCSI
control block (SCB).
[0087] The embodiment of FIG. 17(b) depicts the system 1700 wherein
bus 1712 is also connected to a network 1742 through network
interface 1740. Such a network 1742 can also serve as a source or
destination for data to be encrypted or decrypted (e.g., network
data traffic such as network data packets that may need
encryption/decryption). It should also be noted that system 1700
can be configured such that bus 1712 connects to a network 1742
(through network interface 1742) but not to a data store 1704
(through disk controller 1706) if desired by a practitioner of the
present invention in view of the use(s) to which the practitioner
intends to put the invention.
[0088] FIG. 18(a) depicts a printed circuit board or card 1800 that
can be connected to the PCI-X bus 1712 of a computer system (e.g.,
a commodity computer system or other) for use in
encrypting/decrypting data. In the example of FIG. 18(a), the
printed circuit board includes an FPGA chip 1802 (such as a Xilinx
Virtex 4 FPGA) that is in communication with a memory device 1804
and a PCI-X bus connector 1806. A preferred memory device 1804
comprises SRAM and DRAM memory. A preferred PCI-X bus connector
1806 is a standard card edge connector.
[0089] FIG. 18(b) depicts an alternate configuration for a printed
circuit board/card 1800. In the example of FIG. 18(b), a private
bus 1808 (such as a PCI-X bus), a disk controller 1810, and a disk
connector 1812 are also installed on the printed circuit board
1800. Any commodity disk interface technology can be supported, as
is understood in the art. In this configuration, the firmware
socket 1720 also serves as a PCI-X to PCI-X bridge to provide the
processor 1708 with normal access to the disk(s) connected via the
private PCI-X bus 1808.
[0090] FIG. 18(c) depicts another alternate configuration for a
printed circuit board/card 1800. In the example of FIG. 18(b), a
private bus 1808 (such as a PCI-X bus), a network interface
controller 1820, and a network connector 1822 are also installed on
the printed circuit board 1800. Any commodity network interface
technology can be supported, as is understood in the art. In this
configuration, the firmware socket 1720 also serves as a PCI-X to
PCI-X bridge to provide the processor 1708 with normal access to
the network(s) connected via the private PCI-X bus 1808.
[0091] It should be further noted that the printed circuit
board/card 1800 may also be configured to support both a disk
controller/connector 1810/1812 and a network interface
controller/connector 1820/1822 to connect the board 1800 to disk(s)
and network(s) via private PCI-X bus 1808, if desired by a
practitioner of the invention.
[0092] It is worth noting that in either of the configurations of
FIGS. 18(a)-(c), the firmware socket 1720 can make memory 1804
accessible to the PCI-X bus, which thereby makes memory 1804
available for use by an OS kernel for the computer system as the
buffers for transfers from the disk controller and/or network
interface controller to the FAMs. It is also worth noting that
while a single FPGA chip 1802 is shown on the printed circuit
boards of FIGS. 18(a)-(c), it should be understood that multiple
FPGAs can be supported by either including more than one FPGA on
the printed circuit board 1800 or by installing more than one
printed circuit board 1800 in the computer system. Further still,
it should be noted that the printed circuit boards 1800 of the
embodiments of FIGS. 18(a)-(c) can use an ASIC chip on which the
encryption/decryption engines are deployed rather than an FPGA chip
1802. if desired by a practitioner of the invention.
[0093] Exemplary applications for the present invention include but
are not limited to general purpose data encryption (e.g., files,
images, documents, etc.), disk encryption, streaming message (e.g.,
packets, cells, etc.) encryption, and streaming image encryption
(e.g., streaming reconnaissance imagery, etc.).
[0094] While the present invention has been described above in
relation to its preferred embodiment, various modifications may be
made thereto that still fall within the invention's scope. Such
modifications to the invention will be recognizable upon review of
the teachings herein. As such, the full scope of the present
invention is to be defined solely by the appended claims and their
legal equivalents.
* * * * *