Browse Prior Art Database

Random Numbers Produced Via a Technique Employing Both a White Noise Generator and the Data Encryption Algorithm

IP.com Disclosure Number: IPCOM000122645D
Original Publication Date: 1991-Dec-01
Included in the Prior Art Database: 2005-Apr-04
Document File: 3 page(s) / 132K

Publishing Venue

IBM

Related People

Matyas, SM: AUTHOR

Abstract

This article discloses a method of generating random numbers which itself makes use of two number generation techniques: (1) hardware "white noise" generator, and (2) a pseudo-random generator based on the Data Encryption Algorithm (DEA). The method has the advantages of each random number generation technique. The "white noise" generator ensures that the generated numbers are unpredictable and nonrepeatable, even if the hardware design is known. Cryptographic feedback techniques based on the DEA are used to "smooth" the statics and remove any bias that may be inherent in the "white noise" generator. A DEA 1-way function is also employed to ensure that a strong generator of pseudo- random numbers is achieved even if the "white noise" generator should malfunction.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 51% of the total text.

Random Numbers Produced Via a Technique Employing Both a White Noise
Generator and the Data Encryption Algorithm

      This article discloses a method of generating random
numbers which itself makes use of two number generation techniques:
(1) hardware "white noise" generator, and (2) a pseudo-random
generator based on the Data Encryption Algorithm (DEA).  The method
has the advantages of each random number generation technique.  The
"white noise" generator ensures that the generated numbers are
unpredictable and nonrepeatable, even if the hardware design is
known.  Cryptographic feedback techniques based on the DEA are used
to "smooth" the statics and remove any bias that may be inherent in
the "white noise" generator.  A DEA 1-way function is also employed
to ensure that a strong generator of pseudo- random numbers is
achieved even if the "white noise" generator should malfunction.

      Methods of generating random numbers and random cryptographic
keys are well known, and prior publications described these methods
(1,2,3,4,5).  Prior methods, however, describe only ways of
generating random numbers using the DEA.  The present method enlarges
upon those prior methods by describing a technique which combines a
hardware "white noise" random number generator with a pseudo-random
number generator based on the DEA.

      The figure illustrates a proposed random number generator which
consists of a hardware generation component 1 (shown on the left) and
a software generation component 2 (shown on the right).  The
generator produces 64-bit random numbers.  A cryptographic function
available on the card allows a random number to be requested.  A
64-bit output register 3 is used to store the so-produced random
number. The process of producing a 64-bit random number consists of
adding "1" to two secret 64-bit inputs (Input1 and Input2) ignoring
any overflow (i.e., addition modulo 264), encrypting the resulting
64-bit numbers with secret keys KL and KR (KL / KR), respectively,
Exclusive ORing the two 64-bit resulting ciphertexts together with
the 64-bit random number produced by the Hardware Random Number
Generator to produce a 64-bit output at Output Register 3.  (Using
modulo 264 counters seeded with 64-bit random numbers has the
advantage that the counter can wrap to zero with no loss in security.
The only concern would be if the counter could wrap on all 264
possible values, which for practical purposes can never happen.  A
test to detect such a wrapping condition could be placed in the
procedure, if desired, but is unnecessary.)

      The proposed random number generator assumes that KL, KR,
Input1, and Input2 are appropriately initialized with secret seed
values which are all different, and ideally all independent.
However, suitable seed values can be derived from as few a...