Dear crypto enthusiasts!
On January 24th, I triggered a discussion about true random sources.
Since them, I put in writing the design I was working on.
The result is here: http://www.connotech.com/secret_rand_src_design.html
with the title "Secret Random Source Design Notes" (abstract reproduced
below).
Regards,
--
- Thierry Moreau
CONNOTECH Experts-conseils inc.
9130 Place de Montgolfier
Montreal, QC, Canada H2M 2A1
Tel. +1-514-385-5691
Secret Random Source Design Notes
abstract
This document addresses the software and system design of a secret
random data source for use by cryptography-based information security
applications. The intended field of use is high assurance systems, but
with a bias towards innovation where standard or prevailing practice
appear to have limitations.
The proposed arrangement is implementaion oriented and tied to a
prevailing software building methodology in the Linux (or POSIX
compliant systems) developer community. It is described with design
options allowing a secre random source function for server systems and
client applications.
Basically, the arrangement relies on a true random source (not strictly
part of the arrangement provisions) to seed a PRNG (Pseudo Random Number
Generator) of a hybrid design made of the BBS (Blum Blum and Shub)
cryptographically strong generator and an LFSR-based combined generator.
Each instance of this hybrid PRNG can be tailored with parameters to
balance a very long period length, relative efficiency, and theoretical
security. This parameterization occurs at the software build time.
The seed size determined by the PRNG parameters is correlated to the
PRNG period length. The relationships between the parameters, seed size,
and period are investigated in the perspective of entropy preservation
from the true random source up to the PRNG output sequence delivered to
cryptographic applications. The proposed arrangement allows a true
random source with a non-compact encoding of its entropy contents,
provided sufficiently large PRNG parameters. Encoding compactness and
distribution uniformness are independent characterizations of a true
random source. In cases where only a non-uniform distribution true
random source is available, the proposed arrangement entropy
preservation assurance may be relied upon for making good use of a not
so good true random source.
The document discusses various aspects of theoretical or academic work
that can support the proposed arrangement. This extends to a more
abstract analysis of randomness in the information theory, where we were
unable to see how the more abstract work can guide implementation
oriented design decisions. The document remains focused on addressing
practice oriented details that may impact the overall system or software
security.
--- + end + ---
_______________________________________________
cryptography mailing list
[email protected]
http://lists.randombit.net/mailman/listinfo/cryptography
