On extracting private randomness over a public channel

Yevgeniy Dodis, Roberto Oliveira

Research output: Contribution to journalArticle

Abstract

We introduce strong blenders. A strong blender BLE(· ·) uses weak sources X, Y to produce BLE(X, Y) that is statistically random even if one is given Y. Strong blenders generalize strong extractors [15] and extractors from two weak random sources [25, 6]. We show that non-constructive strong blenders can extract all the randomness from X, as long as Y has logarithmic min-entropy. We also give explicit strong blenders which work provided the sum of the min-entropies of X and Y is at least their block length. Finally, we show that strong blenders have applications to cryptographic systems for parties that have independent weak sources of randomness. In particular, we extend the results of Maurer and Wolf [12] and show that parties that are not able to sample even a single truly random bit can still perform privacy amplification over an adversarially controlled channel.

Original languageEnglish (US)
Pages (from-to)252-263
Number of pages12
JournalLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume2764
StatePublished - 2003

Fingerprint

Entropy
Randomness
Extractor
Privacy
Amplification
Logarithmic
Generalise

ASJC Scopus subject areas

  • Biochemistry, Genetics and Molecular Biology(all)
  • Computer Science(all)
  • Theoretical Computer Science

Cite this

@article{c3e8bfc1a8ad4cd9824b9757b93e3f65,
title = "On extracting private randomness over a public channel",
abstract = "We introduce strong blenders. A strong blender BLE(· ·) uses weak sources X, Y to produce BLE(X, Y) that is statistically random even if one is given Y. Strong blenders generalize strong extractors [15] and extractors from two weak random sources [25, 6]. We show that non-constructive strong blenders can extract all the randomness from X, as long as Y has logarithmic min-entropy. We also give explicit strong blenders which work provided the sum of the min-entropies of X and Y is at least their block length. Finally, we show that strong blenders have applications to cryptographic systems for parties that have independent weak sources of randomness. In particular, we extend the results of Maurer and Wolf [12] and show that parties that are not able to sample even a single truly random bit can still perform privacy amplification over an adversarially controlled channel.",
author = "Yevgeniy Dodis and Roberto Oliveira",
year = "2003",
language = "English (US)",
volume = "2764",
pages = "252--263",
journal = "Lecture Notes in Computer Science",
issn = "0302-9743",
publisher = "Springer Verlag",

}

TY - JOUR

T1 - On extracting private randomness over a public channel

AU - Dodis, Yevgeniy

AU - Oliveira, Roberto

PY - 2003

Y1 - 2003

N2 - We introduce strong blenders. A strong blender BLE(· ·) uses weak sources X, Y to produce BLE(X, Y) that is statistically random even if one is given Y. Strong blenders generalize strong extractors [15] and extractors from two weak random sources [25, 6]. We show that non-constructive strong blenders can extract all the randomness from X, as long as Y has logarithmic min-entropy. We also give explicit strong blenders which work provided the sum of the min-entropies of X and Y is at least their block length. Finally, we show that strong blenders have applications to cryptographic systems for parties that have independent weak sources of randomness. In particular, we extend the results of Maurer and Wolf [12] and show that parties that are not able to sample even a single truly random bit can still perform privacy amplification over an adversarially controlled channel.

AB - We introduce strong blenders. A strong blender BLE(· ·) uses weak sources X, Y to produce BLE(X, Y) that is statistically random even if one is given Y. Strong blenders generalize strong extractors [15] and extractors from two weak random sources [25, 6]. We show that non-constructive strong blenders can extract all the randomness from X, as long as Y has logarithmic min-entropy. We also give explicit strong blenders which work provided the sum of the min-entropies of X and Y is at least their block length. Finally, we show that strong blenders have applications to cryptographic systems for parties that have independent weak sources of randomness. In particular, we extend the results of Maurer and Wolf [12] and show that parties that are not able to sample even a single truly random bit can still perform privacy amplification over an adversarially controlled channel.

UR - http://www.scopus.com/inward/record.url?scp=30744451262&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=30744451262&partnerID=8YFLogxK

M3 - Article

VL - 2764

SP - 252

EP - 263

JO - Lecture Notes in Computer Science

JF - Lecture Notes in Computer Science

SN - 0302-9743

ER -