Three recently proposed schemes use secret sharing to support privacy-preserving data outsourcing. Each secret in the database is split into n shares, which are distributed to independent data servers. A trusted client can use any k shares to reconstruct the secret. These schemes claim to offer security even when k or more servers collude, as long as certain information such as the finite field prime is known only to the client. We present a concrete attack that refutes this claim by demonstrating that security is lost in all three schemes when k or more servers collude. Our attack runs on commodity hardware and recovers a 8192-bit prime and all secret values in less than an hour for k = 8.
%0 Book Section
%1 springerlink:10.1007/978-3-642-31540-4_12
%A Dautrich, Jonathan
%A Ravishankar, Chinya
%B Data and Applications Security and Privacy XXVI
%C Berlin / Heidelberg
%D 2012
%E Cuppens-Boulahia, Nora
%E Cuppens, Frédéric
%E Garcia-Alfaro, Joaquin
%I Springer
%K data-outsourcing secret-sharing
%P 145-160
%R 10.1007/978-3-642-31540-4_12
%T Security Limitations of Using Secret Sharing for Data Outsourcing
%U http://dx.doi.org/10.1007/978-3-642-31540-4_12
%V 7371
%X Three recently proposed schemes use secret sharing to support privacy-preserving data outsourcing. Each secret in the database is split into n shares, which are distributed to independent data servers. A trusted client can use any k shares to reconstruct the secret. These schemes claim to offer security even when k or more servers collude, as long as certain information such as the finite field prime is known only to the client. We present a concrete attack that refutes this claim by demonstrating that security is lost in all three schemes when k or more servers collude. Our attack runs on commodity hardware and recovers a 8192-bit prime and all secret values in less than an hour for k = 8.
%@ 978-3-642-31539-8
@incollection{springerlink:10.1007/978-3-642-31540-4_12,
abstract = {Three recently proposed schemes use secret sharing to support privacy-preserving data outsourcing. Each secret in the database is split into n shares, which are distributed to independent data servers. A trusted client can use any k shares to reconstruct the secret. These schemes claim to offer security even when k or more servers collude, as long as certain information such as the finite field prime is known only to the client. We present a concrete attack that refutes this claim by demonstrating that security is lost in all three schemes when k or more servers collude. Our attack runs on commodity hardware and recovers a 8192-bit prime and all secret values in less than an hour for k = 8.},
added-at = {2012-09-11T10:11:23.000+0200},
address = {Berlin / Heidelberg},
affiliation = {University of California, Riverside, USA},
author = {Dautrich, Jonathan and Ravishankar, Chinya},
biburl = {https://www.bibsonomy.org/bibtex/2c255a9a820d632b038a71dac2b9160a6/matthiashuber},
booktitle = {Data and Applications Security and Privacy XXVI},
description = {Abstract - SpringerLink},
doi = {10.1007/978-3-642-31540-4_12},
editor = {Cuppens-Boulahia, Nora and Cuppens, Frédéric and Garcia-Alfaro, Joaquin},
interhash = {967b494da3f0f87be671132e8ac8ebab},
intrahash = {c255a9a820d632b038a71dac2b9160a6},
isbn = {978-3-642-31539-8},
keyword = {Computer Science},
keywords = {data-outsourcing secret-sharing},
pages = {145-160},
publisher = {Springer},
series = {Lecture Notes in Computer Science},
timestamp = {2012-09-11T10:11:23.000+0200},
title = {Security Limitations of Using Secret Sharing for Data Outsourcing},
url = {http://dx.doi.org/10.1007/978-3-642-31540-4_12},
volume = 7371,
year = 2012
}