Distributed Secure Multiparty
Computations in Hostile
Environments
by Joel Dominic, James Hunt, and Adam Mcculloch
Content
 The Encryption Method Used
 The Problem
 Our Solution
 Conclusion
Asymmetric Encryption
Sender's
Encryption Key
Plain Text
Encrypted
Data
Receiver’s
Decryption Key
Why Use Multiparty Computations
 Allows low end machines to quickly complete computations
that would otherwise take a great amount of time.
 It frees up processing power on the base machine without
effecting the time it takes to solve a problem.
 It allows cooperative work within a non-geographic
specific network.
Why Secure Multiparty Computations
 When computing sensitive information, some malicious
entities outside the network may attempt to sniff out
secure information.
 Malicious entities may penetrate the network and try to
feed false results or otherwise disrupt the computation.
 Not all parties involved in computing the data may not
have the authorization to view the end result.
Securely Computing a Function Over a
Network
Our Goal
Securely computing data across a
network not only securing the integrity
of the data itself but also the place
each machine has in the computation
Disclaimer
Our solution should only be used in situations
where the data to be computed must remain
secure but is NOT time critical.
Our approach results in a secure computation of
the input set but the methods used will severely
impact time and resource efficiency.
Step One
Take the all the operations to be solved and break
them into a sequence of independent stand-alone
(atomic) functions.
Function 1
Complex
Operation
Function 2
Function 3
Step Two
 The Host machine generates n+1 asymmetric key
combinations (where n is the number of functions)
 Each function is then wrapped in a mobile agent in
such a way that:
• Encrypt keys 2 .. n+1 are added before functions 1 .. n
• Decrypt keys 1 .. n are added after functions 1 .. n
• The host holds onto encrypt key 1 and decrypt key n+1
 Each function is then randomly distributed to a
machine in the computation network
Step Three
1.
The host machine broadcasts the initial encrypted
input across the network.
2. Every machine in the network receives the
encrypted information and attempts to decrypt it
 If successful:



the atomic function is performed
the information is re-encrypted and broadcast out to the network
Mobile agent ignores all other input for this stage
 If unsuccessful:


the information is discarded
if there is no more input for this stage, mobile agent broadcasts
‘junk data’
Conclusion
Our solution allows for data to be securely
computed across a network without compromising
the integrity of the data OR the functions
involved
References
Ishai, Yuval, Eyal Kushilevitz, Rafail Ostrovsky, and Amit Sahai. "ZeroKnowledge Proofs from Secure Multiparty Computation." SIAM
Journal on Computing 39.3 (2009): 1121. Print.
Backes, Michael, and Dominique Unruh. "Computational Soundness
of Symbolic Zero-knowledge Proofs." Journal of Computer Security 18
(2010): 1077-155. Print.