Upload
others
View
7
Download
0
Embed Size (px)
Citation preview
Lightweight, Scalable and Secure Computations for Engineering Design
Siva C. Chaduvula, Mikhail J. Atallah, Jitesh H. PanchalPurdue University
RQ: How can designers execute their computations securely and efficiently in a co-design setting?10
.7 96.3
953.7
3.95 29.7
396.6
1 0 100 1000VECTORLENGTH
BANDWIDTH(KB)Sharemind ProposedApproach
Computational overhead imposed by existing cryptographic approaches make them unsuitable for
intensive and iterative engineering computationsInformation shared is prone to leakage and misuse
through collaborators
2. Privacy Preserving Elicitation of Customer Needs
References1. Chaduvula SC, Atallah MJ, Panchal JH. Secure Co-design: Achieving Optimality without Revealing. Journal of Computing and Information Science in Engineering, 2018, In Press, doi:
10.1115/1.40394312. Wang, S., Bhandari, S., Chaduvula, S.C., Atallah, M.J., Panchal, J.H. and Ramani, K., 2017. Secure collaboration in engineering systems design. Journal of Computing and Information Science
in Engineering, 17(4), p.041010. 3. Sharemind: Bogdanov, D., Niitsoo, M., Toft, T. and Willemson, J., 2012. High-performance secure multi-party computation for data mining applications. International Journal of Information
Security, 11(6), pp.403-418.
Boeing Mitsubishi1. Security Mismatch
X
3. Common Collaborator
Y XBoeing Mitsubhishi
2. Future Competitor
X Boeing Airbus
Collaborator
Internal attacker
External attacker
General Electric
X X
Confidential Input : X
Note: Enterprises mentioned in this slide are purely for illustrative purposes
Source: Boeing Images: Image taken from http://wildabouttravel.boardingarea.com/2017/02/the-787-10/
Engineering collaborations involve sensitive information flow among designers from
different enterprises from multiple countries
Mot
ivat
ion
App
roac
h &
Res
ults
Bro
ader
Impa
cts 1. Secure Outsourcing 3. Secure eProcurement
Cloud service provider
Designer
Additive Share 1
AdditiveShare 2
Collaborator 1Confidential value : 11
11+1811
-1811
Collaborator 2
Third party
Fast Lightweight And SecureComputations (FLASC)
• Adding/multiplying a parameter with a largerandom number hides the parameter
• Adding or multiplying a large number is orders ofmagnitude faster than existing cryptographicprimitives that rely on modular exponentiation
SPOs Performance
Scalability (VIP)
Secure Primitive Operations (SPOs)
Application Logic
Vector Inner product (VIP)
Less than Zero (LTZ)
Shuffle and re-split (SRS)
Equal to Zero(ETZ)
Direct Min (Dmin)
Matrix product (MP)
Linear Solver (LS)
Sort (S)
Set Operations Order statistics
Branching (B)
Matrix Inverse (MI)
Exponentiation (EXP)
Addition(ADD)
Eigenvector(EV)
Gradient Descent (GD) SVD (SVD)
Matrix Rank (MR)
Auctions (A)