SPAR-MPC Workshop 10 Feb

Overview of Workshop Goals
and RFI Responses
Emily Shen
SPAR-MPC Workshop
28 May 2014
SPAR-MPC Vision
• Multi-party computation (MPC) protocol generation tools:
– Input: description of computation, security requirements,
application parameters
– Output: efficient, secure MPC protocols tailored to application
• Properties of tools:
– Accessible to end users
– Can achieve many points in security vs. efficiency space
– Can generate new, efficient protocols quickly
SPAR-MPC - 2
ES 05/28/14
SPAR-MPC Workshop
10 Feb: Request for
Information (RFI)
31 Mar: IARPA
received 21
responses
15 Apr: IARPA +
MIT LL reviewed
responses
28-29 May:
SPAR-MPC
Workshop
Workshop objective: Discuss remaining questions to understand challenges
and feasibility of potential SPAR-MPC directions
SPAR-MPC - 3
ES 05/28/14
Workshop Goals
Understand challenges and approaches for:
1. Security vs. efficiency tradeoffs, including allowing leakage
2. Toolbox components, compilation, and composition
3. Program structures, problem decomposition and scoping
4. Metrics for progress and success
SPAR-MPC - 4
ES 05/28/14
Workshop Goals
Understand challenges and approaches for:
1. Security vs. efficiency tradeoffs, including allowing leakage
2. Toolbox components, compilation, and composition
3. Program structures, problem decomposition and scoping
4. Metrics for progress and success
SPAR-MPC - 5
ES 05/28/14
Security vs. Efficiency Tradeoffs
RFI response ideas:
• Leakage challenges
• Other security and functionality variations
Remaining questions:
• Further understand challenges of representing, reasoning about
leakage and other relaxations
Breakout sessions:
• Group 1: User specification of application requirements
• Group 2: Crypto language for leakage
• Group 3: Translation between application and crypto
requirements
• Group 4: Real-world implications of formal security properties
SPAR-MPC - 6
ES 05/28/14
Security and Functionality Variations
• Adversaries (besides semi-honest and malicious)
–
–
–
–
Covert
Mobile
Rational
Non-coordinating
• Security properties (besides confidentiality and integrity)
–
–
–
–
Anonymity
Fairness
Limited information revealed by output
Graceful degradation of security
• Functionality relaxations:
– Imperfect correctness, approximate computation
– Restricted inputs
SPAR-MPC - 7
ES 05/28/14
Workshop Goals
Understand challenges and approaches for:
1. Security vs. efficiency tradeoffs, including allowing leakage
2. Toolbox components, compilation, and composition
3. Program structures, problem decomposition and scoping
4. Metrics for progress and success
SPAR-MPC - 8
ES 05/28/14
Toolbox Components, Compilation, and
Composition
RFI responses ideas:
• Build and benchmark a library of MPC tools with common API
• Build new, complex protocols from building blocks
• Develop optimizing compiler
• Study composition and concurrency
Remaining questions:
• What would the components and API of MPC toolbox look like?
• What are the challenges of compilation and composition?
Breakout session:
• Group 5: Crypto toolbox components, compilation, and
composition
SPAR-MPC - 9
ES 05/28/14
Workshop Goals
Understand challenges and approaches for:
1. Security vs. efficiency tradeoffs, including allowing leakage
2. Toolbox components, compilation, and composition
3. Program structures, problem decomposition and scoping
4. Metrics for progress and success
SPAR-MPC - 10
ES 05/28/14
Program Structure, Problem
Decomposition and Scoping
RFI response ideas:
• MPC applications and protocols described by parameters of
model of computation, security, and functionality
Remaining questions:
• What parameter subspaces are most appropriate and feasible
to address?
• How to decompose the problem and structure SPAR-MPC
program?
Breakout sessions:
• Group A: Crypto research risks/rewards
• Group B: User language/interface design risks/rewards
• Group C: Potential SPAR-MPC program structures
SPAR-MPC - 11
ES 05/28/14
Workshop Goals
Understand challenges and approaches for:
1. Security vs. efficiency tradeoffs, including allowing leakage
2. Toolbox components, compilation, and composition
3. Program structures, problem decomposition and scoping
4. Metrics for progress and success
SPAR-MPC - 12
ES 05/28/14
Metrics for Success and Progress
RFI response ideas:
• Use metrics for individual MPC protocols
• Benchmark library of MPC tools developed with common API
• Compare performance of MPC protocols with insecure
computation
Remaining questions:
• Metrics for MPC protocol generation tools and overall research
program
Breakout groups:
• Group D: Use cases and likely impact
• Group E: Metrics for success and progress
SPAR-MPC - 13
ES 05/28/14
SPAR-MPC Breakout Agenda
Wednesday, 28 May
Thursday, 29 May
Numbered Breakout Sessions
Lettered Breakout Sessions
1. User specification of
application requirements
A. Crypto research risks/rewards
2. Crypto language for leakage
B. User language/interface design
risk/rewards
3. Translation between application
and crypto requirements
C. Potential SPAR-MPC program
structures
4. Real-world implications of
crypto properties
D. Use cases and likely impact
5. Crypto toolbox components,
compilation, and composition
SPAR-MPC - 14
ES 05/28/14
E. Metrics for success and
progress
Today’s Breakout Sessions
Wednesday, 28 May
Numbered Breakout Sessions
1.User specification of application requirements
2.Crypto language for leakage
3.Translation between application and crypto requirements
4.Real-world implications of crypto properties
5.Crypto toolbox components, compilation, and composition
SPAR-MPC - 15
ES 05/28/14