Bridge Gates for Multi-Party Computation Input Verification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Multi-Party Computation (MPC) processes do not guarantee the correctness of inputs provided by parties, especially when one party is adversarially controlled, which can compromise the entire process.
Innovation Solution
A method involving two parties and a server where initial garbled labels are generated and sent to the server, followed by the creation of bridge gates to translate these labels into garbled values for a fresh garbled circuit, allowing the server to verify the correctness of inputs through commitments and pseudorandom generator seeds, ensuring secure and accurate computation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If MPC processes are performed without input verification, then the computation can proceed freely and efficiently, but the correctness of inputs from adversarially-controlled parties cannot be guaranteed
Solution Approach 1:
The system performs preliminary actions by generating commitments to input values before the actual MPC computation begins. Each party creates a commitment to their input during an initialization phase, which is then used during the evaluation phase to verify that the inputs used in computation match the committed values. This preliminary commitment mechanism ensures input correctness without adding complex verification during the computation itself.
Solution Approach 2:
The system introduces commitments as an intermediary mechanism between the parties' inputs and the MPC computation. These commitments act as a mediator that allows verification of input correctness without revealing the actual inputs or requiring complex real-time verification during computation. The server uses these commitments to verify that parties are using their declared inputs throughout the computation process.
2Reliability
If bridge gates are generated to translate initial garbled labels to fresh garbled values, then input correctness can be verified, but the computation process becomes more complex
Solution Approach 1:
The bridge gates are generated in advance during the initialization phase, before the actual MPC computation begins. This preliminary generation allows the verification mechanism to be prepared beforehand, so that during the evaluation phase, the server can efficiently verify input correctness using pre-computed bridge gates without adding significant overhead to the computation process.
Solution Approach 2:
The system creates bridge gates that copy or translate the initial garbled labels into the format needed for the fresh garbled circuit. These bridge gates serve as a copy mechanism that preserves the input values while transforming them into the appropriate representation for verification, allowing the server to check input correctness without requiring the original parties to repeatedly provide their inputs.
3Reliability
If commitments and pseudorandom generator seeds are used to verify inputs, then security and integrity are enhanced, but the initialization and evaluation phases require additional steps
Solution Approach 1:
The commitments and pseudorandom generator seeds are established during the initialization phase before the evaluation phase begins. This preliminary setup allows all verification mechanisms to be prepared in advance, so that during the evaluation phase, the additional steps required for verification are minimized and can be performed efficiently using the pre-established commitments and seeds.
Data Source
AI summary
A method of performing a Multi-Party Computation (MPC) process between two parties and a server, the parties generating initial garbled labels to an initial garbled circuit and sending the initial garbled labels corresponding to an input to the server, the parties generating a fresh garbled circuit and generating multiple bridge gates for translating the initial garbled labels to garbled values for the inputs to the fresh garbled circuit, where each of the bridge gates is associated with a specific input wire of the fresh garbled circuit and maps a value of the initial garbled labels to a value of garbled labels of the fresh garbled circuit, where the server computes fresh garbled values for the fresh garbled circuit using the bridge gates and the initial garbled values and evaluates the fresh garbled circuit using the fresh garbled labels.