Mediated Secure Computation for Collusion-Resistant Privacy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing secure multiparty computation techniques face challenges in preventing collusion among agents, leading to invalid results, especially when agents wish to keep their inputs private and remain anonymous, while conventional methods assume a majority of honest parties.
Innovation Solution
A mediated secure computation system that assigns unique identifiers to nodes in a network, encrypts logical circuits with a common random number unknown to the server, and combines encrypted portions to execute a logical circuit, ensuring privacy and preventing collusion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional secure multiparty computation is used, then privacy of inputs is maintained, but collusion among agents cannot be prevented leading to invalid results
Solution Approach 1:
A mediator is introduced between the agents and the computation system. The mediator receives encrypted inputs from agents, executes the computation, and distributes results. This intermediary structure prevents direct collusion between agents while maintaining privacy, as the mediator acts as a trusted third party that coordinates the computation without allowing agents to directly influence each other's inputs or outputs.
Solution Approach 2:
The computation system is divided into separate functional components: input encryption modules for each agent, a central computation module, and result distribution modules. Each agent's input is encrypted separately and processed independently through the computation module, preventing agents from colluding to manipulate the computation while maintaining individual privacy and computational integrity.
2Loss of information
If agents keep their inputs private and remain anonymous, then privacy is maintained, but preventing collusion becomes more difficult
Solution Approach 1:
The mediator serves as a trusted intermediary that receives encrypted inputs from anonymous agents, executes the computation, and distributes results. This structure maintains agent anonymity and input privacy while preventing collusion, as the mediator controls all communication channels and computation execution, ensuring that agents cannot collude to manipulate results without being detected.
Solution Approach 2:
The system incorporates feedback mechanisms where the mediator verifies computation results and provides confirmation to agents. This feedback loop ensures that agents who attempt collusion or provide incorrect inputs can be detected and penalized, maintaining both privacy and reliability simultaneously.
3Adaptability or versatility
If a majority of parties are assumed honest, then conventional SMC protocols work, but systems with malicious minority parties fail
Solution Approach 1:
The mediator acts as a trusted third party that coordinates computation between agents with potentially malicious behavior. The mediator verifies inputs, executes computations securely, and distributes results, enabling the system to tolerate malicious minority parties while maintaining computational correctness. This intermediary structure provides the reliability needed for systems where not all parties can be trusted.
Solution Approach 2:
The system incorporates verification and validation mechanisms that prevent malicious inputs from affecting computation results. The mediator performs beforehand checks on agent inputs and computation processes, cushioning against potential malicious behavior by malicious parties before they can compromise the overall computational correctness.
Data Source
AI summary
Techniques are described for mediated secure computation. A unique identifier value may be assigned to each one of a plurality of nodes included in a network. An encrypted portion of a logical circuit may be received at a server from each of the nodes, the logical circuit including one or more gates, each gate associated with one or more logical input wires and one or more logical output wires, the logical circuit associated with a function, wherein each encrypted portion is encrypted based on a random number value that is common to the plurality of nodes and unknown at the server. A result may be obtained based on executing the logical circuit, based on combining the encrypted portions of the logical circuit received at the server.