Server-Aided Multiparty Computation for Encrypted Decimal Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing multi-party computation methods are inefficient for handling decimal numbers and complex functions, requiring high computational costs and times, and compromise data security and privacy, especially when using a limited number of data providers.
Innovation Solution
A multi-party computation method utilizing Additive Secret Sharing and a 'server-aided' model with three untrusted but non-colluding servers to process encrypted datasets, allowing secure computation of public functions while maintaining privacy, using protocols like Araki and bit-decomposition to optimize operations on decimal numbers and complex functions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional MPC protocols are used to ensure secure multi-party computation, then security and privacy protection are improved, but communication overhead and computational complexity increase significantly
Solution Approach 1:
The patent segments the secret sharing scheme into different levels (first-level and second-level secret sharing) and divides the computation into distinct phases (key generation, encryption, computation, decryption). This segmentation allows each component to be optimized independently, reducing overall communication overhead while maintaining security.
Solution Approach 2:
The patent introduces an intermediary mechanism using additive homomorphic encryption and secret sharing protocols that allow computations to be performed on encrypted data without direct communication between all parties. The encrypted shares act as intermediaries that carry information securely through the computation process.
2Reliability
If traditional MPC protocols are used to ensure secure multi-party computation, then security and privacy protection are improved, but computational complexity increases significantly
Solution Approach 1:
The patent performs preliminary actions by pre-generating secret shares and encryption keys before the actual computation. The key generation phase creates first-level and second-level secret shares in advance, and the encryption structure is prepared beforehand, reducing computational burden during the actual computation phase.
Solution Approach 2:
The patent replaces traditional mechanical MPC computation mechanisms with homomorphic encryption-based computation. Instead of requiring complex interactive protocols and multiple rounds of communication, the system uses mathematical properties of homomorphic encryption to perform computations directly on encrypted data, significantly reducing computational complexity.
3Reliability
If data is encrypted to protect privacy, then security is improved, but data utility and analytical capability deteriorate
Solution Approach 1:
The patent changes the parameter of data representation from plaintext to homomorphically encrypted data, which maintains the mathematical structure necessary for computation while protecting privacy. The encrypted data retains computational properties that allow useful operations to be performed without decryption.
Solution Approach 2:
The patent substitutes traditional data processing mechanisms with homomorphic encryption-based computation. This allows analytical operations to be performed on encrypted data directly, maintaining data utility while ensuring privacy protection throughout the computation process.
Data Source
Figure 1
AI summary
A method of multi-party computation, for processing and secure handing of a plurality of data associated with one or more users, comprising the steps of: providing a predetermined multi-party computation algorithm; each user being able to send a first dataset to a data processing unit (1) via a respective second data processing unit (2) distinct from the first data processing unit (1) and in signal communication with the first data processing unit (1); each first dataset being associated with the user and comprising one or more encrypted numerical values; processing each first dataset that has been sent using at least one reference function residing in the first data processing unit (1) to generate a respective encrypted result for each reference function; requesting the first data processing unit to send the result using a predetermined function shared by the users and a respective second data processing unit sending the result to the second requesting data processing unit (2); wherein the step of sending the first dataset comprises the sub steps of detecting the presence of decimal numerical values and integer numerical values among the numerical values of the first dataset; associating an integer mantissa and an exponent of a floating-point representation with each decimal numerical value that has been detected; encrypting each integer numerical value and each mantissa using the predetermined multi-party computation algorithm.