Privacy-Preserving Distributed Computing via Approximate FHE and Coded Computing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Distributed computing systems face challenges in maintaining data privacy when processing sensitive information, as existing homomorphic encryption schemes introduce noise and errors that deteriorate result accuracy with multiple worker nodes, preventing practical use in large-scale computations.
Innovation Solution
Implementing an approximate Fully Homomorphic Encryption (FHE) scheme combined with Berrut Approximated Coded Computing (BACC) that encodes and encrypts data portions, allowing servers to perform operations on encrypted data without decryption, and using a subset of results to achieve accurate final results with reduced error.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If homomorphic encryption schemes are used to preserve data privacy in distributed computing, then data privacy is improved, but result accuracy deteriorates due to noise accumulation with multiple worker nodes
Solution Approach 1:
The patent segments the distributed computing task into multiple independent coded computing tasks, each processed by different worker nodes. By dividing the overall computation into smaller segments that can be independently encrypted and processed, the system maintains privacy while managing noise accumulation at a manageable scale for each segment.
Solution Approach 2:
The patent combines homomorphic encryption with coded computing techniques to create a composite privacy-preserving computing framework. This composite approach integrates the cryptographic protection of homomorphic encryption with the error-correcting capabilities of coded computing, allowing the system to tolerate noise accumulation while maintaining both privacy and accuracy.
2Productivity
If more worker nodes are added to distributed computing systems to improve computational capacity, then productivity is improved, but result accuracy deteriorates due to increased noise from additional homomorphic encryption operations
Solution Approach 1:
The patent changes key parameters including the encryption scheme parameters (modulus, polynomial degree), coding parameters (redundancy level, code rate), and aggregation parameters (threshold for result reconstruction). By optimizing these parameters together, the system can scale to more worker nodes while controlling noise accumulation through parameter tuning.
Solution Approach 2:
The patent introduces coded computing intermediaries that act as noise filters between the encrypted computations and the final result aggregation. These intermediaries process the encrypted outputs from worker nodes, applying error-correcting codes that can tolerate a certain level of noise accumulation, thereby enabling the system to incorporate more worker nodes without proportionally increasing noise impact.
3Ease of operation
If approximate coding schemes like BACC are used to reduce computational complexity, then ease of operation is improved, but result accuracy deteriorates
Solution Approach 1:
The patent adjusts the approximation parameters of BACC (such as the degree of approximation, tolerance thresholds, and reconstruction accuracy levels) to find an optimal balance between computational complexity and result accuracy. By carefully tuning these parameters in conjunction with the homomorphic encryption parameters, the system achieves sufficient accuracy for practical applications while maintaining computational efficiency.
Data Source
AI summary
A client device encodes at least two datasets using one or more encoding functions to generate encoded data portions that are encrypted using a first key according to an approximate Fully Homomorphic Encryption (FHE) scheme to generate encrypted data portions that are sent to a plurality of servers. Encrypted results are received from at least a subset of servers of the plurality of servers. Each encrypted result is calculated by a respective server using at least two encrypted data portions received by the server. The encrypted results are decrypted using a secret key according to the approximate FHE scheme to derive decrypted encoded results that are decoded using an approximate decoding function. In one aspect, an encrypted result is calculated by each server by evaluating a multivariate function using the at least two encrypted data portions received by the server.