Encrypted Big Data Analysis via Federated Learning and Trusted Execution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current big data secure sharing analysis services lack effective protection of sensitive information and commercial secrets, especially in multi-party collaborative analysis, leading to data leakage and inadequate privacy protection, which hinders the development of various disciplines.
Innovation Solution
A whole-lifecycle encrypted big data analysis method and system that processes data from multiple sources in a trusted execution environment, encrypts data locally, and performs federated model training, ensuring that raw data never crosses borders, with a secure computing server analyzing and validating the data without direct exchange between sources, using techniques like Oblivious RAM Trees and collision checking.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If data is shared and analyzed across multiple institutions, then big data analysis capability is improved, but data security and privacy protection deteriorate
Solution Approach 1:
The patent introduces a trusted third-party platform that acts as an intermediary between multiple data institutions. This platform enables secure data sharing and collaborative analysis by mediating the interaction between data providers and analyzers, ensuring that raw data remains protected while still allowing meaningful analysis to occur through the intermediary's secure environment
Solution Approach 2:
The patent creates encrypted copies of data that can be shared and analyzed without exposing the original sensitive information. Through techniques like homomorphic encryption and secure multi-party computation, the system allows operations on encrypted data copies, enabling analysis capability while maintaining the security and integrity of the original data sources
2Reliability
If data encryption is implemented for privacy protection, then data security is improved, but data processing efficiency deteriorates
Solution Approach 1:
The patent implements preliminary data preprocessing and formatting at the data source before encryption and transmission. By preparing data in advance with proper schemas and formats, the system reduces the computational burden during encrypted processing, thereby maintaining higher processing efficiency while still providing strong encryption protection
Solution Approach 2:
The patent employs advanced encryption techniques and algorithms that optimize the balance between security strength and processing speed. By carefully selecting and adjusting encryption parameters, the system achieves adequate security protection while minimizing the performance overhead associated with encrypted data processing
3Reliability
If federated learning is used for collaborative analysis, then data privacy protection is improved, but system complexity deteriorates
Solution Approach 1:
The patent designs a unified federated learning framework that can handle multiple types of data, models, and analysis tasks through a common architecture. This universal system reduces complexity by providing standardized interfaces and processes that work across different institutions and data types, rather than requiring separate custom solutions for each scenario
Data Source
AI summary
This is a whole-lifecycle encrypted big data analysis method and system for the data from the different sources. The method performs unified local modeling of data from multiple data sources and transmitting them to a secure computing server after encryption. This secure computing server then processes and analyzes the data, including feature extraction, model training and model validation. The system includes multiple data sources corresponding to the described method and a secure computing server, for providing data, analyzing and processing the data. By processing and encrypting the data locally, and by supporting secure data sharing and federated learning of multiple data sources, this invention achieves protection of the data sources and addresses the privacy and security issues of the raw data for cross-institution big data collaborative analysis.