Private Pipeline for Encrypted Data Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current computer systems face data privacy risks due to the need for unencrypted data in cache memory during computations, making them vulnerable to exfiltration and side-channel attacks, despite regulations requiring encryption at all stages of data lifecycle.
Innovation Solution
The implementation of private processing pipelines that enable computations on encrypted data using masked operations, ensuring data remains encrypted throughout its lifecycle, resistant to first-order side-channel attacks, and compatible with industry-standard cryptographic practices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If data is decrypted in cache memory for computation, then computation speed is improved, but data security deteriorates due to vulnerability to exfiltration and side-channel attacks
Solution Approach 1:
The patent introduces an encrypted cache memory as an intermediary storage medium between main memory and processing units. This encrypted cache serves as a mediator that allows data to be stored and processed in encrypted form, eliminating the need to decrypt data before computation while still enabling fast access. The encrypted cache memory with associative lookup capabilities allows the system to retrieve and process encrypted data directly, thus maintaining both computation speed and data security.
Solution Approach 2:
The patent changes the fundamental parameter of data state from decrypted to encrypted during cache storage and processing. By modifying the encryption status parameter of data in cache memory, the system enables computations to be performed on encrypted data directly. This parameter change eliminates the security vulnerability of storing decrypted data in cache while maintaining computational efficiency through hardware-supported encrypted operations.
2Object-affected harmful factors
If data is encrypted throughout its lifecycle, then data security is improved, but computation complexity increases
Solution Approach 1:
The patent replaces complex software-based encrypted computation mechanisms with hardware-based encrypted cache memory and associative lookup circuits. Instead of using software encryption libraries and complex algorithmic operations, the system uses hardware-implemented encrypted storage with built-in associative search capabilities. This substitution of mechanical/software systems with hardware systems significantly reduces computation complexity while maintaining security.
Solution Approach 2:
The patent segments the memory system into encrypted cache memory with associative lookup capabilities, separate from traditional decrypted cache hierarchies. By dividing the memory architecture into security-sensitive encrypted storage and standard processing areas, the system isolates complex encrypted operations to specific hardware components, reducing overall system complexity while maintaining security benefits.
3Object-affected harmful factors
If conventional encryption methods are used, then data security is improved, but performance overhead increases due to decryption/encryption operations
Solution Approach 1:
The patent performs encryption actions in advance by storing data in encrypted form directly in the encrypted cache memory before computation occurs. Data is encrypted once during write operations and remains encrypted during storage and retrieval, eliminating the need for repeated decryption and re-encryption cycles. This preliminary encryption action maintains security while avoiding performance overhead associated with multiple encryption operations.
Solution Approach 2:
The patent maintains continuous encrypted state of data throughout the entire computation lifecycle in cache memory. Instead of alternating between encrypted and decrypted states, the data remains continuously encrypted during storage, retrieval, and processing operations. This continuous encryption approach eliminates unnecessary decryption/encryption transitions, reducing performance overhead while maintaining uninterrupted security protection.
Data Source
AI summary
An example private processing pipeline may include: a masked decryption unit to perform a masked decryption operation transforming input data into masked decrypted data; a masked functional unit to produce a masked result by performing a masked operation on the masked decrypted data; and a masked encryption unit to perform a masked encryption operation transforming the masked result into an encrypted result.