Secure In-Memory Vector Encoding for Sniffer-Resistant Similarity Searches
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data transfer methods compromise sensitive data security due to interception by sniffers and the need for frequent encryption/decryption across networks and systems.
Innovation Solution
Implementing a secure, in-memory unit on an associative processing unit (APU) for encoding, encrypting, and decrypting data using neural proxy hash encoding, reducing the need for frequent encryption/decryption and protecting data from sniffers by using encoded vectors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If data is encrypted using software on CPU before transfer, then data security is improved, but processing time and system complexity increase due to frequent encryption/decryption operations
Solution Approach 1:
The patent replaces traditional CPU-based software encryption with hardware-level encryption implemented in the APU's in-memory unit. This substitution of mechanical/software encryption with hardware encryption enables faster processing while maintaining security, directly resolving the contradiction between data security and processing time.
Solution Approach 2:
The patent introduces an in-memory unit as an intermediary component between the CPU and memory. This intermediary handles encryption and decryption operations locally in memory, reducing the need for frequent CPU intervention and network transmission of encrypted data, thereby decreasing processing time while maintaining security.
2Ease of operation
If data is transmitted across network for processing, then access to data is improved, but security is compromised due to interception by sniffers
Solution Approach 1:
The patent extracts sensitive data processing operations from the networked environment and places them locally in the APU's in-memory unit. By performing encoding, encryption, and decryption locally before data leaves the device, the system eliminates the vulnerability window where sniffers could intercept data during network transmission.
Solution Approach 2:
The patent performs encryption and encoding actions preliminarily before data is transmitted across the network. The in-memory unit encrypts data and creates encoded vectors before transmission, so that even if data is intercepted during network transfer, it remains secured against sniffer interception.
3Productivity
If encoded vectors are used for similarity search, then search efficiency is improved, but device complexity increases due to additional encoding components
Solution Approach 1:
The patent merges the encoding, encryption, and similarity search functions into a single integrated in-memory unit within the APU. This consolidation combines multiple previously separate components into one unified structure, improving search efficiency while minimizing the increase in device complexity through functional integration.
Data Source
AI summary
A system including a secure, in-memory unit implemented on an associative processing unit (APU), for creating encrypted vectors. The in-memory unit includes a data store and an encryptor. The data store stores data and the encryptor encrypts the data into an encrypted vector. Optionally, the unit includes a neural proxy hash encoder that encodes the data into an encoded vector, and, in this embodiment, the encryptor encrypts the encoded vector into an encrypted encoded vector. The neural proxy hash encoder includes a trained neural network which includes a plurality of layers that encode the data into feature sets. The trained neural network encodes image files, audio files, or large data sets. The APU is implemented on SRAM, non-volatile, or non-destructive memory.