Searchable Encryption Scheme with External Tokenizer for Encrypted Database Queries
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current encryption schemes for sensitive data in relational databases break applicative logic, such as search and pattern matching functions, and do not support order-preserving and pattern matching, posing challenges in protecting personally identifiable information (PII) while allowing privileged users to perform their jobs without accessing plaintext data.
Innovation Solution
A searchable encryption scheme with an external tokenizer is implemented at the JDBC layer, which maps plaintext strings to encrypted values and supports SQL operations like LIKE queries without modifying application code, ensuring data security by maintaining encryption even during searches and sort operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If current encryption schemes are applied to sensitive data in relational databases, then data security is improved, but applicative logic functions such as search and pattern matching are broken
Solution Approach 1:
The patent introduces an external tokenizer as an intermediary component that maintains a mapping between plaintext tokens and their encrypted representations. This tokenizer sits between the application and the encrypted database, enabling search operations by translating plaintext search terms into encrypted form without requiring decryption of the stored data, thus preserving both security and search functionality
Solution Approach 2:
The system segments the encryption functionality into two distinct parts: deterministic encryption for the database storage layer and probabilistic encryption with an external tokenizer for the search layer. This segmentation allows each layer to be optimized independently - the database layer maintains strong security through deterministic encryption while the search layer enables pattern matching through the tokenizer's mapping mechanism
2Ease of operation
If deterministic encryption is used to support search operations on encrypted data, then pattern matching capability is improved, but order-preserving capability is lost
Solution Approach 1:
The external tokenizer acts as a mediator that handles pattern matching operations separately from the encrypted storage layer. It maintains a plaintext-to-encrypted mapping that enables LIKE queries and other pattern matching operations without requiring the encrypted data itself to preserve order, thus allowing deterministic encryption to be used while maintaining pattern matching through the intermediary tokenizer
3Reliability
If encryption is applied to PII data in databases, then protection against data leakage is improved, but access control for privileged users is worsened
Solution Approach 1:
The external tokenizer serves as a controlled intermediary that privileged users must interact with to perform searches on encrypted data. Users can execute search operations through the tokenizer without gaining access to plaintext data, maintaining security while enabling operational access. The tokenizer's mapping mechanism allows authorized operations without compromising the encrypted storage
Data Source
AI summary
A server receives a first query to perform one or more operations on an encrypted database and intercepts the first query. A set of data referenced by the first query is determined to include data that is to be encrypted. Based on metadata, one or more data columns of the data that is to be encrypted are identified as well as an encryption scheme to be applied to each of the data columns. Each of the identified data columns is encrypted based on the identified encryption scheme to produce encrypted data. One or more of the operations of the first query is determined to be unsupported on encrypted data. The one or more operations are executed on a token vault at a tokenizer to retrieve token data referencing the encrypted data. The first query is replaced with a modified query based on the retrieved token data. The modified query is executed on the encrypted database to retrieve an encrypted query result.