Pseudonymized Data Storage With Separated Key Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data security solutions face challenges in providing comprehensive protection for sensitive data across diverse technologies and data formats, with privacy-on-read approaches being complex and computationally expensive, and privacy-on-write methods lacking efficient data recovery and analysis capabilities.
Innovation Solution
Implementing a privacy-on-write approach that combines pseudonymization and encryption to protect sensitive data, using deterministic hashing and symmetric encryption, storing data in transformed and encrypted forms, and maintaining encryption keys separately to ensure data integrity and recovery.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If privacy-on-read approaches are implemented via table or SQL views, dynamic masking, table and field permissions, use of proxies, or access controls, then access to sensitive data is protected at access, read, or query time, but the creation and enforcement of role-based access control permission models becomes extremely complex
Solution Approach 1:
The patent applies preliminary action by implementing privacy-on-write, where data is anonymized or pseudonymized at the time of writing/storing rather than protecting it at read time. This transforms the complex access control problem into a simpler transformation problem that can be applied uniformly across all data formats and technologies.
Solution Approach 2:
The patent extracts the privacy protection mechanism from the access control layer and places it at the data writing layer. By separating the anonymization/pseudonymization function from the access control logic, the system eliminates the need for complex role-based access control permission models while maintaining data protection.
2Reliability
If physically removing data is used to delete sensitive data in privacy-on-read models, then data is completely removed from storage, but the process becomes slow and computationally expensive requiring scanning and searching every record in every dataset
Solution Approach 1:
The patent uses copying by creating transformed copies of sensitive data through anonymization and pseudonymization. Instead of physically removing original data, the system creates protected versions that can be deleted efficiently through metadata updates or transformation reversal, avoiding the need to scan and search every record.
Solution Approach 2:
The patent applies parameter changes by transforming data parameters through anonymization and pseudonymization functions. Deletion becomes a matter of changing the transformation parameters or removing references to transformed data, rather than physically removing each record, dramatically improving deletion speed and efficiency.
3Reliability
If data is anonymized or pseudonymized at an early stage to protect security, then sensitive data is inherently protected and can be accessed by any technology, but sensitive data should still be recoverable in a protected way
Solution Approach 1:
The patent segments the data protection system into multiple layers: anonymization for full protection, pseudonymization for protected recoverability, and controlled access mechanisms. This segmentation allows different levels of data protection and recoverability depending on the specific use case and security requirements.
Solution Approach 2:
The patent uses composite approaches by combining anonymization and pseudonymization techniques with controlled access mechanisms. The system creates a composite protection model where data can be both protected and recoverable through the coordinated use of multiple transformation and access control methods.
Data Source
AI summary
Techniques relating to data privacy are disclosed. These techniques include identifying data fields relating to a first data record, and generating transformed data fields by transforming the data fields using one or more transformations. Each of the transformations includes at least one of: (i) a hash function or (ii) a surrogate key function. The techniques further include generating encrypted data fields by encrypting the data fields using a respective encryption key, and storing the transformed data fields, the encrypted data fields, and of encryption keys. This includes storing the transformed data fields in one or more first storage locations, storing a first mapping between each of the transformed data fields and a corresponding encrypted data field in one or more second storage locations, and storing a second mapping between each of the transformed data fields and a corresponding encryption key in one or more third storage locations.