Data Masking System Using Random Index Rotation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional data masking techniques face challenges in achieving consistency and security, particularly due to the use of static maps which are insecure and time-consuming, and dynamic maps which lead to inconsistent masked data versions, making it difficult to securely protect sensitive data.
Innovation Solution
A processor-implemented method and system that utilizes a random key and regular expressions to categorize sensitive data elements, generate random and masked indexes, and perform rotation operations on matrices to create and unmask sensitive data, ensuring consistent and secure data masking with the same format across different versions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If static maps are used for data masking, then the original form of sensitive data can be protected, but the processing time increases as the input dataset increases
Solution Approach 1:
The patent divides the masking process into two distinct phases: a precomputation phase where the static map is generated once from the input dataset, and a masking phase where the static map is applied to multiple datasets. This segmentation allows the computationally intensive operations to be performed only once, rather than repeatedly for each dataset, thus reducing processing time while maintaining protection reliability.
Solution Approach 2:
The patent performs preliminary actions by precomputing the static map before actual masking operations. The static map is generated in advance based on the input dataset characteristics, and this precomputed map is then reused for masking multiple datasets. This preliminary computation eliminates the need to perform complex analysis during each masking operation, significantly reducing processing time.
2Productivity
If static maps are used for data masking, then processing can be efficient, but the original input dataset may be predicted through the static maps reducing security
Solution Approach 1:
The patent introduces a static map as an intermediary between the original dataset and the masked output. This static map acts as a mediator that transforms data without revealing the original values. The static map is designed to be non-invertible, meaning that even though it is reused for efficiency, it cannot be used to predict or reconstruct the original input dataset, thus maintaining security while enabling efficient processing.
Solution Approach 2:
The patent creates a static map that is a derived representation of the input dataset characteristics, but not a direct copy. This static map captures patterns and structures needed for consistent masking while deliberately omitting information that would allow reconstruction of original data. The static map is a simplified, one-way representation that enables efficient masking operations without compromising security.
3Reliability
If dynamic maps are used for data masking, then security is improved, but consistency in masked dataset cannot be achieved across different iterations
Solution Approach 1:
The patent merges the advantages of both static and dynamic approaches by combining a precomputed static map with dynamic masking operations. The static map provides consistent transformation rules that ensure the same input always produces the same masked output, while the actual masking operation remains dynamic and secure. This combination achieves both consistency across iterations and security against prediction attacks.
Data Source
AI summary
This disclosure relates generally to systems and methods for masking and unmasking of sensitive data. The present systems and methods solve the problems of consistency of the data masking, by using a random index and a masked index with use of regular expression concept. An additional random key produce different masked data versions, however the original form of the sensitive data is achieved with any masked data version. Plurality of masked data versions are generated by masking the sensitive input dataset, where the plurality of masked data versions comprises same format of the sensitive input dataset. The generated masked data versions are secured and hard to predict the original form of the sensitive input dataset by authorized or unauthorized environments. Also, the present method consume less processing time, as the masking process and the unmasking process make operations on the indexes rather than with the original dataset.