Non-linear Token Mapping for Dictionary Encryption Security
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing encryption dictionaries used in tokenization algorithms are vulnerable to attacks where an attacker can estimate the first letter of a plaintext word from the token value, compromising security, especially in multitenant environments where sensitive data needs to be protected.
Innovation Solution
A method for generating an encryption dictionary that uses a non-linear mapping technique, increasing token space for frequently occurring letters and decreasing it for less frequent ones, while preserving the ability to perform sorting and searching operations on encrypted data, by adding random constant values and using a coin flipper to introduce steps and varying slopes in the token-value relationship.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If a standard encryption dictionary maps words to random tokens, then the encryption process is simple and efficient, but an attacker can derive a course estimation of the first letter of a plaintext word from the token value
Solution Approach 1:
The patent applies local quality by making the token space distribution non-uniform, specifically tailoring it to match the frequency distribution of letters in the plaintext language. Frequently occurring letters are assigned smaller token spaces, while less frequent letters receive larger token spaces. This localized adjustment of token space allocation based on letter frequency undermines the attacker's ability to estimate plaintext from token values, while maintaining efficient encryption operations.
2Object-affected harmful factors
If token space is increased for frequently occurring letters, then security against guessing attacks is improved, but the overall token space size increases
Solution Approach 1:
The patent changes the parameter of token space allocation from a uniform distribution to a non-uniform distribution that reflects letter frequency. By adjusting the token space size for each letter based on its frequency in the plaintext language, the system achieves better security against guessing attacks without requiring a proportional increase in overall token space. This parameter change optimizes the balance between security and resource usage.
3Object-affected harmful factors
If a non-linear mapping technique is used to enhance security, then the vulnerability of guessing plaintext is reduced, but the complexity of the encryption dictionary generation increases
Solution Approach 1:
The patent applies preliminary action by pre-calculating and storing the non-linear mapping between plaintext letters and token values in an encryption dictionary. This dictionary is generated once using the letter frequency distribution, and then reused for multiple encryption operations. The preliminary creation of this frequency-based mapping table reduces the complexity of individual encryption operations while maintaining enhanced security against guessing attacks.
Data Source
AI summary
Computer systems and applications are provided for encrypting data that preserves the ability to process the encrypted data. The method includes receiving data in unencrypted form. The method further includes encrypting the data in accordance with an encryption dictionary generated by arranging the plurality of plaintext symbols in lexicographical order; defining a first subset comprising a first plurality of the lexicographically arranged symbols; defining a second subset comprising a second plurality of the lexicographically arranged symbols; defining a first set with a first plurality of unique random tokens within a first token space for use with the first plurality of symbols; and defining a second set with a second plurality of unique random tokens within a second token space for use with the second plurality of symbols such that the second plurality of unique random tokens is non-linear with respect to the first plurality of unique random tokens.