MRAM Cell True Random Number Generation via Probability Trimming
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current electronic devices rely on pseudorandom number generators (PRNGs) which are not truly random, lacking the security and reliability needed for secure data storage and transmission, particularly in cryptographic applications.
Innovation Solution
The integration of magnetoresistive random-access memory (MRAM) cells into integrated chips (ICs) to generate true random numbers by switching resistive states with a probability of 0.5, utilizing a probability trimming series to determine optimal switching values for RNG signals, thereby reducing manufacturing complexities and costs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If PRNGs are used to generate random numbers, then the device complexity is reduced and manufacturing costs are lowered, but the reliability and security of the random numbers deteriorate
Solution Approach 1:
The patent replaces software-based PRNG algorithms with a hardware-based physical process (magnetic tunnel junction resistance switching) to generate true random numbers. This substitution eliminates the fundamental limitation of PRNGs (dependence on seed values) while using a well-established MRAM technology, thereby improving reliability without requiring entirely new complex hardware architecture.
Solution Approach 2:
The patent makes MRAM cells serve dual functions: their primary function for data storage and an additional function for true random number generation. By utilizing the inherent stochastic nature of magnetoresistive switching in MRAM cells, the same hardware component performs both memory operations and cryptographic random number generation, eliminating the need for separate dedicated RNG hardware and reducing overall device complexity.
2Reliability
If MRAM cells are used for random number generation, then true random numbers are generated with improved security, but the manufacturing precision requirements increase
Solution Approach 1:
The patent performs preliminary characterization and calibration of MRAM cells during the manufacturing process to determine individual cell switching thresholds and probabilistic behaviors. This preliminary action enables the system to compensate for manufacturing variations by adapting control parameters to each cell's specific characteristics, thereby achieving reliable 50/50 random switching despite manufacturing precision variations.
Solution Approach 2:
The patent implements feedback mechanisms where the system monitors and measures the actual switching behavior of MRAM cells during operation and adjustment phases. Based on this feedback, the system dynamically adjusts write pulse parameters (amplitude, duration, shape) to maintain optimal switching probability close to 0.5, compensating for manufacturing variations and ensuring high-quality random number generation without requiring extremely tight manufacturing tolerances.
3Ease of manufacture
If probability trimming series is implemented, then the manufacturing cost is reduced, but the time required for optimization increases
Solution Approach 1:
The patent performs probability trimming and optimization of MRAM cell switching parameters during the manufacturing process itself, rather than after deployment. By conducting characterization and calibration tests during factory production, the system determines optimal control parameters for each cell or cell group in advance, enabling mass production with standardized parameters and reducing or eliminating the need for time-consuming post-manufacturing optimization and field tuning.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach provides a secure and reliable method for generating true random numbers, enhancing data security and reducing errors in electronic devices without increasing manufacturing complexities or costs.
Implementation Method 1
magnetoresistive random-access memory (MRAM) cell configured to generate a random bit... switching resistive states with a probability of 0.5
Data Source
AI summary
In some embodiments, a method for generating a random bit is provided. The method includes generating a first random bit by providing a random number generator (RNG) signal to a magnetoresistive random-access memory (MRAM) cell. The RNG signal has a probability of about 0.5 to switch the resistive state of the MRAM cell from a first resistive state corresponding to a first data state to a second resistive state corresponding to a second data sate. The first random bit is then read from the MRAM cell.