Random Number Generator Skewness Control via Memory Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing random number generators often exhibit skewness, leading to statistical dependences between bits due to deviations in the probability of 0 and 1 values from the ideal ½, which complicates achieving statistically independent and evenly distributed random numbers, especially when physical noise sources require control to maintain constant generation probabilities.
Innovation Solution
A device and method that store generated bits in memory regions associated with the difference between 0 and 1 counts, allowing for algorithmic post-processing only when a predetermined number of bits is reached, ensuring statistical independence and constant probability, thus eliminating skewness without reducing entropy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If skewness control is implemented by changing generator parameters to correct bit frequency deviation, then the probability of 0 and 1 values approaches 1/2, but statistical dependences arise between bits
Solution Approach 1:
The patent segments the bit stream into multiple parallel lanes, each processed independently through its own skewness control mechanism. This segmentation isolates the statistical dependences to individual lanes while maintaining overall randomness quality, resolving the contradiction between skewness correction and statistical independence.
Solution Approach 2:
The patent introduces an intermediary processing stage between the physical random number generator and the output, where skewness control is applied. This intermediary layer corrects the bit frequency deviation without directly modifying the physical generator, thereby maintaining the statistical independence property while achieving accurate 50/50 bit distribution.
2Device complexity
If physical noise sources are used for random bit generation, then hardware complexity is reduced, but skewness and deviation from ideal 1/2 probability occur
Solution Approach 1:
The patent implements self-service skewness control where the system automatically monitors and adjusts its own bit distribution without external intervention. The skewness control mechanism uses feedback from the generated bit stream to dynamically adjust parameters, allowing the physical noise source to maintain ideal probability distribution while keeping the overall system simple.
Solution Approach 2:
The patent dynamically changes operational parameters of the physical noise source based on detected skewness in the generated bit stream. By adjusting parameters such as sampling rates, threshold voltages, or oscillator frequencies in response to measured deviations, the system maintains 1/2 probability distribution while preserving the simplicity of physical hardware generation.
Data Source
AI summary
Random numbers can be generated in a statistically independent manner and with identical probability if the bits generated by a controlled bit generator are stored by a storage in a plurality of memory regions, wherein the bits are each stored in such memory regions associated with a difference of the bits of the values 1 and 0 generated up to the time of storage, and if all the bits stored within a memory region are subjected to algorithmic post-processing as soon as a predetermined number of bits within a memory region is exceeded. The fact that the bits are not stored and processed sequentially, i.e. in the order of generation, allows using a sequence of bits within which the individual bits are statistically independent of one another for the algorithmic post-processing. Thus, a way of performing algorithmic post-processing of the bits without reducing the entropy is provided.