IMPROVEMENT OF STATISTICAL CHARACTERISTICS OF NUMERICAL RANDOM SEQUENCE

Abstract

At the present stage, a program generation of random numbers is at risk of being hacked due to the increasing computing power of modern systems. Hardware generation is based on reliable stochastic physical phenomena but provides low productivity or bad statistical parameters.   The work proposes to use a linear cellular automata processing for improvement of some statistical parameters of random number sequences, generated by web camera. 

  The investigation presents the results of improvement of the statistical characteristics of a sequence of numbers obtained from an ordinary webcam, in terms of compliance with one of the requirements of crypto resistance: uniform distribution of elements by value.   Previously it was found that stochastic processes occurring in the webcam matrix cause a chaotic distribution of the values ​​in a generated sequence of random numbers.  This obstacle can be overcome utilizing the processing power of linear cellular automata, especially rules 30, 90, 105. 150.   These   cryptoprimitives are known as chaotic ones that consume low computing power.  

The success of the application was assessed in comparison with the generation of random numbers by a software method, in particular a class SecureRandom of Java programming language.   It is shown that by choosing several iterations, it is possible to obtain the required level of uniformity of distribution of sequence elements by value.

Estimation the level of uniformity of a distribution is carried out quickly using the statistical library of the Java programming language and can be implemented on a regular smartphone, the Android operating system, without the use of cumbersome statistical packages.  The results of the study can be used in the design of a hardware random number sequence generator.