A Novel Method for Improving the Uniformity of Random Number Generator Based on Data Oriented Modeling

In this paper a new method is introduced for generating uniform random numbers based on data oriented modeling. This modeling method of random variable is proposed in [1]. By using this method, uniform random variable is modeled with several data. Up to now, statisticians have used mathematical functions to model random variables. This approach conforms to human brain structure, in other words it is easier for human to use mathematical functions for modeling. In this approach for modeling random variable X, distribution function F(x) is denoted its scattering. In our approach random variable X is modeled with data structure, sizeable number of data. Data-oriented model of uniform random variable proposed in [1] and named UDPD. UDPD is a weighted digraph which is explained in detail in next section of this paper. The introduced method in this paper is a simulation method of UDPD model. This simulation method based on its result named uniformity improving method, or UIM in short. The implementation results show that the good uniformity of numbers generated with UIM in comparison with Matlab’s rand. Basic definition to outline UDPU and implementation of UIM is explained in section 2. In the third section non-uniformity of current Matlab’s method and UIM is compared by calculating two evaluation factors and it is shown that uniformity of random numbers generated by UIM is better than Matlab’s method.