Sang - Ho Shin 1 and Kee - Young Yoo

Cellular automata (CA) were initiated in the early 1950s to develop complex structures of capable self-reproduction and self-repair. Since thenmany researchers have taken attend in the study of CA. Initially, CA concept was first introduced by von Neumann von Neumann (1966) for the proposal of modeling biological self-reproduction. His primary interest was to derive a computationally universal cellular space with self-reproduction configurations. Afterward, a new phase of activities was started by Wolfram Wolfram (1983; 1984), who pioneered the investigation of CA as a mathematical model for self-organizing statistical systems. Wolfram was proved that the randomness of the patterns generated by maximum-length CA is significantly better than other widely used methods, such as linear feedback shift registers. The intensive interest in this field can be attributed to the phenomenal growth of the VLSI technology that permits cost-effective realization of the simple structure of local-neighborhood CAWolfram (1986). The PRNGs based on CA have been studied for the last decade and a variety of CA PRNGs have been proposed Guan & Tan (Jul. 2004); Guan & Zhang (Feb. 2003); Seredynski et al. (2004); Tan & Guan (2007); Xuelong et al. (Oct. 2005). Recent interest has been focused on the two-dimensional (2-D) CA PRNGs with genetic algorithm Chowdhury et al. (1993); Guan et al. (2004); Quieta & Guan (2005); Tomassini et al. (2000), since, statistically, the point has been established that the quality of randomness of a 2-D CA is significantly better than a 1-D CA. In this paper, an efficient PRNG based on hybrid between one-dimension (1-D) and two-dimension (2-D) CA is proposed. In the phase of the evolution of 2-D CA cells, the proposed CA PRNG is based on von Neumann neighborhood, in that this method refers to the five cells and new control values (rule decision input value & linear control input value) to decide a rule. In the phase of the evolution of 1-D CA cells, on the other hand, <90, 150> rule combination is used because this rule combination is better than the others Hortensius et al. (1989). In the meantime, the proposed CA PRNG is compared with previous works Guan et al. (2004); Tomassini et al. (2000) to check the quality of randomness. The proposed CA PRNG could generate a good quality of randomness because the proposed CA PRNG is better than Sang-Ho Shin1 and Kee-Young Yoo2 1School of Electronic Engineering and Computer Science, Kyungpook National University Gyengdaejeongmunro 67, Buk-Gu, Daegu, 702-701 2School of Computer Science and Engineering, Kyungpook National University Gyengdaejeongmunro 67, Buk-Gu, Daegu, 702-701 South Korea An Improved PRNG Based on the Hybrid between Oneand TwoDimensional Cellular Automata 15