The analysis of one dimensional multiple-valued linear cellular automata

The authors present an analysis of multiple-valued linear cellular automata (CA) and their properties over GF(q). An application for pseudorandom pattern generation over a finite alphabet is discussed. For these cellular automata, the legal computational rules are defined and classified. The desired cellular automata must also have a maximal-length cycle in their state transition graph. An efficient recurrence relation and similarity transformations are presented. Three separate methods are outlined to produce minimal-cost CA with the above properties.<
>