Modified GA method for variable ordering in BDD for MIMO digital circuits

Abstract

Reduced Ordered Binary Decision Diagram (also called ROBDD or simply BDD) are graphical representation of Boolean functions, which can be easily manipulated using modern VLSI CAD tools. The order of input variables has a considerable effect on the size of the constructed BDD. Many algorithmic approaches have been proposed to find the optimal variable ordering. In this paper, a new modified genetic algorithm based method for variable ordering is proposed which use different modified crossover and modified mutation operators. The results have shown that the proposed method yields a good improvement in the size of the BDD constructed for many digital benchmark circuits.