Implementation of multiple-valued functions using literal-splitting technique

Abstract

One of the main practical problems of implementing multiple-valued logic circuits is reduced noise margins. In this paper we show that this problem can be avoided by implementing multiple-valued functions using the following technique. First, the input domain of an m-valued n-variable function f is partitioned into m m-valued input binary-valued output characteristic functions. Then, each of these functions is treated as a function of n/spl middot/m binary variables, representing the literals of the variables off. The fact that any multiple-valued function can be implemented that way is known. The main contribution of this paper is to show a technology, for which this type of implementation is natural and results in no hardware overhead.