Autocorrelation techniques for multi-bit decoder PLAs

Abstract

The area required by a PLA can be reduced using multibit rather than single-bit input decoders. In such a PLA, the input variables are partitioned into disjoint subsets, each of which is used as the input to a decoder. The output of the decoders, instead of the input signals and their complements, are used in the core of the PLA. The problem of assigned pairs of input variables to two-bit decoders is addressed. Variable pair selection is based on the total autocorrelation of a system of Boolean functions. Preliminary results on the use of autocorrelation coefficients in the assignment of triples of variables to decoders are also presented.<>