Output Grouping Method Based on a Similarity of Boolean Functions

A method allowing us to efficiently group the multi-output Boolean function outputs is presented in this paper. Some kind of decomposition is usually involved in the logic synthesis process. Here the circuit has to be repeatedly divided into subcircuits, until these subcircuits become technological library elements or technology-dependent components, in general. Our output-grouping method can be used to compute which functions of a set of Boolean functions could share most of logic. This idea can be exploited in single-level decomposition, where a two-level multi-output circuit is divided into several stand-alone blocks, while retaining their two-level nature. The total size of these circuits should be kept minimal. After such single-level decomposition the circuits are further processed by a common synthesis process. Obtained results are shown in this paper, for a set of standard MCNC and ISCAS benchmarks. Two general problems to be solved were considered for tests: an output-grouping based decomposition and a parity predictor synthesis, which is used in an on-line diagnostic design.