2019 IEEE 49th International Symposium on Multiple-Valued Logic (ISMVL)

1 / 2018-11-06 09:55:57

Approximate Reversible Logic Synthesis

Reversible Logic,Approximate Computing,Boolean Matrix Factorization,Quantum Cost

Approximate reversible logic synthesis generates inexact implementations of reversible functions in exchange for benefits in design qualities such as quantum cost and circuit complexity. In spite of all that, the synthesis of approximate reversible circuits has not yet been studied. In this paper, we present a novel approximate reversible logic synthesis approach based on Boolean matrix factorization (BMF) using bit-vector encoding that select Column Subset (CSS). The basic idea of our approach is to pick columns in the truth table of a reversible circuit and approximate using BMF, so it is able to approximate complete reversible circuits. Experimental results demonstrate the approach can produce approximate reversible circuits with complexity reduction ranging from 23% to 49% under small error constrains.