Optimal practical adders using perceptrons

Networks of perceptrons provide simple and elegant solutions for many classes of Boolean functions, including addition. That is why, a large number of solutions have been proposed for the addition of two binary numbers. Such solutions have been optimized with respect to several cost functions like: depth, size, fan-ins, weights, and thresholds. In this paper it is also reviewed that constructive solutions for the addition of two binary numbers, and subsequently introduce a novel optimal hybrid structure for threshold logic practical adders (i.e., adders having from 16- to 128-bit inputs). Quantitative comparison of the new structure with the latest solutions in terms of depth, size, weights, thresholds, fan-in, as well as fan-out, is presented for a particular case.