New Voters for Redundant Systems

Publisher Website
Google Scholar
Abstract
This paper describes a class of redundant voters which compute and provide a “correct” output signal given a set of redundant input signals. The output is determined as a continuous smooth function of the redundant inputs using a weighted “average.” This type of voter tends to minimize switching transients since the isolation of faulty signals is achieved through a continuous numerical weighting. Equalization schemes which are normally required in voter designs to inhibit adverse transients during a channel failure are negated by this method. This function approach to designing a voter and fault isolation system provides for direct calculation of the characteristics of the output signal, given known characteristics of the inputs. Using this proposed method, a voter can be designed to provide a weighted average of the remaining good signals of a redundant system in the presence of one or more failed signal channels. Computed examples of voter output for assumed values of voter input signals are given for one representative candidate voter when used in configurations for both triply redundant and quadruply redundant systems.

This publication has 0 references indexed in Scilit: