Fault secure multiple-valued logic networks
James E. Smith, Jean Dussault · International Symposium on Multiple-Valued Logic · 1978
Fault secure multiple-valued logic networks have their outputs encoded in an error-detecting code so that assumed failures either result in no output error or in a detectable (noncode) output. In this paper we discuss a way of constructing fault secure combinational networks as well as a generalization that we define to be “strongly fault secure.” The only constraint on the networks themselves is that they be made of positive unate gates and use unordered output encodings.The fault assumptions are quite general; they include stuck-at faults as well as many other failures which seem reasonable in a multiple-valued technology. Protection is provided against both permanent and intermittent faults. Various unordered codes and their properties are discussed as is the construction of check circuits that are themselves fault secure or strongly fault secure.