Analytical derivation of software failure regions

Abstract

This paper proposes an analytical method for deriving software failure regions, which are regions of the input space that are mapped to failures by specific faults. Previous studies have used empirical rather than analytical approaches to derive failure regions. A manual technique is presented and proven to produce the necessary and sufficient conditions of a fault being executed and leading to a failure. Semiautomated tools to assist in the manual technique are discussed, as is the use of failure regions in regression testing