Markov analysis of software specifications

Abstract

A procedure for modeling software usage with the finite state, discrete parameter Markov chain is described. It involves rigorous analysis of the specification before design and coding begin. Many benefits emerge from this process, including the ability to synthesize a macro level usage distribution from a micro level understanding of how the software will be used. This usage distribution becomes the basis for a statistical test of the software, which is fundamental to the Cleanroom development process. Some analytical results known for Markov chains that have meaningful implications and interpretations for the software development process are described.