Software Quality Specification Using Markov Chain Statistical Technique
D. G. Iyanda, L. A. Ogundele, T.A Mojeed · IOSR Journal of Computer Engineering · 2017
Informal software engineering methods have been using combination of diagrams, text, tables and simple notation to create analysis and design models with application of little mathematical rigour.On the other hand, formal methods allow a software engineer to create a model that is more complete, consistent, and unambiguous than those produced using conventional methods (Roger, 2005).However, it had been observed that the fundamental step in the Markov analysis of a software specification is to define the underlying probability law for the usage of the software under consideration.From an analytical point of view, this is a traceable stochastic process and a good basis for statistical testing.Also, from the software engineering point of view, definitions and properties of the model ensure the testability of the software.All the available models have their drawbacks and none of them used mathematical method, hence our Markov chain statistical approach.In this approach, the analysis of the specification, performed prior to design and coding, yields an irreducible Markov chain called Usage Markov Chain Statistics.Each usage state is labelled with a stimulus from the input domain of the software.If the control is exerted depend both on the position at time t, and position at time t-1, then , there is succession of state such that the dependency of the past state reaches a fixed distance into the past, for instance, if it reaches back three states, then it forms a Markov Chain statistical approach.