Evolution of requirements models by simulation

Abstract

Simulation is a common means for validating requirements models. Simulating formal models is state-of-the-art. However, requirements models usually are not formal for two reasons. Firstly, a formal model cannot be generated in one step. Requirements are vague in the beginning and are refined stepwise towards a more formal representation. Secondly, requirements are changing, thus leading to a continuously evolving model. Hence, a requirements model will be complete and formal only at the end of the modeling process, if at all. If we want to use simulation as a means of continuous validation during the process of requirements evolution, the simulation technique employed must be capable of dealing with semi-formal, incomplete models. We present an approach how we can handle partial models during simulation and use simulation to support evolution of these models. Our approach transfers the ideas of drivers, stubs, and regression from testing to the simulation of requirements models. It also uses the simulation results for evolving an incomplete model in a systematic way towards a more formal and complete one.