Model-Based Verification: Abstraction Guidelines

ion is used to reduce the complexity of a model by including only the parts of the system necessary for the issues being investigated. The goal of abstraction is to prune away unnecessary detail. This enables the modeler to explore, substantiate, or disprove intended behaviors of a system while maintaining the validity of the model [Frantz 95]. Engineers and scientists routinely use abstraction in problem solving. This section of the technical note presents some abstraction techniques and relevant examples. The techniques discussed are: • variable elimination • enumeration • reduction • non-determinism • grouping based on commonalities • decomposition Various stages of abstraction and modeling of a system are generally referred to as being representative of a ‘level of abstraction.’ The convention for these levels is that the removal of detail results in a higher level of abstraction whereas adding detail results in a lower level of abstraction. The term ‘granularity’ is often used to refer to the level of detail. Levels of abstraction and levels of granularity are often used interchangeably. Variable elimination [Heitmeyer 98, Bharadwaj 99] removes parts of the system that are not relevant to the properties and behavior to be demonstrated or proven. Irrelevant variables can be identified by looking at dependencies and then removed. Again, consider a process control computer system and its applications software. Let’s assume we are investigating the