From natural language standard documents to state machines: advantages and drawbacks

Abstract

Problems in requirements documents are among the root cause of a number of accidents in space missions. A common approach toward the minimization of these problems is to transform the requirements into models that represent the system’s behavior. However, this solution requires dealing with issues such as choosing the best modeling formalism, defining to what extent the transformation process should be automated, and assuring the quality of the requirements documents to be used as input. In space missions, requirements are frequently tailored from standard documents, such as the Packet Utilization Standard, which are composed of mandatory and optional requirements. This paper presents a semi-automatic method to transform standard requirements documents into extended finite state machines. To evaluate it, we apply the method to a set of requirements from the Packet Utilization Standard. We evaluate the method using some Packet Utilization Standard services. In light of the results, the paper discusses advantages and potential problems of each approach.