Proceedings of the 2002 ACM symposium on Applied computing
| Hi-index | 0.00 |
Requirements are descriptions of the application domain, of the problems to be solved there, and of the system(s) to be built in that domain to solve these problems. Many projects have failed because their requirements were inadequately analyzed or described. Use cases have offered analysts and developers an approach for capturing the end-user requirements in an object-oriented fashion. However, experience is showing some problems with use cases such as lack of a precise definition that led most companies to re-invent their own version, lack of notion of atomicity, and lack of addressing all interaction types. This research is concerned with developing a more effective alternative to Use Case Analysis (UCA) for modeling the functional requirements of Human-Machine Safety-Critical Real-time Systems. The new alternative modeling approach, Behavioral Pattern Analysis (BPA), is an Event-Oriented approach in which events are considered the primary entities of the world domain requirements model. While the term 'event' is used in UML, and in almost all of the other modeling approaches, to mean an occurrence of stimulus that can trigger a state transition, the Event described in BPA is a real-life conceptual entity that is unrelated to any implementation. It describes What, Who, How, When, Where, and Why for an interaction or set of interactions between entities. The appeal to events is natural in the definition of requirements because events provide more complete and clearer understanding of the defined requirements. Such completeness and clarity are keys to the effectiveness of requirements modeling. It is the researcher's belief that the system model should be event driven rather than use case driven. Three real-life applications were used to validate the effectiveness of the new BPA approach in modeling the functional requirements of safety-critical real-time systems. Both, the UCA and the BPA approaches were used to model the functional requirements of these applications. To validate the BPA approach improved effectiveness, the first application was used in a pilot case study by the researcher to provide a proof of concept, and the other two applications were used in another sixteen case studies that were arranged for the evaluation purpose. In all of these case studies, the resulting models from both approaches were compared using safety, CMM repeatability, and the ANSI/IEEE Std 830-1984 effectiveness standards as the basis of the comparison. The case studies result indicated that at least eighty seven percent (87%) of the participating Subject Matter Experts (SMEs) believe that BPA is more effective than UCA in the definition of requirements of safety-critical and real-time systems. (Abstract shortened by UMI.)