A methodology for object-oriented modeling and design of real-time, fault-tolerant systems

Many methodologies for software modeling and design include some form of static and dynamic modeling to describe the structural and behavioral views respectively. Modeling and design of complex real-time software systems requires notations for describing concurrency, asynchronous event handling, communication between independent machines, timing properties, and accessing real time. Function-oriented structured analysis methodologies such as Ward and Mellor's SA/RT and Harel's Statecharts have provided extensions for real-time system modeling. Dynamic modeling of real time systems using object-oriented methodologies also requires extensions to the traditional state machine notations in order to convey the real time system characteristics and constraints. Shaw's Communicating Real Time State Machines (CRSM's), Harel's O-Chart notations, and the Octopus methodology provide methods for modeling real-time systems consistent with object-oriented methods. This thesis proposes an object-oriented analysis and design methodology that augments the traditional Object Modeling Technique (OMT) dynamic model with real-time extensions based on high-level parallel machines and communication notations from CRSM. An example of the proposed methodology is provided using a realistic but hypothetical example of an automated passenger train system. A design refinement step is included for fault tolerant considerations. An evaluation of the proposed methodology with its extended notations is provided.