Requirements modelling of real-time systems.
Real-time systems are characterised by the critical nature of their missions, and the
demanding environment with which they interact. Real-time systems are used for
dedicated applications. Every application is the subject of special requirements
enforced by the customer. Considering the vital role that these systems play, it is
imperative that a systematic approach be adopted in modelling their unique
requirements. In this thesis I propose such a treatment.
Real-time systems are time critical. Temporal requirements are the timing
restrictions imposed by the application environment. Previous studies in
requirements modelling of real-time systems have focused on adding the notion of
time to modelling techniques of traditional systems without regard to the realities of
requirements modelling. The information should be presented in the way the user
handles it, and not the way which is convenient to the software engineer. I attempt
to understand the needs of the users better by modelling the real world as close to
the user's perspective as possible, and propose the Real World Model (RWM).
RWM is assumed to be developed by users, and requirements engineers. An
engineering approach to building the model is provided.
A real-time system has a well defined use to its community. A requirements model
must rely on the user level activities, and aid the human understanding and
communication. In the RWM, a real-time system is viewed as a set of concurrently
acting automata, each representing a system entity. This model supports temporal
reasoning in easily described ways, for all classes of timing properties. A
generalised classification of timing constraints is provided.
A requirements modelling language facilitates the description of requirements, and
serves as a medium of communication among developers and stakeholders.
Jarke et al [Jarke 94] observe that there is a need for a requirements language that
manages the relationship between the meta-level domain scheme, and the scenarios
that actually instantiate the scheme under development. Here I propose Timed
Requirements Language (TRL) to bridge this gulf between the world of
stakeholders, and the world of specifiers. TRL has natural looking expressions for
formulating the needs. TRL has a number of novel features including the treatment
of causality, and the description of static, and dynamic constraints all integrated into
one uniform framework. TRL has been used with a number of systems. The
generality of the language is validated through its application to specific systems.