Design and run-time requirements modelling for adaptive systems

This thesis explores the construction, enrichment and use of requirements models for adaptive systems. This thesis proposes the enrichment of adaptive systems' requirements models with additional tracing information, preserving the rationale behind configuration decisions. The preserved rationale can be utilised at design time, allowing decisions to be re-taken in the context of new or changed information and allowing for the identification of areas of uncertainty in understanding. The preserved rationale can also be used by a system itself, at run time, allowing it to adapt its behaviour to contexts not fully envisaged at design time. This thesis presents the ReAssuRE modelling process, which defines modelling perspectives from which some classes of adaptive system may be viewed. ReAssuRE models embed the described tracing information, and may be interpreted and reasoned with by a suitably constructed system at run time. This thesis presents tool support, allowing adaptive behaviour to be derived directly from ReAssuRE models. Finally, this thesis presents proof of concept components that allow a system to reason with ReAssuRE models, transforming them in response to monitored data, and derive new adaptive behaviour.EThOS - Electronic Theses Online ServiceGBUnited Kingdo