Software Plug and Play

Abstract

This thesis considers Software Plug and Play (PNP) as a potentially useful architectural abstraction (AA) that can solve many software architecture (SA) problems faced by vertical markets in the industry. While PNP as a technology has been utilized to address many extensibility concerns in hardware, there is no systematic support for similar extensibility notion at the SA level. To address these concerns, this thesis proposes the creation of frameworks and component models to support PNP. Accordingly, this thesis considers PNP as a user-defined framework, supported by a run-time component library that is embedded in the component schema, in order to support functional and structural extensibility of a running software system. newlineAs practitioners, it was decided that the inductive path of Clayton Christensen s approach for theory building would be an appropriate way to build the architectural foundations for PNP. Accordingly, this approach was adopted to empirically study six graded systems that had decreasing requirements on their environments. Through this empirical and experimental work, the notion of software components, components composition and SA was illustrated. Further, the architecture principles that were adopted to enable PNP in these systems aided in identifying the corresponding PNP AA. Based on this experimental experience, this thesis contributes the following to the field of computer science and SA: newline1. The concept that PNP is a collection of standards, encoded in the run-time components and embedded in the component schema, for supporting extensibility of a running software system. newline2. The characterization of PNP in terms of underlying architecture principles, and corresponding AA. newline3. The formulation of a generic PNP component model, that supports this characterization, which when instantiated aids in realizing pure PNP architectures. newline4. An expression of the Plug and Socket interfaces, that are constituent elements of the PNP component model, by means of abstract data types.