Full Text Available
Note: Clicking the button above will open the full text document at the original institutional repository in a new window.
Feature-based generation of pervasive systems architectures utilizing software product line concepts
As the need for pervasive systems tends to increase and to dominate the computing discipline, software engineering approaches must evolve at a similar pace to facilitate the construction of such systems in an efficient manner. In this thesis, we provide a vision of a framework that will help in the...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Published: |
AUC Knowledge Fountain
2012
|
| Subjects: | |
| Tags: |
No Tags, Be the first to tag this record!
|
| Summary: | As the need for pervasive systems tends to increase and to dominate the computing discipline, software engineering approaches must evolve at a similar pace to facilitate the construction of such systems in an efficient manner. In this thesis, we provide a vision of a framework that will help in the construction of software product lines for pervasive systems by devising an approach to automatically generate architectures for this domain. Using this framework, designers of pervasive systems will be able to select a set of desired system features, and the framework will automatically generate architectures that support the presence of these features. Our approach will not compromise the quality of the architecture especially as we have verified that by comparing the generated architectures to those manually designed by human architects. As an initial step, and in order to determine the most commonly required features that comprise the widely most known pervasive systems, we surveyed more than fifty existing architectures for pervasive systems in various domains. We captured the most essential features along with the commonalities and variabilities between them. The features were categorized according to the domain and the environment that they target. Those categories are: General pervasive systems, domain-specific, privacy, bridging, fault-tolerance and context-awareness. We coupled the identified features with well-designed components, and connected the components based on the initial features selected by a system designer to generate an architecture. We evaluated our generated architectures against architectures designed by human architects. When metrics such as coupling, cohesion, complexity, reusability, adaptability, modularity, modifiability, packing density, and average interaction density were used to test our framework, our generated architectures were found comparable, if not better than the human generated architectures. |
|---|