The fundamental goal of software development is to deliver high quality products that are correct, consistent and complete, yet with low cost and time. As software systems continue to increase in size and complexity, achieving this goal becomes difficult. Much attention has recently been focused on the design and analysis of non-functional requirements (i.e., performance, security, adaptability, etc.) for software systems at the software architecture design level. The significant benefits of such work include detecting and removing defects earlier, reducing development time and cost while improving the quality.

My research proposes a framework, Formal Design Analysis Framework (FDAF), to effectively design and analyze non-functional requirements for software systems at the architecture design level. One of the difficulties of modeling non-functional requirements is that in general, a system's non-functional requirements tend to be global and it is relatively difficult to encapsulate them within model elements such as modules, classes, and operations at design level. To address this problem, the FDAF uses Aspect-Oriented Software Development (AOSD) techniques, where non-functional requirements are treated as crosscutting concerns and encapsulated in modular units called aspects. The base design notation used in the FDAF is the well-established Unified Modeling Language (UML). The FDAF also extends the UML to support aspect-oriented design and aims to provide designers a repository of reusable aspects to use. However, without a precise semantics, the UML lacks the ability to formally represent a system's specification and consequently, to develop rigorous tool supports for automated analysis. This has driven the FDAF to investigate existing formal methods, which are mathematical approaches to software development and they support the rigorous specification, design, and verification of system properties. The FDAF supports the automated translation from a UML design into other formal specifications, thus a UML-based design with non-functional requirements can be analyzed in existing formal tools. Currently, the design and analysis of two performance aspects, response time and resource utilization, are supported. The formal notations used for their analysis are Architecture Description Languages Rapide and Armani, respectively.

As a system's non-functional requirements may conflict with each other, i.e., higher security leads to lower performance, the long term goal of the FDAF is to support aspect-oriented design and analysis of multiple, (possibly) conflicting non-functional requirements.

Annie (Lirong) Dai is a PhD candidate at the University of Texas at Dallas. She received her M.S. degree in Computer Science from the University of Texas at Dallas and her B.S. in Computer Science from Sichuan Union University in China. Her research interests span a range of topics in software engineering including software architecture development, the design and analysis of non-functional requirements in the software development lifecycle, and Aspect-Oriented Software Development. [an error occurred while processing this directive]