Software Architecture Evaluation Based On Evolution

With the expansion of software scale, software complexity rises and the design of the system framework is becoming more and more critical thus more and more people use architecture to develop certain software products. Architecture can significantly improve the efficiency of software development, reduce costs, ensure software quality and software reuse rate, only when the architectural design is reasonable without error. A good architecture is one of the key factors to develop a software smoothly, thus the research about the measurement of software architecture is more and more important.Software architecture is a product in the early stages of software development, which guides the subsequent software development work, with high abstractness, and the changes of architecture will directly affect the quality of all softwares which implements this architecture. Evolution is an activity throughout the modern software life cycle, and the user demand expansion, resources and environment change and technology development are likely to push architecture evolution. The evolution with rational operation can make the software architecture optimized, and further improve the quality of the software of application of the architecture. However, the evolution with bad operation may lower the software quality which is an application of this architecture, or even cause fatal flaw, so the research about the evaluation of software architecture evolution is of great significance.We propose a new quality attribute in this thesis, software architecture maturity, which is used to measure the level of software architecture close to mature architecture on the dynamic behavior and static structure. On this basis, we design a method to measure evolutionary maturity. This method takes the architecture description documents as input, and assesses the software architecture fromstatic and dynamicaspects, and evaluates architecture maturity from evolutive perspective. The study includes following content:1) Software Architecture Measurement Model. Architecture Maturity and related concepts as well as their division principles are proposed. Elaborated SAM model approach, including the static architecture maturity model, dynamic architecture maturity model and integrated architecture maturity model.2) The process about the evaluation of evolution. From the first to the last version of the schema version architecture, evolution is not a direct step, but a series of adjacent evolution version superposition. First, the basic data collection of the current version; and then complete the work of the current version of architecture maturity metric, and the last part is the evolution of the assessment which analyze the physical differences and change reasons between adjacent architecture versions;3) The method of evolution assessment which assesses physical difference and causes. Specifically, the analysis of the physical differences between the two neighboring architecture versions analyzes any differences between the two; as for reason analysis which is used to identify the maturity of measurement results change reasons.The method proposed in this thesis can accurately identify all the differences and finds the impact of evolutive operations on the architecture, which provides important information for architects for decision making. Three experiments are designed to verify the validity of the software architecture maturity model and evolution assessment methods which demonstrates the maturity measurement and the evolution assessment methods, and the results verify the effectiveness of this method.