Research On Adaptable Software Architecture Evaluation Method

Since the last decade of 20th century, one of themostly problems influencing the large software system is quality because of the incessant increasing size, complexity and demand for quality. The researchers and practitioners have recognized that quality attributes of large software systems are largely constrained by the architecture of system. Therefore, choosing the appropriate software architcture has determinant influence on the key quality attributes of system. It is very important to evaluate a system's architecture according to quality requirments as early as possible. The purpose of the evaluation is to analyze the software architecture to identify potential risks and verify that the quality requirements have been addressed in the design.As an aspect of quality features of architectur, adaptability has definite particularity. The research on adaptable software architecture evaluation method is quite faulty, and there is a lack of qualitative and quantitative metric indexes and systemic evaluation methods. Therefore, investigating adaptable software architecture evaluation method is of important significance and wide development space.This thesis researches and discusses the adaptable software architecture evaluation method of large software system. The main contents of these studies are as following:(1) It summarizes, analyses and concludes the theory foundation, summarizes the development of the representative software architecture evaluation methods, and discovers the important significance to investigate adaptable software architecture evaluation method.(2) Based on goal-scenario coupling, this thesis presents a structured method of goal description and identification. And according to the goal modeling gained from the above method, this thesis gives a new goal-driven software architecture modeling method based on the theory and principle of graph grammar presented by Le Metayer.(3) Based on GQM approach, integrating the research status, this thesis presents a process-oriented metrics for software architecture adaptability. This method extends and improves the GQM method. It develops process-oriented processes for metrics modeling, introduces data and validation levels, increases structured description of metrics, and defines new indexes of metrics.(4) This thesis gives a new reasonable decision making method based on IAt-IP and DEA. An unascertained measure model for evaluating the adaptation of software architecture is established on the basis of the unascertained measure theory. An example of practical application is given to show the effectiveness of these methods.(5) Based on the basic principle and concrete content of the adaptable software architecture evaluation method, the TQM system is taken as an example application to demonstrate the method.