A Study On Chaotic-Fractal Description Of Software Defects And Quality Evolution Metrics

With the development of complexity research and the increase of software complexity, it will become an inexorable trend how to use complexity theory to study the complexity of software quality, which is one of the difficult problems in the field of software engineering. After the summaries of the related researches in the field, this dissertation gives the fractal growing description of software defects by the chaotic-fractal theory, and uses the modern systems theory to analyze the internal causes of software defect emergence at the level of software system, and designs the process method of evolutional metrics of software quality by the grey theory and fuzzy theory.This dissertation achieves the research results in theory as follows.It abstracts the mathematical description through a fractal growing experiment of software defects, and gives the formula of software defect number of the n-th time iteration, and draws the Darwin Tree of software defect growth during an entire software process, and describes the chaotic process of iteratively stretched and folded fork and sums up the chaotic characteristics of to-initial-value-sensibility, density and self-similar-dimension of software defects.At software system level, the research results of this dissertation show that environmental changes make some software defects appear, which are dynamic, changeable and relative, and the internal causes of software defect emergence is that negative entropy of software is inadequate and software is imperfect, and software defects are less if software self-organization and orderliness are more perfect, and there is implicit influence from software defects with the stabilization of software development during its process while the order parameters during software process play a key role.This dissertation initiates a new software engineering method "Emulative software engineering", which is a process framework of software quality evolutional metrics based on the software quality PDCA ring. In order to measure the evolutional process, the judgment formula of convex-concave and convergence power of evolutional grey sequence are deduced, and the computation and graph methods ofevolutional grey iteration are designed. This grey method is an analysis one for small sample finite sequence, which can be used for judging the convergence rate of a convex-concave sequence, or an irregular curve sequence, or several irregular curve sequences, and for comparing one with another, and can be also employed in other field. In addition, the fuzzy evaluative formula for software process supervision and software process improvement are also deduced. The contrast experiments show that the above evolutional method is obviously helpful to software process improvement.