An Application of Causal Analysis to the Software Modification Process

The development of high quality large-scale software systems within schedule and budget constraints is a formidable software engineering challenge. The modification of these systems to incorporate new and changing capabilities poses an even greater challenge. This modification activity must be performed without adversely affecting the quality of the existing system. Unfortunately, this objective is rarely met. Software modifications often introduce undesirable side-effects, leading to reduced quality. In this paper, the software modification process for a large, evolving real-time system is analysed using causal analysis. Causal analysis is a process for achieving quality improvements via fault prevention. The fault prevention stems from a careful analysis of faults in search of their causes. This paper reports our use of causal analysis on several significant modification activities resulting in about two hundred defects. Recommendations for improved software modification and quality assurance processes based on our findings are also presented.