Software evaluation for developing software reliability engineering and metrics models

摘要

Today's software is extremely complex, often constituting millions of lines of instructions. Programs are expected to operate smoothly on a wide variety of platforms. There are continuous attempts to try to assess what the reliability of a software package is and to predict what the reliability of software under development will be. The quantitative aspects of these assessments deal with evaluating, characterizing and predicting how well software will operate. Experience has shown that it is extremely difficult to make something as large and complex as modern software and predict with any accuracy how it is going to behave in the field. This thesis proposes to create an integrated system to predict software reliability for mission critical systems. This will be accomplished by developing a flexible DBMS to track failures and to integrate the DBMS with statistical analysis programs and software reliability prediction tools that are used to make calculations and display trend analysis. It further proposes a software metrics model for fault prediction by determining and manipulating metrics extracted from the code.