A Large Scale Validation of a Methodology for Assessing Software Reliability (NUREG/CR-7042)
On this page:
Download complete document
Manuscript Completed: November 2010
Date Published: July 2011
Y. Shi, M. Li, W. Kong, J. Dai
Reliability and Risk Laboratory
Nuclear Engineering Program
The Ohio State University
NRC Project Managers:
S. Arndt, N. Carte, R. Shaffer, and M. Waterman
NRC Job Codes Y6591, N6878
Office of Nuclear Regulatory Research
U.S. Nuclear Regulatory Commission
Washington DC 20555-0001
This report summarizes the results of a research program initiated by the U.S. Nuclear Regulatory Commission at the University of Maryland1 to validate a method for predicting software reliability. The method is termed the Reliability Prediction System (RePS). The RePS methodology was initially presented in NUREG/GR-0019, “Software Engineering Measures for Predicting Software Reliability in Safety Critical Digital Systems” and validated on a small control system application with a set of five RePSs in NUREG/CR-6848, “Validation of a Methodology for Assessing Software Quality.” The current effort is a validation of the RePS methodology with respect to its ability to predict software quality (measured in this report and in NUREG/GR-0019 in terms of software reliability) and, to a lesser extent, its usability when applied to safety-critical applications.
The application under validation, herein defined as APP, is based on a safety-related digital module typical of what might be used in a nuclear power plant. The APP module contains both discrete and high-level analog input and output circuits. These circuits read input signals from a plant and send outputs that can be used to provide trips or actuations of system equipment, control a process, or provide alarms and indications. The transfer functions performed between the inputs and outputs are dependent on the software that is installed in the module.
The research described in this report provides evidence that twelve selected software engineering measures in the form of RePSs can be used (with different degrees of accuracy) to predict the reliability of software in safety-critical applications. These twelve measures are ranked based on their prediction ability. The rankings are then compared with those obtained through an expert opinion elicitation effort, as described in NUREG/GR-0019, and with those obtained through a small-scale validation, as described in NUREG/CR-6848.