A Comparison of Error Categorization Schemes for Use in Software System Safety Programs.

reportActive / Technical Report | Accession Number: ADA288741 | Open PDF

Abstract:

Software safety is becoming increasingly important in the development of DoD advanced weapon systems. To make software safer, hazard conditions must be avoided along with the errors that accompany them. The first step in identifying errors is classifying error data. The area of software error classification is not as advanced as other software development areas. The technical literature lacks examples of comprehensive taxonomies that can be applied to various computer software domains and applications. The predominant approach is to organize errors into categories particular to the program currently in work. The typical error scheme is made of narrow categories that are not interrelated. Errors have been classified by symptom, by cause, by life cycle phase, by severity, and by software control. The focus of this research was to determine the best way to classify errors in order to aid system safety in software development. The research identified common areas used in industry that aid in error classification. A telephone survey of experts in safety and software was used to obtain input on the most effective classification schemes. The research also proposed a taxonomy that will be ideal for DOD software development. Since software is becoming a larger part of advanced weapon systems, development of error-free and safe software to operate and support these weapon systems is increasingly important

Security Markings

DOCUMENT & CONTEXTUAL SUMMARY

Distribution:
Approved For Public Release

RECORD

Collection: TR
Identifying Numbers
Subject Terms