Experience report: Error distribution in safety-critical software and software risk analysis based on unit tests

Abstract

Systematic verification and validation measures are of essential importance in particular for safety-critical software. After a short introduction into the test environment and error categories, the paper presents the results of a unit test performed on a C++ software package for the European Space Agency in the ARC Seibersdorf research test lab. The authors analyse the error distribution and relations between software metrics and software faults and recommend guidelines for a less error-prone design of safety-critical software.