Reliability improvement and assessment of safety critical software

Abstract

In order to allow the introduction of safety-related Digital Instrumentation and Control (DI&C) systems in nuclear power plants, the software used by the systems must be demonstrated to be highly reliable. The most widely used and most powerful method for ensuring high software quality and reliability is testing. An integrated methodology is developed in this thesis for reliability assessment and improvement of safety critical software through testing. The methodology is based upon input domain-based reliability modeling and structural testing method. The purpose of the methodology is twofold: Firstly it can be used to control the testing process. The methodology provides path selection criteria and stopping criteria for the testing process with the aim to achieve maximum reliability improvement using available testing resources. Secondly, it can be used to assess and quantify the reliability of the software after the testing process. The methodology provides a systematic mechanism to quantify the reliability and estimate uncertainty of the software after testing.