SICA: a Software Complexity Analysis Method for the Failure Probability Estimation

This paper presents a method, called SICA (SImple Complexity Analysis), for the complexity analysis of application software in computer based reactor protection systems of nuclear power plants. The complexity measures are utilised in the estimation of software failure probabilities. Complexity of software can be defined in several ways. The challenge is to find a practical and justifiable metric, which can be assumed to correlate with the reliability. The goal has been to develop a simple complexity analysis method, because reactor protection systems contain typically very many software modules, and their analysis can be time-consuming. The complexity analysis is performed based on functional diagrams used for requirements specification. Software modules are divided into three complexity categories: low, medium and high. In SICA, categorisation of modules is performed based on the number of feedback loops, the number of connected complex function blocks, the number of connected function blocks, and the number of inputs and outputs. The complexity analysis is demonstrated with application software module examples. The decision rules of the SICA method are simple to apply and the complexity category of a software module can be determined by a visual assessment.