Software Reliability Modeling using Fault Tree Analysis and Stochastic Petri Nets

Abstract

This thesis estimates the performance of software system using Stochastic Petri Nets (SPN) and Fault Tree Analysis (FTA).This paper explains how a working Petri net model and fault tree model is developed for a soft real time system. Evaluating reliability of software at the early stage of the development is become key concern for most of the organization. Assessing the reliability at early stage helps in removing many bugs and hence improving the efficiency of the architecture. Reliability is computed based on failure rates of systems by using a modified architecture based approach for reliability modelling.Fault Tree Analysis is a method for identifying and documenting the combinations of lower-level software events that allow a top-level event (or root node) to occur. But fault trees are event oriented. The dependency between the components can not be easily incorporated in the model. Similarly redundancies,time delay conditions and other dynamic behaviour can not be easily modelled using fault trees, since they are static in nature. SPN model is used to counter the problems of Fault Trees. A ATM system is taken as an example to model the system in both fault tree and stochastic petri net.