An improved path-based reliability prediction model for component-based embedded software

With the increasing use of component-based software development in embedded systems, reliability prediction for such software has become a challenge. This paper consequently contributes to this problem through an improved path-based reliability prediction model. We use Hoare's Communicating Sequential Processes (CSP) to specify the software architecture. So the execution path could be derived from CSP trace model accurately and completely. A probabilistic model named Component Transition Graph (CTG) is constructed. Based on CTG, the execution frequency of path is calculated. By analyzing the structure of loop in the architecture, we propose a solution for reliability estimation containing loops in path which is the most important contribution of our work. The reliability prediction and sensitivity analysis techniques and how the model could be used to identify the critical component interfaces and transition are illustrated with examples.