AADL-Based Stochastic Error Propagation Analysis for Reliable System Design of a Medical Patient Table

Abstract

This paper introduces a new method for stochastic error propagation analysis of mechatronic systems designed with Architecture Analysis & Design Language (AADL). The analysis is based on a formal Dual-graph Error Propagation Model (DEPM). This model captures control and data flow aspects and reliability properties of system components and allows the quantitative system reliability evaluation using underlying Markov chain models. This paper describes an automatic transformation algorithm from AADL to DEPM that identifies data and control flow transitions between devices, processes and threads of an AADL model and generates a DEPM for further error propagation analysis. An integrated third-party scheduling tool Cheddar helps to generate control flow sequences that are transformed into a stochastic control flow graph of the DEPM. The generated DEPM allows a user to specify fault activation probabilities for particular system components and numerically analyze error propagation to critical outputs. The introduced method is illustrated with a reliability analysis of a mobile medical patient table.