Performance Prediction on Software Reliability of Embedded Computer Systems

Abstract

The role of an embedded computer system has created a significant revolution during last decades. The study of reliability in software and hardware systems is being developed as an interdisciplinary science. In this thesis, we present reliability as well as availability models and how these models can be helpfulfor developing reliable software systems.Here, we describe the basic and theoretical work in the field of reliability analysisof an embedded computer systemincluding availability, failure frequency, mean time to failure, mean time to repair etc.To achieve high reliability and availability of the embedded computer system, performance modeling is required. In this thesis, authors aim is to develop several models including software reliability growth model. The study of various hardware and software models have been done analytically as well as numerically in different chapters.Moreover, governing equations of an embedded system are constructed using appropriate birth-death rates. To obtain the system state probabilities governing the model of the embedded system, we employ various methods such as Runge-Kutta method, recursive method,matrix method andLaplace transform in the thesis.Furthermore, the analytical results are also compared with adaptive neuro-fuzzy inference system (ANFIS) and genetic algorithm (GA). The steady state probabilities of the model are obtained which are further used to obtain the various performance indices of the software-hardware system. To validate the analytical results, numerical results are performed using MATLAB software. The effect of different parameters on the system performance indices are demonstratedin the form of tables and graphs.The research work of this thesis has been segmented into eight chapters, of which a brief summary has been outlined below. newline In the first chapter, wereview some notable research works done by various prominent researchers related to ourwork. The research objectives, motivation, methodology and various performance measures for embedded system ar