A Method for Selecting a Model to Estimate Software Reliability at the Design Phase of Component-Based Real-Time System Development

Abstract

The rapid growth of software-based functionalities has not only increased the complexity of Component-Based Real-Time Systems (CBRTS) but also made it difficult for designers to quantify its reliability. Architecture-Based Software Reliability Models (ABSRMs) are useful for estimating the architectural reliability of component-based systems so that the behavior of its software components can be examined at different development phases. However, due to the availability of a number of models spread over different ABSRM categories, it is difficult to know which one is suitable for a particular system. Traditional model selection methods have two limitations: (i) they can only be applied to identify a model from a specific ABSRM category; and (ii) they omit a well-defined and verified mechanism to identify the ABSRMs which are capable of addressing the reliability requirements at an early development phase. Reliability estimation at the design phase enables developers to analyze the impact of the software component's reliability and its interactions on a CBRTS, study the sensitivity of CBRTS reliability to the reliabilities of its software components and interfaces and guide the process of identifying potential troublesome software components which require more attention during the integration exercises. In this paper, we describe a method for selecting an ABSRM to estimate software reliability at the design phase of a CBRTS development. To demonstrate the usefulness of our proposed method, we apply it to an automotive Automatic Parking System.

Comprehensive evaluation of software system reliability based on component-based generalized G-O models

“…Therefore, a component interface with a simple architecture can be regarded as completely reliable. On the other hand, a component interface with complex architecture can be split off, analyzed, improved, and then assembled with other components ( Garg & Lai, 2018 ).Thus, we can regard the component interfaces as components themselves and use the SRGMs to describe their reliabilities; after integration, developers must test the new system to eliminate interface faults and verify the functions of components and the new system. For large-scale software systems, developers usually adopt integration testing based on an operation profile.…”