A Methodology for Model-Based Reliability Estimation

Introduction

System software has become vastly more complicated in recent years, and ranges from e-commerce to industrial embedded systems, which support both functional and non-functional requirements of the system. In fact, the satisfaction of non-functional requirements or software quality, such as performance, reliability, security and adaptability, towards functional requirements are often treated as critical. Performance-sensitive systems are widely used to support various application domains, such as telecommunications, embedded systems, and industrial control systems. The unreliability of such systems is often related to underperformance of systems and could affect the whole system. Therefore, it is largely necessary to ensure promising software quality from as early as possible in the design phase to satisfy today’s demanding standards.

Software qualities generally can be defined as a combination of quality attributes of a system in a certain accepted degree (IEEE 1061, 1992). According to ISO 25000 (ISO/IEC, 2005), software quality is defined as a set of qualities of a system that satisfies the given system’s requirements’. In recent years, several researchers and standard organizations have proposed the software quality taxonomy in order to define exactly which software quality is justified as being important for the software systems (McCall et al., 1977; Boehm, 1978; Dromey, 1995; ISO 25000, 2005).

Among software quality attributes, considerable significance has been given to reliability attributes (Kramer and Volker, 1997). Software reliability is described as the capability of the system/component/services to perform a given task under a specific condition and time frame (IEC 9126-1, 2001). The system operation is said to be reliable when continuously operating with minimal failure. Software reliability generally acknowledges the probability that a system may fail within a certain period. The failure elements can be traced from the system’s error and fault information (Eusgeld et al., 2008). Faults in the system will impose a certain error that will lead to a system failure if no specific approach is applied to handle this matter. A system failure might be due to faults in implementation, unreliable system resources or misplaced system tasks (Salfner and Lenk, 2010; Valis and Bartlett, 2010). Therefore, wherever the operation of the system is disturbed, the consequences may result in an enormous risk to either people or the environment. On the other hand, the risk can be greatly reduced if an adequate reliability analysis is carried out before deploying the system (Krishna and Mall, 2010).