Abstract
Assessment of software reliability has emerged as an area of utmost importance in recent times with the proliferation of software-based systems. Conventional software reliability assessment is done by applying software models that incorporate the unfortunate drawback of requiring previous failure data to formulate the model based on statistical methodologies. But highly dependable software systems used for safety-critical applications, such as computer relays for power system transmission line protection, produce little failure data. This article presents a methodology using statistics of extremes to embark upon software success estimation. The estimate so obtained is an indicator analogous to the conventional reliability index. This estimated reliability is further employed for assessing software's health more effectively by means of well-being analysis. The most significant contribution of this article is to compute the software well-being indices for healthy, marginal, and risky states. A case study for software used in computer relaying of power system transmission line protection validates the efficacy of the proposed methodology, especially for safety-critical applications.