A Stochastic Analysis Of Computing Models In Computer System Reliability Based On Software

Abstract

The backdrop concentrates on the research and creation of models for software dependability, with an emphasis on the utilization of Non-Homogeneous Poisson Process (NHPP) prototypes to forecast that rate of software malfunction. Software reliability analysis and simulation methodologies are getting better thanks to the efforts of researchers and academics in computer science, software engineering, and related fields. The significance of precise modelling for software reliability in terms of boosting software quality, making the most of testing, and boosting software systems' overall reliability. Software reliability can be anticipated and evaluated using a variety of models and methods, such as failure detection, fault repair techniques, and dependable growth assumptions.

Objective

In order to predict and identify software failure rates, the study paper will examine and assess a variety of software reliability models, including stochastic models such as Non-Homogeneous Poisson Process (NHPP) models. The paper's objectives are to examine multiple models in respect to the Theil statistic and performance indicators such as variance, bias, root median square prediction mistake and the mean square mistake. The analysis's findings will show how precise and well-suited the recommended methods are to more normative growth models of software reliability. In order to anticipate software dependability, the study also discusses the construction of multiple change-point extended queueing models that take time delays in fault detection and rectification processes and defect recovery rates into account.

Developed throughout the previous three decades, software reliability growth models (SRGMs) are a valuable tool for estimating and forecasting software dependability. The majority of frequent SRGMs seem to think that errors will be corrected immediately. These expectations might not always be realistic or correct in real life. To debug software, programmers need to be able to replicate mistakes, identify their root causes, make the required adjustments, and then run the programme again. It takes time to complete this process. It's possible that the issue's clearance rate fluctuates as well may vary over time and between sites, according to some research or observations. Consequently, we shall look into the subject of how queueing models for mistake detection and correction define software development processes. We introduce an enhanced model for infinite server queuing that features numerous change-points to accurately estimate and assess software stability. The suggested model may be more accurate than traditional SRGMs in capturing the variance in fault rectification rates and influencing software development behaviour, according to empirical results and real failure statistics.