Statistical Software Reliability Modeling for Real-Time PLC- Based Systems

This paper discusses a new way of integrating a suite of commercially available, off-the-shelf software tools and hardware to develop a scalable testing platform called Automated Real-Time Test System (ARTS). The main function of ARTS is to provide automated functional testing, in order to increase the test coverage of a Personnel Safety System (PSS). Data that resulted from five years of manual testing of the PSS was used as input to several software reliability analysis tools. The results of this analysis allowed us to estimate the reliability of the PSS system's software as a function of test coverage. Two statistical modeling tools selected for the analysis were: 1) SMERFS-statistical modeling and estimation of reliability functions for software, and 2) CASRE-computer aided software reliability estimation. A data table of the Brooks-Motley and S-Shaped Reliability growth models are provided for comparison purposes and were use to help in selecting the best model that fit our data.