| Summary: | 碩士 === 佛光大學 === 資訊學系 === 97 === With the rapid of the deployment of computer systems, people in the modern society are increasingly dependent on the hardware and software systems. When the demand for computer systems increases, the possibility of crises from computer failure will also increase. Therefore, the reliability of computer systems has become an important concern for our daily life. Numerous Software Reliability Growth Models (SRGMs) for the software failure phenomenon have been developed to measure software reliability, and some of them are based on Nonhomogeneous Poisson Process (NHPP). However, some of the above assumptions of traditional SRGMs are not applicable in an actual software environment. Therefore, we propose a new model to remove the first drawback that the fault-correction process is perfect and can thus establish a corresponding time-dependent delay function to fit the fault-correction processes. The objective here is to remove this assumption in order to make the SRGMs more realistic and accurate. Furthermore, we apply neural network with back-propagation to match the histories of software failure data. To illustrate the proposed approach, we use the interval-domain data set which was the System T1 data of the Rome Air Development Center (RADC) reported by Musa. The system T1, developed by Bell Laboratories, was used for a real-time, command and control system and took 21 weeks to do the software test. After the experiments, we can see that the proposed models fit the data well and the estimated number of initial faults is very close to the real observed number of initial faults.
|
|---|
