Study On Software Reliability Model Of The Synthetical Use Of Two Kinds Of Time

The development of computer technology, especially network technology, push ITtechnology to a new height, and it has even been called the third industrial revolution.As the development of people，s demand to computer, software products increase as themountain spring spews, with the increase of the application. Software reliability, whichis the important quantifiable standards of software quality, has been being paid moreand more attention to.Firstly, this paper introduces the significance of the software reliability, and thenintroduces the research status of software reliability at home and abroad. Secondly, itexpounds the basic concepts of software reliability, mathematical foundation ofsoftware reliability modeling, process of modeling, metrics standards, and some classicsoftware reliability models. This paper combines the calendar time and test-executiontime to process the statistical data from software testing phase, and get the prediction，which is more in line with the actual situation. Lastly, two-dimensional softwarereliability model is established by the calendar time and test-execution time, which isbased on the Non-Homogeneous Poisson Process.①Calendar time and test-execution time are used together. In the data processingof one-dimensional model, test-execution time is used to calculate the software failureintensity, and calendar time is the foundation of the weight distribution. Cumulativenumber of software failure is the abscissa, and cumulative software failure intensity isthe ordinate. Value coefficient is given, which is in order of time sequence, grows by thespeed of exponent. A straight line could be gained by a virtual "center of gravity" andthe last point, which is from all the data points except the last the one, according to thevalue coefficient. The straight line is used to predict the next data point.②Two-dimensional software reliability model is established by calendar time andtest-execution time. On the basis of traditional one-dimensional software reliabilitymodels, the Two-dimension software reliability model is established from thetest-execution time, which is used to express the testing-effort expenditure,Non-Homogeneous Poisson Process, Marshall-Olkin distribution and the bivaritateWeibull distribution.③Some experiments have been done to compare with others. The prediction datacompares with actual values by some standards. In one-dimensional case, G-O model is used in the comparison, which is good to predict the situation in the future. Intwo-dimension case, it compared with the GO, GL, LT, and WT.