Naučno-tehničeskij Vestnik Informacionnyh Tehnologij, Mehaniki i Optiki (Jul 2016)

DYNAMIC SOFTWARE TESTING MODELS WITH PROBABILISTIC PARAMETERS FOR FAULT DETECTION AND ERLANG DISTRIBUTION FOR FAULT RESOLUTION DURATION

  • A. D. Khomonenko,
  • A. I. Danilov,
  • A. A. Danilo

DOI
https://doi.org/10.17586/2226-1494-2016-16-4-655-662
Journal volume & issue
Vol. 16, no. 4
pp. 655 – 662

Abstract

Read online

Subject of Research.Software reliability and test planning models are studied taking into account the probabilistic nature of error detection and discovering. Modeling of software testing enables to plan the resources and final quality at early stages of project execution. Methods. Two dynamic models of processes (strategies) are suggested for software testing, using error detection probability for each software module. The Erlang distribution is used for arbitrary distribution approximation of fault resolution duration. The exponential distribution is used for approximation of fault resolution discovering. For each strategy, modified labeled graphs are built, along with differential equation systems and their numerical solutions. The latter makes it possible to compute probabilistic characteristics of the test processes and states: probability states, distribution functions for fault detection and elimination, mathematical expectations of random variables, amount of detected or fixed errors. Evaluation of Results. Probabilistic characteristics for software development projects were calculated using suggested models. The strategies have been compared by their quality indexes. Required debugging time to achieve the specified quality goals was calculated. The calculation results are used for time and resources planning for new projects. Practical Relevance. The proposed models give the possibility to use the reliability estimates for each individual module. The Erlang approximation removes restrictions on the use of arbitrary time distribution for fault resolution duration. It improves the accuracy of software test process modeling and helps to take into account the viability (power) of the tests. With the use of these models we can search for ways to improve software reliability by generating tests which detect errors with the highest probability.

Keywords