Resumen de Statistical approach to software reliability certification

Isaac Corro Ramos, Alessandro Di Bucchianico, Kees Van Hee

• We present a sequential software release procedure that certi es with some con dence level that the next error is not occurring within a certain time interval. The failure detection process is modeled as a black box software reliability growth model. We focus on the two main classes of models: General Order Statistics and Non-homogeneous Poisson Process.

  Decisions are based on the error-free period since the last error observation. Such periods depend on the test history and by choosing them properly we can certify that: the release time is optimal and the global risk taken in the procedure can be controlled. Our procedure is a special case of sequential testing one-stage look ahead policy. We study our problem from a Bayesian point of view without making any assumption on the initial number of errors in the software.We show how to implement our procedure and study its performance via simulation.