Abstract
In this article, we discuss a continuous time software reliability model under the non homogeneous Poisson process (NHPP) assumption for error occurrence in each fault to suit periodic debugging, in which errors are not corrected at the instants of their detection but at some pre-specified debugging times. We describe maximum likelihood estimation for the model parameters and provide a computational method to estimate those parameters. This in turn helps to estimate the reliability of the software. We also discuss some asymptotic properties of the estimated model parameters, specially the number of errors initially present in the software. Finally, we investigate the finite sample properties of the estimates under a specific family of NHPP models, specially that of the initial number of errors, through simulation.