Software reliability growth models of non-homogeneous Poisson process

Many software reliability growth models (SRGM) based on fault coverage only consider accumulative fault coverage ignoring that gained by every test case. In order to describe the software testing process, two non-homogeneous Poisson process (NHPP) SRGM incorporating fault coverage are proposed. The first model supposes every test case has equal fault detective ability and can get the same fault coverage. The second model is proposed considering that faults remaining in software have different detected rate and that faults with highest rates are removed first. Fault detectability of a test case is related to the order that it is executed. The goodness-of-fit of the proposed model is examined using a software failure data set. Comparing with some other existing NHPP SRGM, the new models can provide better goodness-of-fits.