This paper proposes a method for assessing software reliability models and its application to the Musa and Littlewood-Verrall models. It is divided into two parts.

In the first part, a taxonomy of the criteria to be considered for assessing a software reliability model is developed and a method for applying this assessment is presented.

In the second part, the result of an assessment of the Musa and Littlewood-Verrall models is presented.

--From the Abstract

Liu and Howell [1] define three broad classes of software reliability models: psychological, spatial, and temporal. The psychological models assume a uniform probability of error per bit of information in the program, the spatial models assume a uniform data density in error space, while the temporal models assume that the errors will be discovered with a given time distribution. The taxonomy presented in this paper is useful only for spatial and temporal models.

The paper goes into extreme and possibly excessive detail in the definition of its taxonomy. There are terms and variables defined in figures and used in the text without further definition. These two combine to make the first part of the paper extremely heavy going.

The second part of the paper is a mechanistic application of the method defined in the first part. The comparison may prove useful to those interested in either model or in the method.

The subject matter is presented much more clearly and at a higher level of abstraction in [2]. Consequently, I can only recommend this paper to those who are either interested in the comparison of the Musa and Littlewood-Verrall models, or those who need a method to compare other software reliability models. Researchers can find better references elsewhere in the literature.