Recently, a class of experimental designs has been devised that guarantee input domain coverage up to all combinations of k test factors taken t at a time.With such designs, all pairwise combinations (or triplets or quadruplets, etc.) are selected at least once.To evaluate their applicability to software testing, we analyzed the extent to which software coverage (i.e., code execution) achieved by these designs for t=1, ... ,k is representative of that achieved by exhaustively testing all factor combinations.The block coverage obtained for t;5;2 was comparable with that achieved by exhaustively testing all factor combinations but higher-order values of t were required for path coverage.Implications of these results for software testing are discussed.

Software systems typically contain design and code defects that manifest themselves as software failures at various points during program execution. These software faults can be viewed as causing failures to occur according to some chance mechanism that is often taken to be a Poisson process. Individual per-fault failure sequences (resulting from delayed fault detection and software correction) during test execution were analyzed to determine whether they satisfied Poisson process assumptions. The system analyzed, a large, complex, industrial software system, was executed under a defined operational profile corresponding to a controlled command and autonomous event mix characteristic of the system's actual usage. The failure events, defined as severe, system-wide affecting events attributed to software faults, typically occurred following many cycles of interacting system features in the expected user mode. Execution times between failure events were calculated and then statistically analyzed for conformance to Poisson process assumptions. The results indicated that first-time failure events conformed to the assumptions of a Poisson process and were consistent with a reliability growth Poisson model but that per-fault failure sequences attributed to specific faults tended to violate Poisson assumptions due to failure clustering.>

A set of criteria is proposed for the comparison of software reliability models. The intention is to provide a logically organized basis for determining the superior models and for the presentation of model characteristics. It is hoped that in the future, a software manager will be able to more easily select the model most suitable for his/her requirements from among the preferred ones.

Graphical techniques are presented for estimating Hough transform performance for detecting straight lines in noisy digital pictures. The methods show explicitly the influence of noise due, for example, to quantizing errors on the accuracy of estimating the underlying sets of collinear points. Exact bounds are obtained for sets of collinear points showing that under quite general conditions only the endpoints are of significance. Theorems governing grid size for the accumulator method are given.