In Sang Chung

KIPS Transactions on Software and Data Engineering, Vol. 2, No. 7, pp. 451-460, Jul. 2013
10.3745/KTSDE.2013.2.7.451,   PDF Download:

Abstract

In order to achieve high test coverage, it is usual to generate test data using various techniques including symbolic execution, data flow analysis or constraints solving. Recently, a technique for automated test data generation that fulfills high coverage effectively without those sophisticated means has been proposed. However, the technique shows its weakness in the generation of test data that leads to high coverage for programs having branch conditions where different memory locations are binded during execution. For certain programs with flag conditions, in particular, high coverage can not be achieved because specific branches are not executed. To address the problem, this paper presents dynamic branch coverage criteria and a test data generation technique based on the notion of dynamic branch. It is shown that proposed technique compared to the previous approach is more effective by conducting experiments involving programs with flag conditions.

Statistics
Cite this article