Test Generation Techniques for Combinatorial Geospatial Computations

Abstract

Software in the industry is usually tested using old proven techniques. State-of-theart software testing technologies are often reserved for research. This study aims to investigate the advantages and drawbacks of three test generation techniques in an applied domain, where an entire tool may be ruled out due to incompatibilities with the software and its stack. They are tested on a software development kit for real-time, geospatial applications that is written in C++. The study performs a quantitative comparison using mutation testing-inspired fault injection, and a qualitative comparison focused on the applicability, ease of use and usefulness during its entire development and integration process. Property-based testing, model-based testing and combinatorial interaction testing are implemented in the software, and the learnings are presented in this study.