Sequence and sequence-less t-way test suite generation strategy based on the elitist flower pollination algorithm
Interaction, or t-way, testing, where t indicates the interaction strength, is an approach to generate test suite for detecting fault due to interaction. In line with the emerging field called Search based Software Engineering, many recently developed t-way strategies have adopted meta-heuristic alg...
| Summary: | Interaction, or t-way, testing, where t indicates the interaction strength, is an approach to generate test suite for detecting fault due to interaction. In line with the emerging field called Search based Software Engineering, many recently developed t-way strategies have adopted meta-heuristic algorithms as the basis of their implementations such as Simulated Annealing, Genetic Algorithm, Ant Colony Optimization Algorithm, Particle Swarm Optimization, Harmony Search and Cuckoo Search, owing their superior performance in term of test size reduction as compared to general computational based strategies, such as General t-way, Test Vector Generator, In Parameter Order General, Jenny, and Automatic Efficient Test Generator. Although useful, all aforementioned t-way strategies have assumed sequence-less interactions amongst input parameters. In the case of reactive systems, such an assumption is invalid as some parameter operations, or events, occur in sequence and hence, creating a possibility of bugs or faults triggered by the order, or sequence, of input parameters. If t-way strategies are to be adopted in such a system, there is also a need to support test data generation based on sequence of interactions. Addressing these aforementioned issues and complementing the existing sequence based strategies such as t-SEQ, Sequence Covering Array Generator and Bee Algorithm, this thesis presents a unified strategy based on the new meta-heuristic algorithm, called the Elitist Flower Pollination Algorithm (eFPA). Unlike existing work, eFPA presents the novel approach of integrating both sequence and sequence-less t-way test suite generation within a single strategy. Concerning the sequence benchmark experiments, eFPA has superior performance when compared with the existing sequence based strategies (with 100% rejection of the null hypothesis). As for sequence-less benchmark experiments, eFPA outperforms most existing strategies (with 92.85% rejection of the null hypothesis). Additionally, we also conclude that eFPA generates better results as compared to the original FPA owing to its enhanced exploration capability through the additional elitism mechanism. In fact, the design of eFPA adds new value into the domain software testing as it is the first t-way strategy that adopts elitism-FPA as its core implementation. |
|---|