Optimal Test Cases Generation Using Evolutionary Soft Computing Techniques

Software testing is essential in the development of software applications, aiming to uncover errors by validating and verifying a program's functionality. In the contemporary information technology era, software has become fundamental to human activity, highlighting the necessity for high-quality and reliable software. Minor bugs in software can lead to significant disasters, as illustrated by the failures of ARIANE-5 and the Mars Climate Orbiter, which were attributed to small errors with massive impacts. Traditional testing methods like Unit and Integration Testing often fall short in detecting low-frequency bugs, which can cause system crashes or other non-crashing errors. To address this, Random Testing has been introduced as an effective tool for identifying these elusive bugs. The paper proposes an enhancement to random testing through the generation of optimal test cases using the Object Behavior Dependence Model, aimed at reducing faults more efficiently. These test cases are processed through algorithms like Genetic Algorithm, Adaptive Genetic Algorithms, and Genetic Algorithms, with a focus on the Genetic Algorithm for minimizing illegal inputs and redundant test cases. This algorithm employs coverage metrics to evaluate test cases, refining the input space by integrating previous and current inputs. The utilization of optimal test cases generated through this method is expected to significantly enhance the performance and reliability of software applications.