Optimus: mecanismo de otimização de sequência de casos de testes em sistemas autoadaptativos

Abstract

Self-adaptive systems (SAS) can modify themselves automatically according to their environment. These dynamic adaptations give the system more flexibility, but it can also result in failures during execution, performance problems, and unwanted operations. For SAS, traditional testing approaches are ineffective due to the dynamic aspects of these systems, making fault detection a complex task. Then, various testing approaches for these systems have been proposed in the literature to solve the main challenges, one of which is runtime testing. However, there is still lack of information regarding the coverage and cost of running tests at runtime. Thus, this research proposes a mechanism to reduce the cost of execution and help cover runtime tests to contribute to the identification of faults in SAS. To evaluate this, a proof of concept was developed along with simulations of artificially generated SAS systems covering different complexities and levels of variability (low, medium and high). The mechanism performed efficiently in terms of execution time and was able to select effective test cases in relation to the objectives in different scenarios. The main contributions of this work are a mechanism for generating sequences of test cases, which aims to minimize the cost of execution and increase test coverage using the context variability metric, and a tool for generating test cases in binary and calculating their costs.