Abstract
In recent years, bio-inspired optimization algorithms have demonstrated the ability to produce optimal solutions to numerous complex computational problems in science and engineering. In this work, a comparative analysis of bio-inspired algorithms is presented to understand and quantify the performance of algorithms in guiding the search process toward better solutions over all feasible solutions. Three evolutionary algorithms, namely, the Genetic Algorithm (GA), Particle Swarm Optimization (PSO) and Differential Evolution (DE), are implemented to generate an optimized test sequence set for digital sequential circuits to investigate how they explore and the exploitation realized in their search spaces. The merit of using a comparative method to analyze and optimize search spaces is to reach a reliable and quantifiable conclusion about the relative performance of each algorithm.
An improvement in the quality of solutions was achieved, particularly in terms of testing time, number of test vectors, and fault coverage of tested sequential circuits in comparison with other algorithmic test generators that have been presented in the literature. The study shows how to effectively reduce the search space without negatively affecting the results and guides the search process over a large space. In addition, the experiments highlight the limitations of each optimization algorithm and offer some constructive methods of improvement. Moreover, several recommendations and guidelines regarding the use of optimization algorithms as test pattern generators to improve their performance and increase their efficiency are presented. Finally, we emphasize the relevance of bioinspired algorithms in solving complex computational problems, data manipulation, and optimization objectives. (C) 2020 Elsevier B.V. All rights reserved.