Abstract
The experiments conducted in previous studies demonstrated the successfulperformance of BSA and its non-sensitivity toward the several types ofoptimisation problems. This success of BSA motivated researchers to work onexpanding it, e.g., developing its improved versions or employing it fordifferent applications and problem domains. However, there is a lack ofliterature review on BSA; therefore, reviewing the aforementioned modificationsand applications systematically will aid further development of the algorithm.This paper provides a systematic review and meta-analysis that emphasise onreviewing the related studies and recent developments on BSA. Hence, theobjectives of this work are two-fold: (i) First, two frameworks for depictingthe main extensions and the uses of BSA are proposed. The first framework is ageneral framework to depict the main extensions of BSA, whereas the second isan operational framework to present the expansion procedures of BSA to guidethe researchers who are working on improving it. (ii) Second, the experimentsconducted in this study fairly compare the analytical performance of BSA withfour other competitive algorithms: differential evolution (DE), particle swarmoptimisation (PSO), artificial bee colony (ABC), and firefly (FF) on 16different hardness scores of the benchmark functions with different initialcontrol parameters such as problem dimensions and search space. Theexperimental results indicate that BSA is statistically superior than theaforementioned algorithms in solving different cohorts of numericaloptimisation problems such as problems with different levels of hardness score,problem dimensions, and search spaces.