The Role of Random Walk-Based Techniques in Enhancing Metaheuristic Optimization Algorithms—A Systematic and Comprehensive Review

Abstract

Metaheuristic algorithms (MHAs) occupy considerable attention among researchers because of their high performance and robustness in optimizing several engineering problems. Random walk (RW) techniques showed a significant role in improving the performance of these algorithms. Therefore, this paper aims to provide a systematic and comprehensive review of the role of three substantial random-walk (RW) strategies in enhancing the performance of MHAs. These strategies are the Gaussian, Levy Flight and Quantum random walks. The PRISMA methodology is applied through the articles obtained from four famous scientific databases. The study provides the integration mechanisms as well as the best controlling parameters’ values while integrating these RW strategies into Particle Swarm Optimization (PSO) to produce the Gaussian PSO (GPSO), Levy Flight PSO (LFPSO) and Quantum PSO (QPSO). An experimental study has been conducted to assess the performances of these algorithms in addition to the standard PSO on 23 unimodal, multimodal and fixed-dimension multimodal benchmark functions. Statistical measures have been calculated based on 30-run optimization processes. The comparisons showed that the QPSO, LFPSO, GPSO and PSO have successfully reached the optimal values of 23 standard benchmark functions with average percentages of 65%, 31%, 13% and 11%, respectively. Accordingly, the QPSO has gained the outstanding rank, especially for unimodal and multimodal functions followed by the LFPSO while the standard PSO comes in the last position preceded by the GPSO. From the results, it can be concluded that integrating random walk strategies into existing or new metaheuristic algorithms is capable of enhancing the optimization process and hence provides reliable results when applied to engineering applications.