A review on ripple-spreading genetic algorithms for combinatorial optimization problems

In various implementations of genetic algorithms (GAs) to combinatorial optimization problems, permutation representations are often adopted. However, these permutation-representation-based implementations are often confronted with one or more of the following problems: (i) Evolutionary operations may generate infeasible solutions; (ii) The representations are not memory-efficient, and may hamper the scalability of algorithms; (iii) Many classic binary evolutionary operators can hardly apply without significant modifications. To address these issues, a novel scheme for applying binary-representation-based GAs to combinatorial problems has recently been proposed based on a ripple-spreading model. Since this model is the centerpiece of the new scheme, we call it the ripple-spreading genetic algorithm (RSGA). In previous studies, several bespoke RSGAs have been developed to tackle a range of different combinatorial optimization problems. Although these RSGAs differ in design details, they share the same motivation, follow the same methodology and illustrate the same advantages vis-a-vis other methods. Based on the previous studies, this paper aims to present a comprehensive review of the methodology of RSGA. Particularly, by analyzing those existing implementations of RSGA, this paper will discuss some generalized important technologies for designing RSGA.