Swap-Based Discrete Firefly Algorithm for Traveling Salesman Problem

Abstract

Firefly algorithm (FA) is an emerging nature-inspired algorithm which has been used to solve discrete optimization problems such as traveling salesman problem (TSP). However, during the discretization of firefly algorithm, one of the FA’s characteristics, i.e. the movement of a dimmer firefly towards a brighter firefly is unapparent as the movement are random. Thus, in this paper, the usage of swap operation as the movement strategy is proposed. The proposed algorithm, Swap-based Discrete Firefly Algorithm (SDFA), is then integrated with Nearest-Neighborhood initialization, reset strategy and Fixed Radius Near Neighbor 2-opt operator (FRNN 2-opt). The proposed algorithm is tested on 45 TSP instances and is compared with several states-of-the-art algorithm. The findings of this research show that the proposed algorithm performs competitively compared to the Discrete Firefly Algorithm, the Discrete Cuckoo Search, the Discrete Bat Algorithm, the Hybrid Genetic Algorithm and the Discrete Bacterial Memetic Evolutionary Algorithm. On average, SDFA reports a percentage deviation of 0.02% from known optimum for TSP instances with dimension range from 14 to 318 cities.