A MapReduce based Ant Colony Optimization approach to combinatorial optimization problems

Abstract

Ant Colony Optimization (ACO) is a kind of meta-heuristics algorithm, which simulates the social behavior of ants and could be a good alternative to existing algorithms for NP hard combinatorial optimization problems, like the 0-1 knapsack problem and the Traveling Salesman Problem (TSP). Although ACO can get solutions that are quite near to the optimal solution, it still has its own problems. Premature bogs the system down in a locally optimal solution rather than the global optimal one. To get better solutions, it requires a larger number of ants and iterations which consume more time. Parallelization is an effective way to solve large-scale ant colony optimization algorithms over large dataset. We propose a MapReduce based ACO approach. We show how ACO algorithms can be modeled into the MapReduce framework. We describe the algorithm design and implementation of ACO on Hadoop.