DEVELOPMENT OF THE PARALLEL ANT COLONY OPTIMIZATION ALGORITHM BASED ON THE TRAVELING SALESMAN PROBLEM

Abstract

The purpose of this study is to develop software to conduct research on the effectiveness of the parallel ant colony optimization algorithm applied to the shortest-path search. The algorithm is based on the behavior principles of an ant colony as well as individual ants, who search the shortest path to the food source. In order to study the effectiveness of the ant colony optimization algorithm one of the most common optimization problems was applied, specifically – a travelling salesman problem. The algorithm shows its efficiency and stability only on the sufficient amount of resources. Its resource usage and therefore the working time grows rapidly together with the growth of the size of the task. In order to solve this problem the parallelism is applied. It is also вused to reduce the premature convergence to the local optimum, as well as to stimulate diversity and search of the alternative solutions of the same problem. In the proposed software each independent ant colony is developed on the separate computer core. Each colony exchanges the best individuals with other colonies in a certain number of generations. The topology of this interaction has the form of «each with each». According to the calculations, we can conclude, that the efficiency of the ant colony optimization algorithm does not change with the increase in the number of kernels (individual populations). However, there is a slight change of the reliability of the particular settings. But at the same time we can observe the substantial increase in operation speed. Thus, the parallelization of the ant colony optimization algorithm can significantly reduce the execution time of a software application without compromising reliability. The ant algorithm is used to solve complex optimization problems. Typical areas of the algorithm application include scheduling, resource and job allocation, transport routing, etc. For this reason, the problem of increasing the efficiency of the algorithm calculations is quite topical.