A Human Intelligence Inspired Meta Heuristic Optimization Algorithm for TSPs

Abstract

This paper presents a novel heuristic optimization algorithm for discrete optimization problems, the Bandwidth Restricted Transmission-Simulated Optimization Algorithm (BRT-S). This algorithm imitates the intellective behaviors of human in managing network transmission. BRT-S is a constructive heuristic algorithm whose optimization procedures simulate processes of data transferring and management operations over the network. A population of solution agents mimicking message transmitters on networks is deployed to quest for optimal solutions. The algorithm however restricts the resource utilized in solution search mimicking the bandwidth resource is limited in network transmission. As a result, agents must compete with others to obtain solution construction resources. Due to the mimicked bandwidth restriction, not every agent is able to complete a solution construction. Only constructed solutions are subject to objective value evaluations. In each evolution iteration, bandwidth resources are separately modulated by conducting bandwidth deterioration, enhancement, or depletion, on the basis of the solution qualities obtained. To illustrate the operation procedures of the algorithm, a BRT-S computation model for solving the Traveling Salesman Problem is presented and the solving system is implemented for benchmark testing. Numerical results of the tests indicate that given similar computation resources, the algorithm generates better solutions than other meta heuristic algorithms, such as ACO and GA.