Evolutionary Swarm based algorithms to minimise the link cost in Communication Networks

Juan Antonio Gómez-Pulido,Anabela Moreira Bernardino,Eugénia Moreira Bernardino,Juan Manuel Sánchez-Pérez,Miguel Ángel Vega-Rodríguez

doi:10.1080/18756891.2012.718157

Juan Antonio Gómez-Pulido, Anabela Moreira Bernardino + Show 3 more

Open Access

https://doi.org/10.1080/18756891.2012.718157

Copy DOI

Abstract

Abstract In the last decades, nature-inspired algorithms have been widely used to solve complex combinatorial optimisation problems. Among them, Evolutionary Algorithms (EAs) and Swarm Intelligence (SI) algorithms have been extensively employed as search and optimisation tools in various problem domains. Evolutionary and Swarm Intelligent algorithms are Artificial Intelligence (AI) techniques, inspired by natural evolution and adaptation. This paper presents two new nature-inspired algorithms, which use concepts of EAs and SI. The combination of EAs and SI algorithms can unify the fast speed of EAs to find global solutions and the good precision of SI algorithms to find good solutions using the feedback information. The proposed algorithms are applied to a complex NP-hard optimisation problem - the Terminal Assignment Problem (TAP). The objective is to minimise the link cost to form a network. The proposed algorithms are compared with several EAs and SI algorithms from literature. We show that the propose...

Highlights

A great number of engineering models and algorithms have been used to solve complex optimisation problems
In order to test the performance of our approaches, we use a collection of Terminal Assignment Problem (TAP) instances of different sizes
The suggestions from literature helped us to guide our choice of parameter values for Tabu Search (TS) [9], Hybrid GA (HGA) [9], GA with Multiple Operators (GAMO) [12], Hybrid Differential Evolution (HDE) [13], Local Search GA (LSGA) [11], Hybrid ACO (HACO) [19], Bees Algorithm (BA) [20], Hybrid Scatter Search (HSS) [16], Discrete Differential Evolution (DDE) [17] and Hybrid Population Based Incremental Learning (HPBIL) [18]

Summary

Introduction

A great number of engineering models and algorithms have been used to solve complex optimisation problems. EAs and SI algorithms have been extensively applied to solve complex optimisation problems. EAs are a subset of evolutionary computation They are bio-inspired population-based meta-heuristic optimisation algorithms [1]. Ant Colony Optimisation (ACO), Particle Swarm Optimisation (PSO), Bees Algorithm (BA) and Artificial Bee Colony (ABC) algorithm are some of the most known SI approaches. We propose a Genetic Algorithm with a new “Swarm” mutation operator (GAS) and a QueenBee Evolutionary Algorithm (QBEA) to optimise a communication network problem - the Terminal Assignment Problem (TAP). That this paper presents the first attempt (to the authors’ knowledge) to use evolutionary swarm based algorithms to optimise TAP. In TAP the number of concentrators and their capacities and locations are known. In TAP, a communication network will connect N terminals, each with Li demand (weight) to M concentrators, each of Cj capacity. We observe an increasing size and an increasing complexity of communication networks, and for that reason finding an optimal solution for TAP continues to be a hard task

Previous Work

Proposed Algorithms

Evaluation of Solutions

Results

Conclusions

39. Differential