Ant Colony Optimization Approach to Communications Networks Design

1
Ant Colony Optimization Approach to
Communications Networks Design

• Tuesday, November 24, 2009
• By
• Bau Yoon Teck ytbau_at_mmu.edu.my
• http//pesona.mmu.edu.my/ytbau/
• MMU FIT COE for CAIIC

2
About ACO in Comm.

• Objective
• Research Info.
• Abstract
• Ant Colony Optimization
• Summary
• Appendix

3
Objective

• To study current Ant Colony Optimization
  algorithms and existing routing protocols. This
  is to identify the various types of problems in
  communications networks design (especially
  routing) that Ant Colony Optimization (ACO) can
  tackle.
• To apply ACO algorithms for the construction of
  degree-constrained communications networks.
• To perform routing in degree-constrained
  communications network using ACO and its hybrid
  versions.
• To compare the performance of conventional as
  well as existing ACO routing techniques against
  that of the proposed algorithms.

4
Research Info.

• Type of Research Applied /Fundamental
• Beneficiaries of Project Researchers and
  companies
• involved in communications networks design
• Product/Services Software Tools
• Researchers Mr. Bau Yoon Teck
• Source of Funding nil
• Status of Project Ongoing

5
Abstract

• Ant Colony Optimization (ACO) is an algorithm for
  discrete optimization
• and multiagent meta-heuristic approach to
  difficult NP-hard combinatorial
• optimization problems and routing in
  communications networks. The
• inspiring source of ACO is the pheromone trail
  laying and following
• behaviour of real ants, which use pheromones as a
  communication
• medium. We apply ACO algorithms to routing
  problems in
• communications networks under static and dynamic
  conditions. This
• study is divided into three parts. The first part
  aims to identify various
• existing routing protocols and compare their
  performance to that of the
• ACO. The second part of this research involves
  formulating and applying
• the ACO algorithms to construct
  degree-constrained communications
• networks. The ACO routing will then be applied on
  the constructed
• networks, taking into consideration different
  traffic conditions. The final
• part of the study will focus on designing hybrid
  ACO routing protocols
• that incorporates genetic algorithms (GAs) and
  reinforcement learning.

6
Ant Colony Optimization
Caption Ant finding the shortest path in the
graph from their nest (colony) to feeding sources
and back
7
Summary

• In real-life communications networks design and
  optimization situation, the problem often
  requires satisfying additional constraints.

8
Thank You
9
Appendix
Simple ACO algorithm for TSP (1) a set of m
artificial ants are initially located at randomly
selected cities (2) each ant, denoted by k,
constructs a complete tour, visiting each city
exactly once, always maintaining a list Jk of
cities that remain to be visited (3) an ant
located at city i hops to a city j, selected
among the cities that have not yet been visited,
according to probability where a and ß are
two positive parameters which govern the
respective influences of pheromone and distance
(4) when every ant has completed a tour,
pheromone trails are updated Gij ? (1- ?) Gij ?
Gij (5) Steps (1) to (4) are repeated either a
predefined number of times or until a
satisfactory solution has been found.