Optimal Wireless Sensor Network Layout with Metaheuristics: Solving a Large Scale Instance

  • Authors:
  • Enrique Alba;Guillermo Molina

  • Affiliations:
  • Departamento de Lenguajes y Ciencias de la Computación, University of Málaga, Málaga, Spain 29071;Departamento de Lenguajes y Ciencias de la Computación, University of Málaga, Málaga, Spain 29071

  • Venue:
  • Large-Scale Scientific Computing
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

When a WSN is deployed in a terrain (known as the sensor field), the sensors form a wireless ad-hoc network to send their sensing results to a special station called the High Energy Communication Node(HECN). The WSN is formed by establishing all possible links between any two nodes separated by at most RCOMM, then keeping only those nodes for which a path to the HECN exists. The sensing area of the WSN is the union of the individual sensing areas (circles of radius RSENS) of these kept nodes.The objective of this problem is to maximize the sensing area of the network while minimizing the number of sensors deployed. The solutions are evaluated using a geometric fitness function. In this article we will solve a very large instance with 1000 preselected available locations for placing sensors (ALS). The terrain is modelled with a 287×287 point grid and both RSENSand RCOMMare set to 22 points. The problem is solved using simulated annealing (SA) and CHC. Every experiment is performed 30 times independently and the results are averaged to assure statistical confidence. The influence of the allowed number of evaluations will be studied. In our experiments, CHC has outperformed SA for any number of evaluations. CHC with 100000 and 200000 evaluations outperforms SA with 500000 and 1,000,000 evaluations respectively. The average fitness obtained by the two algorithms grows following a logarithmic law on the number of evaluations.