Finding lower bounds of localization with noisy measurements using genetic algorithms

  • Authors:
  • Farhan Ahammed;Javid Taheri;Albert Zomaya

  • Affiliations:
  • The University of Sydney, Sydney, & NICTA Australia, Eveleigh, Australia;The University of Sydney, Sydney, Australia;The University of Sydney, Sydney, Australia

  • Venue:
  • Proceedings of the first ACM international symposium on Design and analysis of intelligent vehicular networks and applications
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

Vehicular Ad-Hoc Networks (VANETs) are wireless networks with mobile nodes (vehicles) which connect in an ad-hoc manner. Many vehicles use the Global Positioning System (GPS) to provide their locations. However the inaccuracy of GPS devices leads to some vehicles incorrectly assuming they are located at different positions and sometimes on different roads. VANETs can be used to increase the accuracy of each vehicle's computed location by allowing vehicles to share information regarding the measured distances to neighbouring vehicles. This paper looks at finding how much improvement can be made given the erroneous measurements present in the system. An evolutionary algorithm is used to evolve instances of parameters used by the VLOCI2 algorithm, also presented in this paper, to find instances which minimises the inaccuracy in computed locations. Simulation results show a definite improvement in location accuracy and lower bounds on how much improvement is possible is inferred.