Maximum Likelihood Coordinate Systems for Wireless Sensor Networks: from physical coordinates to topology coordinates

04/30/2018 ∙ by Ashanie Gunathillake, et al. ∙ 0

Many WSN protocols require the location coordinates of the sensor nodes, as it is useful to consider the data collected by the sensors in the context of the location from which they were collected. Thus, one of the major challenges in WSNs is to determine the coordinates of sensors while minimizing the hardware cost. To address this, numerous localization algorithms have been proposed in the literature. However, outcomes of these algorithms are affected by noise, fading, and interference. As a result, their levels of accuracy may become unacceptable in complex environments that contain obstacles and reflecting surfaces. The alternative is to use topological maps based only on connectivity information. Since they do not contain information about physical distances, however, they are not faithful representatives of the physical layout. Thus, the primary goal of this research is to discover a topology map that provides more accurate information about physical layouts. In doing so, this research has resulted in four main contributions. First, a novel concept Maximum-Likelihood Topology Map for RF WSNs is presented. This topology map provides a more accurate physical representation, by using the probability of packet reception. The second contribution is Millimetre wave Topology Map calculation, which is a novel topology mapping algorithm based on maximum likelihood estimation for millimetre wave WSNs. The third contribution is a distributed algorithm being proposed to calculate the topology coordinates of sensors by themselves as two algorithms above calculate centrally, which requires time. Since a topology map contains significant non-linear distortions, two WSN applications i.e. target searching and extremum seeking, which use a proposed topology map to localize the sensors and perform its specified task are presented as the final contribution of this dissertation.

READ FULL TEXT
POST COMMENT

Comments

There are no comments yet.

Authors

page 40

This week in AI

Get the week's most popular data science and artificial intelligence research sent straight to your inbox every Saturday.