Authors:
Constantin Voloşencu
1
;
Daniel-Ioan Curiac
1
;
Ovidiu Banias
1
;
Alex Doboli
2
and
Octavian Dranga
3
Affiliations:
1
“Politehnica” University of Timisoara, Romania
;
2
State University of New York, United States
;
3
School of Engineering, James Cook University, Australia
Keyword(s):
Perimeter protection, wireless sensor networks, redundancy, knowledge based systems, neural networks, perceptron.
Related
Ontology
Subjects/Areas/Topics:
Mobile Ad-Hoc and Sensor Networks
;
Sensor, Mesh and Ad Hoc Communications and Networks
;
Telecommunications
;
Wireless Information Networks and Systems
Abstract:
This paper provides a strategy for perimeter protection using sensor networks with hardware and analytical redundancy. The sensor network reliability is augmented using a knowledge-based system, which implicates the analysis of the trustworthiness of each sensor. For this, we used two stratagems: one that relies on hardware redundancy based on the Confidence Weighted Voting Algorithm and one that relies on analytical redundancy based on a neural perceptron predictor that uses past and present values obtained from neighbouring nodes. This solution can be also a way to discover the malfunctioning nodes that were subjects of an attack and it is localized at the base station level being suitable even for large-scale sensor networks.