Citation: Poggi, B.; Babatounde, C.;
Vittori, E.; Antoine-Santoni, T.
Efficient WSN Node Placement by
Coupling KNN Machine Learning for
Signal Estimations and I-HBIA
Metaheuristic Algorithm for Node
Position Optimization. Sensors 2022,
22, 9927. https://doi.org/10.3390/
s22249927
Academic Editor: Tian Wang
Received: 12 September 2022
Accepted: 29 November 2022
Published: 16 December 2022
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Article
Efficient WSN Node Placement by Coupling KNN Machine
Learning for Signal Estimations and I-HBIA Metaheuristic
Algorithm for Node Position Optimization
Bastien Poggi
1,
* , Chabi Babatounde
2
, Evelyne Vittori
1
and Thierry Antoine-Santoni
3
1
UMR CNRS 6134 SPE, University of Corsica, 20250 Corte, France
2
UAR CNRS 3514 Stella Mare, University of Corsica, 20250 Corte, France
3
UMR CNRS 6240 LISA, University of Corsica, 20250 Corte, France
* Correspondence: poggi_b@univ-corse.fr
Abstract:
Wireless sensor network (WSN) deployment is an intensive field of research. In this pa-
per, we propose a novel approach based on machine learning (ML) and metaheuristics (MH) for
supporting decision-makers during the deployment process. We suggest optimizing node positions
by introducing a new hybridized version of the “Hitchcock bird-inspired algorithm” (HBIA) meta-
heuristic algorithm that we named “Intensified-Hitchcock bird-inspired algorithm” (I-HBIA). During
the optimization process, our fitness function focuses on received signal maximization between
nodes and antennas. Signal estimations are provided by the machine learning “K Nearest Neighbors”
(KNN) algorithm working with real measured data. To highlight our contribution, we compare the
performances of the canonical HBIA algorithm and our I-HBIA algorithm on classical optimization
benchmarks. We then evaluate the accuracy of signal predictions by the KNN algorithm on different
maps. Finally, we couple KNN and I-HBIA to provide efficient deployment propositions according to
actual measured signal on areas of interest.
Keywords: wireless sensor network; deployment; optimization; KNN; HBIA
1. Introduction
In the past two decades, the number of deployed wireless sensors has increased
dramatically. This spectacular growth is on the one hand correlated with the development
of machine learning applications requiring large amounts of data and on the other hand
made possible by the reduction in the production costs of the equipment.
Following this trend, many new technologies have emerged supporting different
routing protocols, different operating systems, and different micro-controller boards. They
have been successfully applied in several fields of research, and application such as envi-
ronmental monitoring [1,2], smart agriculture [3], health [4], and security [5].
This current heterogeneity of technologies generates new challenges [
6
]. To deal with
it, researchers have to provide new tools to help decision-makers during WSN deploy-
ment. Many issues need to be addressed: “Which technology is the most suitable for my
problem?”; “How to configure WSN parameters?”; “Where to deploy sensors in order to
observe a specific phenomenon?”; “How to ensure node connectivity within the studied
area according to its specificities and constraints?”.
One of the most encountered problems with WSN is their efficient placement in an
area of interest (AoI). On star topology, all sensors must receive signals from one or more
fixed antennas. Many times these antennas are placed in specific places accepted by the
population. Signal propagation is rarely the main criterion considered. Citizens usually
focus first on other criteria, such as visual impact and health protection. Node localization
must reconcile these constraints and provide a minimum level of connectivity.
Sensors 2022, 22, 9927. https://doi.org/10.3390/s22249927 https://www.mdpi.com/journal/sensors
