Skip to Main content Skip to Navigation
Theses

Conception of a wireless sensor network for decision making based on Data mining methods

Abstract : Recently, Wireless Sensor Networks (WSNs) have emerged as one of the most exciting fields. However, the common challenge of all sensor network applications remains the vulnerability of sensor nodes due to their characteristics and also the nature of the data generated which are of large volume, heterogeneous, and distributed. On the other hand, the need to process and extract knowledge from these large quantities of data motivated us to explore Data mining techniques and develop new approaches to improve the detection accuracy, the quality of information, the reduction of data size, and the extraction of knowledge from WSN datasets to help decision making. However, the classical Data mining methods are not directly applicable to WSNs due to their constraints.It is therefore necessary to satisfy the following objectives: an efficient solution offering a good adaptation of Data mining methods to the analysis of huge and continuously arriving data from WSNs, by taking into account the constraints of the sensor nodes which allows to extract knowledge in order to make better decisions. The contributions of this thesis focus mainly on the study of several distributed algorithms which can deal with the nature of sensed data and the resource constraints of sensor nodes based on the Data mining algorithms by first using the local computation at each node and then exchange messages with its neighbors, in order to reach consensus on a global model. The different results obtained show that the proposed approaches reduce the energy consumption and the communication cost considerably which extends the network lifetime.The results also indicate that the proposed approaches are extremely efficient in terms of model computation, latency, reduction of data size, adaptability, and event detection.
Complete list of metadatas

https://tel.archives-ouvertes.fr/tel-01688802
Contributor : Abes Star :  Contact
Submitted on : Friday, January 19, 2018 - 5:06:09 PM
Last modification on : Wednesday, September 16, 2020 - 9:57:01 AM
Long-term archiving on: : Thursday, May 24, 2018 - 8:01:22 AM

File

These-2017-MATHSTIC-Informatiq...
Version validated by the jury (STAR)

Identifiers

  • HAL Id : tel-01688802, version 1

Citation

Massinissa Saoudi. Conception of a wireless sensor network for decision making based on Data mining methods. Networking and Internet Architecture [cs.NI]. Université de Bretagne occidentale - Brest, 2017. English. ⟨NNT : 2017BRES0065⟩. ⟨tel-01688802⟩

Share

Metrics

Record views

711

Files downloads

218