A Fast Clustering Algorithm for Analyzing Big Data Generated in Ubiquitous Sensor Networks

Recently, The emergence of wireless sensor networks (WSNs) plays a major role in the rise of big data as thousands of their applications collect huge amounts of data that require processing. Consequently, WSN faces two major challenges. First, it handles the big data collection, and second, the ener...

Full description

Saved in:
Bibliographic Details
Published in2018 International Arab Conference on Information Technology (ACIT) pp. 1 - 6
Main Authors Zahwe, Oussama, Majed, Ola, Harb, Hassan, Hamze, Mohamad, Nasser, Abbass
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.11.2018
Subjects
Big Data
Clustering algorithms
Data clustering
Data communication
Data models
Energy consumption
Euclidean distance
FKmeans algorithm
Periodic applications
Real sensor data
Wireless sensor networks
Online AccessGet full text

Cover

More Information
Summary:Recently, The emergence of wireless sensor networks (WSNs) plays a major role in the rise of big data as thousands of their applications collect huge amounts of data that require processing. Consequently, WSN faces two major challenges. First, it handles the big data collection, and second, the energy of sensors will be depleted quickly due to the huge volume of data collection and transmission. Hence, current research has been focused on data classification as an efficient technique to reduce big data collection in WSNs thus enhancing their lifetime. This paper proposes a fast data clustering technique called FKmeans, i.e. Fast Kmeans, dedicated to periodic applications in WSNs. FKmeans consists of two stage algorithm and aims to enhance the time cost of distance calculation of traditional Kmeans algorithm thus, ensure fast data delivery to the sink node. The first stage, i.e. center selection stage, selects a small portion of datasets in order to find the best possible location of the centers. The second stage, i.e. cluster formation stage, uses the traditional Kmeans algorithm adopted to the Euclidean distance where the initial centers used are taken from the first stage. Our proposed technique is validated via simulations on real sensor data and comparison with the traditional Kmeans algorithm. The obtained results show the effectiveness of our technique in terms of improving the energy consumption and data delivery delay, without loss in data fidelity.
DOI:10.1109/ACIT.2018.8672680