bullet Sensors & Transducers Journal

    (ISSN 1726- 5479)


2008 e-Impact Factor

25 Top Downloaded Articles

Journal Subscription 2012

Editorial Calendar 2012

Submit an Article

Editorial Board

Current Issue

Sensors & Transducers journal's cover

Sensors & Transducers Journal 2011

Sensors & Transducers Journal 2010

Sensors & Transducers Journal 2009

Sensors & Transducers Journal 2008

Sensors & Transducers Journal 2007

2000-2002 S&T e-Digest Contents

2003 S&T e-Digest Contents

2004 S&T e-Digest Contents

2005 S&T e-Digest Contents

2006 S&T e-Digest Contents

Best Articles 2011




Vol. 14-2, Special Issue, March 2012, pp.64-78




Energy Efficient in-Sensor Data Cleaning for Mining Frequent Itemsets


Jacques M. BAHI, Abdallah MAKHOUL, * Maguy MEDLEJ

Computer Science Laboratory (LIFC), University of Franche-Comté,

Rue Engel-Gros, 90016, Belfort, France

E-mail: jacques.bahi@univ-fcomte.fr, abdallah.makhoul@univ-fcomte.fr, maguy.medlej@univ-fcomte.fr



Received: 7 November 2011   /Accepted: 20 December 2011   /Published: 12 March 2012

Digital Sensors and Sensor Sysstems


Abstract: Limited energy, storage, computational power represent the main constraint of sensor networks. Development of algorithms that take into consideration this extremely demanding and constrained environment of sensor networks became a major challenge. Communicating messages over a sensor network consume far more energy than processing it and mining sensors data should respect the characteristics of sensor networks in terms of energy and computation constraints, network dynamics, and faults. This lead us to think of a data cleaning pre processing phase to reduce the packet size transmitted and prepare the data for an efficient and scalable data mining. This paper introduces a tree-based bi-level periodic data cleaning approach implemented on both the source node and the aggregator levels. Our contribution in this paper is two folds. First we look on a periodic basis at each data measured and periodically clean it while taking into consideration the number of occurrences of the measures captured which we shall call weight. Then, a data cleaning is performed between groups of nodes on the level of the aggregator, which contains lists of measures along with their weights. The quality of the information should be preserved during the in-network transmission through the weight of each measure captured by the sensors. This weight will constitute the key optimization of the frequent pattern tree. The result set will constitute a perfect training set to mine without higher CPU consumption allowing us to send only the useful information to the sink. The experimental results show the effectiveness of this technique in terms of energy efficiency and quality of the information by focusing on a periodical data cleaning while taking into consideration the weight of the data captured.


Keywords: Sensor networks, Periodic data aggregation, Tree based algorithms, Quality of information, Training set to mine, Frequent Pattern tree (FP_Tree)


Acrobat reader logo Click <here> or title of paper to download the full pages article (1.31 Mb)



Sensors & Transducers journal subscription

only 450 $ US per year:




Read more about Wireless Sensors and Wireless Sensor Networks





1999 - 2012 Copyright ©, International Frequency Sensor Association (IFSA). All Rights Reserved.

Home - News - Links - Archives - Tools - Voltage-to-Frequency Converters - Standardization - Patents - Marketplace - Projects - Wish List - e-Shop - Sensor Jobs - Membership - Videos - Publishing - Site Map - Subscribe - Search

 Members Area -Sensors Portal -Training Courses - S&T Digest - For advertisers - Bookstore - Forums - Polls - Submit Press Release - Submit White Paper - Testimonies - Twitter - Facebook - LinkedIn