A faulty node detection scheme for wireless sensor networks that use data aggregation for transport. Issue 14 (12th January 2016)

• Record Type: Journal Article
• Authors: Artail, Hassan; Ajami, Abdelkarim; Saouma, Tania; Charaf, Malak
• Is Part Of: Wireless communications and mobile computing. Volume 16:Issue 14(2016)
• Page Start: 1956
• Page End: 1971
• Publication Date: 2016-01-12
• Subjects: wireless sensor networks -- data aggregation -- faulty node detection -- Markov chain -- data reliability; Wireless communication systems -- Periodicals; Mobile communication systems -- Periodicals; 621.38205
• DOI: 10.1002/wcm.2661
• Languages: English
• ISSNs: 1530-8669

Abstract

Abstract: This paper presents a faulty node detection approach for wireless sensor networks that aggregate measurement data on their way toward the sink (base station). The approach is based on the idea of commanding sensor nodes on the aggregation paths to temporarily stop including their readings in the received aggregated readings from their upstream neighbors. The scheme is dependent on the ability of the sink to detect faulty nodes through changes in the received aggregated readings at the sink using a Markov Chain Controller (MCC). The algorithm that is run in the sink uses the MCC to assign a state to each sensor node based on transitions that are triggered by receiving aggregated path readings, and accordingly deduces the nodes that may be faulty. The experimental results show at least 98% detection rate at the cost of reasonable detection delays and generated wireless network traffic. Copyright © 2016 John Wiley & Sons, Ltd. Abstract : We present a faulty node detection scheme for wireless sensor nodes that report aggregated measurements to the data sinks. The scheme uses time division for stopping nodes selected in random order from contributing readings to the ones from upstream neighbors. It runs at the sink to derive the states of the nodes by comparing the readings before and after turning nodes on and off. The performance results show at least 98% detection rate with reasonable detection delays and energy consumption.