A Comparative Evaluation of Mining Techniques to Detect Malicious Node in Wireless Sensor Networks

A Comparative Evaluation of Mining Techniques to Detect Malicious Node in Wireless Sensor Networks

Mandeep Singh (Lovely Professional University, Jalandhar, India), Navjyot Kaur (Lovely Professional University, Jalandhar, India), Amandeep Kaur (Lovely Professional University, Jalandhar, India) and Gaurav Pushkarna (Lovely Professional University, Jalandhar, India)
Copyright: © 2017 |Pages: 12
DOI: 10.4018/IJCWT.2017040103

Abstract

Wireless sensor networks have gained attention over the last few years and have significant applications for example remote supervising and target watching. They can communicate with each other though wireless interface and configure a network. Wireless sensor networks are often deployed in an unfriendly location and most of time it works without human management; individual node may possibly be compromised by the adversary due to some constraints. In this manner, the security of a wireless sensor network is critical. This work will focus on evaluation of mining techniques that can be used to find malicious nodes. The detection mechanisms provide the accuracy of the classification using different algorithm to detect the malicious node. Pragmatically the detection accuracy of J48 is 99.17%, Random Forest is 80.83%, NF Tree is 81.67% and BF Tree is 72.33%. J48 have very high detection accuracy as compared with BF Tree, NF Tree Random Forest.
Article Preview

Introduction

Wireless Sensor Networks (WSN) consists of sensor nodes containing simple processors, low power consuming antennas and various detectors. The sensor nodes have capability to store and handle information locally because of the software and hardware structure they have. Generally wireless sensor networks use radio communication and approximately all radio-based environment use COTS (Commercial Off-The-Shelf) components (Puccinelli, 2005). The lifetime of network is very critical for most applications and it is affected by energy consumption of the nodes. Although it is frequently assumed that transmitting and receiving data causes the main energy drain (Choi, 2010; Praveena, 2006). The energy is not only consumed by the node when it is communicate, but also in numerous endless control processes, like keeping a routing table for data collection, or ongoing communication at the MAC layer (Cheng, 2011; Ye, 2002).

The energy consumption may possibly be reduced by following ways (Puccinelli, 2005):

  • Energy-efficient routing protocol is avoiding the battery diminution of a node.

  • Certain protocols are disposed to minimize energy consumption on forwarding pathways. The nodes which are situated adjacent to the base station, their lifetime have been reduced.

  • In Medium Access Control (Ye, 2002), the activities like control packet overhead, collisions and idle listening have a direct influence on energy consumption.

WSN is a significant exploration area, mostly with current developments in embedded systems and wireless communications. The nodes in a sensor network are much higher than in ad hoc network. To ensure coverage and connectivity the impenetrable organizations is often anticipated (Dener, 2015; Raza, 2012). Each node consists of a central processing unit, memory, battery, one or more sensors and a radio trans-receiver as shown in Figure 1. The network can be scaled from 10 to 100 of nodes. It perfectly associates with existing wired environment and control systems which has the ability to supervise, watch and follow target remotely (Jenkins, 2014). The frequently supervised factors are temperature, humidity, pressure, wind direction, speed, and chemical concentrations. They are supervising surrounding conditions and assets available in the network.

Figure 1.

Wireless sensor network communication structure

The significant feature of WSN consists of the situation use in emergency statuses as it is self-organizing and self-maintaining (Yueqing, 2010; Sohrabi, 2000) In majority of cases, the nodes are randomly deployed for supervising and target following. The network topology is asymmetrical as a result of randomly deployed (Younis, 2006). The nodes keep in touch with the base station through multi-hop paths and the paths are established with the assistance of routing algorithms. The network commonly comprises of an enormous amount of sensor nodes. The sensor node can do exact sensing and extendibility of the sensing areas (Choi, 2010; Wang, 2009). Once in a while the surroundings are hostile, so node failure is significant, sometimes a node is temporarily unreachable, and communication causes major packet loss (Jenkins, 2014).

The WSNs have significant benefits over traditional communication technologies, which include rapid deployment, low cost, flexibility, and parallel processing (Gungor, 2009).

WSNs have plentiful applications (In Figure 2) such as, forest fire detection, surveillance, and object tracking, health monitoring, surveillance environment monitoring and agriculture monitoring (Low, 2005) are only a few examples. Event-driven and continuous data collections are two categories of these applications.

Figure 2.

Applications of wireless sensor network

There are a number of challenges in WSNs (Praveena, 2006; Mahjoub, 2007):

Complete Article List

Search this Journal:
Reset
Open Access Articles: Forthcoming
Volume 9: 4 Issues (2019): Forthcoming, Available for Pre-Order
Volume 8: 4 Issues (2018)
Volume 7: 4 Issues (2017)
Volume 6: 4 Issues (2016)
Volume 5: 4 Issues (2015)
Volume 4: 4 Issues (2014)
Volume 3: 4 Issues (2013)
Volume 2: 4 Issues (2012)
Volume 1: 4 Issues (2011)
View Complete Journal Contents Listing