An analysis of fault detection strategies in wireless sensor networks

Abstract

Wireless sensor networks have emerged as a key technology which is used in many safety critical applications. The sensors in wireless sensor network have to be deployed in hostile, harsh and unattended environments for long periods of time. This creates a great challenge in providing a good quality of service. This results in introductions of faults, sensor failures, communication failures and changes in topology. Hence, efficient fault detection techniques are required for good quality of service. In this article, we survey various fault detection techniques and provide a new taxonomy to integrate new fault detection techniques. We perform a qualitative comparison of the latest fault detection algorithms. From a qualitative analysis, we select a list of techniques that are analyzed quantitatively. We also discuss the shortcomings, advantages and future research directions for fault detection in wireless sensor networks.

Introduction

Wireless sensor networks consist of a large number of independent sensor nodes connected together to form a network. Each of these individual sensor nodes in a wireless sensor network has sensing and processing capability. Wireless communication is used as a medium for communicating between the nodes. It forms an ad hoc network with peer–peer communication. Wireless sensors are low power devices with limited computational power, memory, battery and storage. They are deployed in hostile and harsh conditions to report time-critical events such as landslide monitoring (Ramesh, 2014), agricultural monitoring (Pantazis et al., 2013), military operations (Akyildiz et al., 2002), infrastructure monitoring, scientific data collection, intruder detection system (Casey et al., 2008), navigation (Varshney, 2004), and environmental monitoring (Casey et al., 2008). Sensor networks are also used to monitor the health of patients in hospitals (Kim and Prabhakaran, 2011).

Since, sensor networks are deployed in hostile and harsh conditions, e.g., rain, snow, wind, thunder, etc., they are susceptible to frequent and unexpected errors. The faults may be due to hardware or software failure. These faults result in erroneous results in normal operation. The occurrence of faults during normal operation results in harsh consequences involving loss of human life, economic and environmental loss as sensor networks are used in safety catastrophic disasters. For example, if the wireless sensor detecting the activity of the volcano malfunctions and gives incorrect readings, it might result in unneeded panic or loss of lives due to the absence of warning.

The presence of faults in wireless sensor data, may increase the network traffic, decrease the fault detection efficiency of the base station and wastes the battery and power. They need to conserve battery as they are supposed to operate for periods of time ranging from hours to years. Moreover, replacing the battery is not feasible since sensors are normally deployed at inaccessible locations. The transmission of collected data from nodes to sink is an expensive process and results in congestion (Ni and Pottie, 2012, Yu et al., 2007, Liu et al., 2006, Rosberg et al., 2010, Anisi et al., 2013). The presence of faulty sensor nodes increases the congestion by transmitting unusable and misleading data. Therefore, fault detection techniques are required for proper management of the sensor network. Fault detection algorithms enhance bandwidth utilization and data reliability. However, the energy consumption by the nodes increases due to complex fault detection techniques. Hence, there is a tradeoff between conservation of energy versus maintaining a high Quality of Service (Yu et al., 2007). Fault detection techniques detect the faulty node in a wireless sensor network and create a record of all faulty nodes, which can be used for fault recovery of the node or replacement of the faulty node or isolate faulty sensor nodes from the network (Mahapatro and Khilar, 2013).

Yu et al. (2007) briefly discuss fault management in WSN. The fault detection strategy classification provided in Yu et al. (2007) is outdated. Moreover, it does not incorporate a discussion of faults in WSN. A discussion on the faults in WSN can be found in Jurdak et al. (2011) where they provide a classification of faults based on the source of faults. Jurdak et al. (2011) provide insight about the suitability of fault detection tools for various faults along with the usability of fault detection tools. The discussion of fault detection tools and techniques in existing surveys is too brief and concise to be constructive. Moreover, the classification of fault detection techniques provided is limited. Although existing classifications provide the basic architecture, they do not provide a further classification based on latest fault detection techniques in WSN. Mahapatro and Khilar (2013) also discuss various fault detection techniques in WSN but none of these surveys provide a quantitative analysis.

In this survey, we provide a discussion on latest fault detection techniques based on our proposed taxonomy. We also classify and mathematically model the faults that occur in wireless sensor networks. By analyzing the existing literature work, we outline prospective future research challenges and issues. We qualitatively and quantitatively analyze and compare various latest fault detection schemes for wireless sensor networks. We use the Intel-Berkeley data set (http://db.csail.mit.edu/labdata/labdata.html) for quantitative analysis. To the best of our knowledge, this survey is the first of its kind to provide quantitative comparisons in this area of research.

The rest of the paper is organized as follows. In Section 2, we discuss various faults that occur in WSN based on proposed taxonomy. In Section 3, we propose a new technique based taxonomy for wireless fault detection. Thereafter, in Section 4 we discuss the state of art fault detection techniques. We present a quantitative and qualitative analysis of various fault detection techniques in Section 5. In Section 6 we outline future research issues and directions for fault detection in WSN and finally Section 7 summarizes our observations on current state and trends in fault detection techniques of WSN.