Table of Content
TABLE OF CONTENT ........................................................................................................................... 1
ABSTRACT ............................................................................................................................................ 2
INTRODUCTION .................................................................................................................................... 3
FAULT TOLERANCE IN WSN .............................................................................................................. 4
LITERATURE REVIEW .......................................................................................................................... 6
RESEARCH OBJECTIVES .................................................................................................................... 7
METHODOLOGY ................................................................................................................................... 8
SIGNIFICANCE .................................................................................................................................... 10
LIMITATION ......................................................................................................................................... 10
POTENTIAL CONTRIBUTIONS (IMPLICATION OF RESEARCH) .................................................... 11
REFERENCES ..................................................................................................................................... 12
APPENDICES ...................................................................................................................................... 13

Page | 1

ABSTRACT
In the last ten years, wireless sensor networks (WSNs) have increasingly gained the attention of researchers. Depending on the applications of WSNs, the sensor nodes are usually dispersed in harsh environments, such as on the ocean floor or in an active volcano, making these nodes more prone to failures. Hence, fault tolerance seems as an essential characteristic that should be considered in the architecture level of these networks. On the other hand, WSNs are battery-powered so that there is a trade-off between preserving the energy and meeting the quality of service requirements of the network. So we are planning for a low computation and efficient algorithm for Distributed Fault detection (DFD) and Fault Tolerance. The main concept is to efficiently use different kinds of the redundancy, time and space. We validate our DFD algorithm through simulation and probabilistic analysis.

Page | 2

INTRODUCTION
Traditional wireless networks are used as a replacement for their wired counterparts.
They provide network infrastructure. In recent years a new type of wireless network emerged very rapidly with a different goal. It provides an effective way for sensing, processing and communicating monitored data. This new type of wireless network is termed as Wireless Sensor Network (WSN). The micro-electro-mechanical systems
(MEMS) technology, wireless communications and digital electronics together enabled the development of low-power, low-cost, multifunctional tiny sensor nodes.
These sensor nodes are densely deployed in the physical phenomenon and collaboratively form a wireless sensor network. These tiny sensor nodes consist of sensing, data processing, and communicating components. The main characteristics of these sensor nodes are in its infrastructure-less and self-organizing capability. The sensor nodes are deployed randomly and they can perform their task in an unattended manner. Another vital feature of sensor network is the cooperation between sensor nodes. Sensor nodes contain on-board processors. Instead of sending raw data sensor nodes transmit partially processed data by carrying out computation locally. Sensor networks are generally used to monitor temperature, humidity, pressure, mechanical stress level, vehicular movement etc. Different types of sensors such as seismic, thermal, infrared, acoustic are used for monitoring.
Sensor Networks can be used for continuous monitoring or event detection. It has widespread application areas which includes:
• Disaster Management
• Military Application
• Environment Monitoring
• Health Care
In spite of all its advantages there are some limitations that influence the performance of WSN. The prominent among such factors are Energy Constraint,
Hardware Constraints, fault tolerance, scalability and so on. Many researchers are currently trying to overcome such limitations and they become successful to rise above these limitations to some extent.
Fault tolerance is the ability of a system to perform its function correctly even in the presence of internal faults. Basic causes behind fault tolerance are fault and failure.
A failure occurs when an actual running system deviates from this specified behaviour. The cause of a failure is called an error. An error represents an invalid system state, one that is not allowed by the system behaviour specification. The error itself is the result of a defect in the system or fault. In other words, a fault is the root cause of a failure. The main concept of all fault tolerance techniques is some form of masking redundancy. This means that components that are prone to failure are
Page | 3

replicated in such a way that if a component fails, one or more of the non-failed replicas will continue to provide service with no appreciable disruption.

Fault Tolerance in WSN
Fault tolerance emerged early as a major concern during design of digital computing systems. Fault tolerance is the ability of a system to perform its function correctly even in the presence of internal faults. Basic causes behind fault tolerance are fault and failure. A failure occurs when an actual running system deviates from this specified behaviour. The cause of a failure is called an error. An error represents an invalid system state, one that is not allowed by the system behaviour specification.
The error itself is the result of a defect in the system or fault. In other words, a fault is the root cause of a failure. The main concept of all fault tolerance techniques is some form of masking redundancy. This means that components that are prone to failure are replicated in such a way that if a component fails, one or more of the non-failed replicas will continue to provide service with no appreciable disruption.
Wireless Sensor Network has due to its wide application areas and limitless future potentials emerge as a premier research topic. WSN consist of thousands of low power low cost tiny sensor nodes. The small size and low cost puts restriction on the hardware and software capabilities of sensor nodes. Sensor nodes are equipped with a limited irreplaceable power source, less processing ability, minimal memory.
Energy is the most important restriction of a WSN. With this limited energy source
WSN have to continue its application from day to years. So there is a trade-off between conserving energy for lifetime increment and maintaining Quality of Service by implementing complex fault tolerance techniques. Sensor nodes are prone to failure due to lack of energy, faults in hardware. Due to its inhospitable working environment often faults in communication occurs in WSN. The harsh environmental condition e.g. fire and rain also cause node misbehaviour, failure of nodes of a particular area, sink failure etc. WSN application fails due to sensor nodes failure; sink failure, communication link error, malicious attack and so on. WSNs are used for many safety and time critical application. In which a failure has an adverse impact on both human and environment. So, Fault tolerance is one of the critical issues in WSN and it need to be tackled with differently. Faults can occur and handled in WSN in four different levels. They are:

Hardware Level
In hardware malfunctioning of any hardware components of a sensor node can cause level faults. Costs of Sensor nodes are low. Therefore its components are not high quality. So they are prone to failure. Sensor nodes have limited irreplaceable
Page | 4

energy source. This strict energy constraint is main hurdle in providing quality of service in WSN application. Sensor nodes are deployed in harsh environment, which can cause failure of sensor nodes as well as sink.

Software Level
Software bugs are common source of errors in WSN. There are two types of software in WSN: system software such as operating system and middleware such as communication, routing etc. It is difficult to provide fault tolerance at hardware level in an economic way. Therefore Fault tolerant approaches are generally applied at middleware level.

Network Communication Level
Failure in network communication occurs due to fault in wireless communication links. This type of failure generally related to harsh physical environment. Error correction and retransmission will enhance the quality of service in this level but they increase overhead and delay.

Application Level
Faults in multiple nodes, sink and network communication link may cause failure of a
WSN application. Addressing fault tolerance at application level is very efficient. But application level fault tolerance is application specific to some extent. But it can address any types of failure like sink, node and communication link. Different types of
Faults in different level cause failure of two main components of WSN: sink and node. Failure of these components can be handled by hardware software or application level.

Page | 5

Literature Review
Numerous fault detection and recovery techniques are proposed in literatures. Some review of existing fault detection and recovery approaches for WSN are presented in this section. A survey on fault tolerance in WSN can be found in different paper with a detailed description on fault detection and recovery is available.
In [1] they had proposed a distributed Bayesian algorithm for sensor fault detection and correction. The algorithm considered that measurement errors due to faulty equipment are likely to be uncorrelated. [2] Had proposed a weighted median fault detection scheme that used spatial correlations among the sensor measurements
(e.g., temperature, humidity).A probabilistic approach to diagnose intermittent WSN faults. The simulation results indicated that the DFD algorithm’s accuracy increased as the number of diagnostic rounds in- creased when neighboring sensor nodes exchanged measurements in each round [3]. DFD algorithm that identified faulty sensor nodes based on comparisons between neighboring sensor nodes’ data. The
DFD algorithm used time redundancy to tolerate transient faults in sensing and communication [4]. [5] Proposed a DFD scheme that detected faulty sensor nodes by exchanging data and mutually testing neighboring nodes. Agnostic diagnosis for detecting silent failures (i.e., failures with unknown types and symptoms). The proposed detection technique exploited the fact that a sensor node’s metrics (e.g., radio on-time, number of packets transmit- ted in a time interval) exhibited certain correlation patterns, the violation of which indicated potential silent failures. The detection accuracy of the proposed scheme was close to 100% for small WSNs, whereas the detection accuracy decreased sharply and the false alarm rate increased as the WSN size increased [6]. Application layer approaches to achieve fault tolerance is discussed in [7]. Technique for handling sink and node failure in a multi-sink WSN is proposed in [8]. In this method all sinks gather knowledge about the topology of WSN by collecting information from each node of the network. Then the sinks exchange their information to create a unique global topology. From this global topology, multiple redundant routes for each sensor node are computed, and forwarding tables are built. Finally, these forwarding tables are transmitted to the respective sensor nodes. Now maximum sensor nodes have more than one routing paths to different sink. So, when a sink fail the failure handled automatically by routing the data to other sinks using alternative routing paths. If a routing path becomes unavailable due to node failure then this situation can be handled by routing the data through redundant path. This technique incorporates a large overhead for collecting topology information, computing the global topology and transmitting the forwarding table. An approach for recovery from node and sink failure is discussed in [9]. It partition the sensor network into different cells, where a cell
Page | 6

contains one data sink s and any sensor that has a smaller hop count to s than to any other data sink. If a sensor has the same hop count to two or more data sinks, then that node is termed as border node and it will belong to the cell of each of these data sinks and it contains addresses related to all cells. If some data sink fails this protocol recover from it by attaching the boarder node only to the non- 58 failed cell and the border node then contain only the address related to the non-failed cell. Then border node broadcast the address related to its non-failed cell and other nodes from the failed sink’s cell attach themselves to boarder node’s non-failed cell. Here node failure is also handled in a very similar manner. Base stations can be attacked and even compromised like any other sensor node on the network. Candidate base stations are base stations with similar characteristics like currently active base stations. They just pretend to be ordinary nodes on the sensor network. But when there is a base station compromise/failure they are resorted to play base station roles. This approach is elaborated in [10]. It is very efficient for failure recovery but providing candidate base station with each primary base station is not cost effective.
A scheme for handling node failure in a clustered WSN is proposed in [11]. Here nodes of a cluster are classified into different types like head node, boundary node, pre boundary node and internal node. When a node is going to die it send a failure report to its entire neighbour. Now each type of receiving node starts its own recovery mechanism. This scheme only deal with node failure due to energy exhaustion and it also assume that the failing node broadcasts its failure report before die out completely. Research Objectives
The main objective of the thesis is to develop new approaches for providing energy efficiency, longer lifetime, and fault tolerance for WSNs, which are mainly used for time critical applications like disaster management. A number of algorithms have been proposed in the literature to solve these problems. A detailed literature survey will be done for developing an idea about the solution. This thesis studies the performances of some existing algorithms and proposes novel algorithms for fulfilling its objective. Performances of all algorithms will be studied and will implement in a simulation environment and will analysis the simulation results.

Page | 7

Research Questions.
General
•

How can fault tolerance be improved in WSNs?

Specific
•
•

How can fault tolerance algorithm be energy efficient?
How can Distributed Fault Detection algorithm be improved?

Methodology

To structure the work to be done, a project management approach will be followed.
The work will be divided in 4 activities that will be subdivided in Tasks.
A1
A2
A3
A4

Curricular Year
Related Work
Development and testing
Elaboration of the Thesis

Back-­‐ ground

Develop-­‐ ment and

Testing

A4 - Elaboration of the thesis

A2 – Related Work

A graphical presentation of the activities showing their interdependencies in shown in Figure below:

Page | 8

Background
The Background is composed of 4 tasks:
•

Methodologies for Scientific Research

•

Knowledge Extraction

•

Simulation Environments

•

Research Planning

Related Work
State of the art
To enlarge the knowledge and improve the concept of this thesis, an ongoing study on the related work is in progress aimed mainly at fault tolerance areas although other complimentary areas are also being studied such as distributed fault detection and energy efficient algorithms. During this task a comparison matrix/methodology will be defined. A common approach would be to use statistics for this purpose.

Elaboration of the thesis
The elaboration of the thesis will be made during all phases of this research but after the results are obtained and tested against the other algorithms, finally it will be the time to conclude the thesis and hopefully the results will be a good contribution to the scientific community that acts not only in WSNs but also other technologies that need Fault tolerance algorithm.

Page | 9

Significance
The world’s need for monitoring has increasing since the recent attacks to critical infrastructures. Monitoring is now ubiquitous and could be present in: Monitoring space (environmental and habitat monitoring, precision agriculture) Monitoring things
(indoor climate control, surveillance and intelligent alarms) Monitoring space and things (the interactions of things with each other and the surrounding space). So in order to achieve a good result there should be some algorithm to detect and tolerate the level of fault that can occur in the system. For example:
•
•
•
•

Wildlife habitats;
Disaster management;
Emergency response;
Healthcare;

Limitation

With the advances in processing power, sensor accuracy, miniaturization and production costs, WSNs will be used in a wider range of applications in the near future. Already deployed WSNs serve as a proof of concept, providing the means to monitor environmental data formerly impossible to collect due to the inaccessibility or characteristics of the environments where such physical phenomena take place.
Research on this field is very promising and has been attracting the scientific community for the last years. Some of the more active research areas are WSN as a service network, Database like network (query), Distributed processing and storage algorithms, Routing protocols, Network Auto-Setup (self-organization and self- maintenance), Fault tolerance algorithm. Recent developments prove that there are many applications feasible with the available technology; assuring that in a near future WSNs will become commodities in what concerns sense and instrument the physical world. Furthermore WSNs are a factor of progress, enhancing productivity in sectors such as agriculture, transport and construction. The fault tolerance algorithm proposed will be tested against the other technologies available and in case of less satisfactory results, the reasons will be assessed and the algorithms will be optimized to achieve better performance.

Page | 10

Potential Contributions (Implication of Research)

The proposed thesis will focus on contributing to the scientific community with a
Fault tolerance algorithm more accurate and effective that the ones in use. The development of Fault tolerance algorithms for wireless sensor networks will try to accomplish the following over the other algorithms:
•

Better accuracy;

•

Better performance;

•

Better efficiency;

•

Scalability;

•

Energy efficient;

Page | 11

References
[1] Bhaskar Krishnamachari and Sitharama Iyengar, "Distributed Bayesian algorithms for fault-tolerant event region detection in wireless sensor network”. IEEE
Transactions on Computers, 2004.
[2] Gao Jian-Liang, Xu Yong-Jun, and Li Xiao-Wei, "Weighted-median based distributed fault detection for wireless sensor networks". Journal of Software, 2007.
[3] P. Khilar and S. Mahapatra," Intermittent fault diagnosis in wireless sensor networks”, 2007.
[4] M. Lee and Y. Choi, " Fault detection of wireless sensor networks. Elsevier
Computer Communications”, 2008.
[5] Peng Jiang, "A new method for node fault detection in wireless sensor networks”,
2009
[6] Xin Miao, Kebin Liu, Yuan He, Dimitris Papadias, Qiang Ma, and Yunhao Liu,
"Discovering silent failures in wireless sensor networks". IEEE Transactions on
Wireless Communications, 2013
[7] H. liu, A. Nayak, I. Stojmenovi´c, “Fault Tolerant Algorithms/Protocols in Wireless
Sensor Networks”, Guide to Wireless Sensor Networks, Springer, 2009.
[8] J. Deng, R. Han, S. Mishra, “Enhancing Base Station Security in Wireless Sensor
Networks”, Department of Computer Science, University of Colorado, Tech. Report
CUCS-951-03, 2003.
[9] H. Lee, A. Klappenecker, K. Lee, L. Lin, “Energy Efficient Data Management for
Wireless Sensor Networks with Data Sink Failure”, IEEE International Conference on
Mobile Ad-hoc and Sensor Systems conference, November 2005.
[10] N. Evarist, Y Lin, “Candidate Base Stations a Security Solution for
Compromised Base Stations in Wireless Sensor Networks”.
[11] G. Venkataraman, S. Emmanuel, S. Thambipillai, “Energy-efficient clusterbased scheme for failure management in sensor networks”, April 2008

Page | 12

Appendices

Time Table
1. Curricular Year
2. Related Work
3.Development and testing
3a.Development of the framework components 3b.Comparative analysis
3c.Elaboration of the thesis
4. Documentation

June 10 to July 15,2015
July 15 to September 30,2015
September 30 to October 30,2015
October 30 to December 25,2015
December 25 to February 15,2015
February 15 to April 15,2015
April 15 to July 05,2015

Page | 13

Page | 14