Vous êtes sur la page 1sur 4

IJSRD - International Journal for Scientific Research & Development| Vol.

4, Issue 05, 2016 | ISSN (online): 2321-0613

Fault Tolerance and Clustering Algorithm in WSN-A Review


Mandeep Kaur1 Parul Garg2
1,2
Department of Computer Science & Engineering
1,2
GZS Campus College of Engineering & Tech. Bathinda, Punjab, India
Abstract WSN is a wireless sensor network consisting of a failures. Failure can be of any form, it might be due to
large no of sensor nodes. These nodes are deployed in the environment factors, hardware failure, any type of damage
network for its feature of monitoring. The process of to the sensors, nodes are running out of power, software
gathering data and communication takes place from nodes to failure or any type of intended attack done to cause failure.
base stations. The base station is the sink to collect all the [12] FT ensures that the work done is reliable and has the
data. These sensor nodes have some limitations like limited ability to tackle any kind of fault. [12]FT mechanisms are
energy in the form of in-built battery, storage capacity. A classified as redundancy based, clustering based and
WSN can go through different types of failure like deployment based. One of the efficient mechanisms of FT is
connectivity, link, node and malfunctions of nodes. WSN clustering. It is an efficient method to save energy and
should be good at identifying the fault and put it right instead control the network.
of discarding it. Energy constraint is one of the main Energy consumption is the most important aspect
problems that occur when sensors are deployed in the to actuate the life of a WSN because sensor nodes work on
environment. So to create a network which consume less in-built batteries. Sometimes energy optimization is more
energy and has a long lifetime is a challenge. In this paper, difficult in sensor networks because it involves not only
fault tolerance (FT) techniques and clustering algorithm is degradation of energy consumption but also stretching the
discussed. life of the network as more as possible. The optimization
Key words: WSN, Fault, Energy, Fault Tolerance, Cluster, can be done by having energy consciousness in every facet
Cluster Head, Clustering Algorithms of design and operation. This secures that energy awareness
is also assimilated into groups of communicating sensor
I. INTRODUCTION nodes and the whole network and not only in the individual
A wireless sensor network consists of so many sensor nodes, nodes. To decrease the overall energy consumption of the
which are deployed in the environment [5]. The sensor sensor network, various types of protocols and algorithms
nodes collect the information from the environment and have been studied. The lifetime of a sensor network can be
transmit the data to the base station. Then the information is enlarged significantly if the operating system, the
transmitted to the end user for its use. In the WSN, sensors application layer and the network protocols are designed to
are dynamic in nature as any sensor nods or links may fail at be energy aware.
any point of operation. Replacing faulty nodes is not an easy B. Clustering Algorithm
task so fault tolerance or reliability measure is a very Cluster formation is the process to form clusters in the
important research challenge that needs to be focused. network. Each cluster consists of CH and cluster members.
A. Fault Tolerance Cluster member send data to the CH and CH
Fault in sensor network cannot be tolerated and recovered forwards the data to the base station (BS). When the data is
directly using the approaches employed with traditional forwarded from one CH to another CH. The process is
wired or wireless network. In WSN failures occur more known as inter-cluster communication.[11] When the data
often than traditional networks, where client machines, transmission takes place from CH to BS. As shown in Fig 1.
servers and routers most of the times operate in normal
conditions. Also, most of the approached used with
traditional networks we do not consider device energy
consumption an issue as they have a fixed power supply to
their system. Various ways that may cause fault in sensor
nodes. [13]Firstly, it is due to energy depletion of the sensor
nodes that means when the node whole energy is consumed
to answer different data request coming from users.
Secondly, it is due to sensor hardware failure which may
include measurement, transmitting or receiving faults. The
third one is due to fault in communication links between
different sensor nodes. A monitor node is added to the
sensor network, where it will detect the fault and report
them to base station. In it mentions the type of fault
occurred in the network and the monitor node maintains the
Fig. 1: Communication b/w clusters
health of the network. The monitor node transmits all the
The process is termed as intra-cluster
monitored traffic and dedicated monitors in a separate
communication. Clustering can be done in one hope or multi
channel. A monitoring node can either be placed directly on
hop way. Members can directly send data to the CH or they
a sensor node or inside the sensor network near to the sensor
may pass the data to the nearer cluster member. And then
nodes. The basic need of a system is to make it work for as
further that member will pass the data to Clustering can be
long as one can. While working it undergoes so many
done in one hope or multi hop way. Members can directly

All rights reserved by www.ijsrd.com 373


Fault Tolerance and Clustering Algorithm in WSN-A Review
(IJSRD/Vol. 4/Issue 05/2016/093)

send data to the CH or they may pass the data to the nearer B. A Cluster-Member Based Fault Tolerant Mechanism
cluster member. And then further that member will pass the [8]This scheme proposed a mechanism in which detection
data to the CH. CH can be selected randomly or using cost and recovery is done locally side by side while network is
function, or calculating the average energy and on that bases operational. The nodes monitor the link with the CH, at time
decide the CH probability. of failure of CH. They detect the failure and re-select the CH
for proper functioning of the network. It works in two phase
II. RELATED WORK Fault detection phase detects the fault as cluster head
[10] In this Leach protocol is introduced to increase the life communicate with each other through beacon message so the
time of network is increased using clustering. [9]In this cluster members send data to another CH when they dont
technique the cluster head exchange all its information with get an acknowledgement from the first CH which means a
the neighbour CH. On failure of any CH, neighbour CH fault has occurred. The second phase is the recovery from
consult each other and negotiate among them to take over fault. Every node is also linked to the neigbouring CH. Once
that cluster under them. [8]In CMATO author proposed the its head is dead it sends data to the neighbouring CH or the
mechanism in which detection and recovery is done locally newly formed CH. So the recovery is possible because it is in
side by side while network is operational. [7] Author the range of two CH. This scheme gives a way to recover
proposed a scheme which is based on cross-layer design of from medium error or the CH error.
Mac layer. When the data is transmitted, sender waits for an C. Low Energy Adaptive Clustering Algorithm-Centralized
acknowledgement. It relies with an ACK after a short inter
frame space. If not, the channel remains silent. In this way, [10] This is a centralized algorithm. This algorithm is the
the cooperative node knows that the intended link fails and extended form of Leach. In this scheme Base station holds
replies an ACK to the sender to become the next replay. It the main responsibility of cluster formation and choosing
works to minimize the impact of packet loss. [6] Author has the CH. As the whole process is centrally operated by the
gone through various clustering algorithms like leach, heed. BS so known as centralized process. Each node directly
And created a view why we require clustering. Clustering communicates with the base station thus a better and
enhances the network in various forms like limits energy centralized view of network is created. Nodes reply to the
consumption, improves network life time. In this various BS there residual energy for the process of cluster
factors are discussed which should be kept in mind so to formation. The CH is selected by the BS. Even in this way
design an effective clustering algorithm. [10] Leach-C is a BS has a full control over the CH formation process. By this
centralized algorithm. [4] In this the author proposed an process performance is enhanced as the control lie in the
algorithm which consists of three protocols in order to hands of BS.
provide scalability, energy efficient, multi level clustering, D. Low-Energy Adaptive Clustering Hierarchy K-Mean
fault tolerance and load balancing. Author uses NS2 for the
Author put forward the clustering technique of leach and
implementation. Leach-ked [3], in this author put forward the
then combines it with leach-C. [3] Leach-ked operates in
clustering technique of leach and then combines it with rounds. Each round consists of two stages cluster building
leach-C. This 5% implies that the CH number is fixed. stage and stable stage. CH is selected from the nodes as 5%
DFCA [2] is a distributed fault tolerant clustering algorithm
are the cluster heads by using k-means. In First phase cluster
addresses both issues of cluster formation based on the
geometric centre is calculated and residual energy, initial
remaining energy of the nodes and fault tolerance when the
energy and distance between the nodes is considered. Once
CH dies. EFDCB [1] is extended form of FDCB algorithm.
the CH is chosen process moves to second phase that is
stable phase. CH sends all nodes notification of elected as the
III. VARIOUS FT AND CLUSTERING TECHNIQUES CH and ask them to send data for transmission. Once the
A. Low-Energy Adaptive Clustering Hierarchy nodes join the CH that is present in there range. They
This protocol randomly with equal probability selects cluster transmit data to the CH. Further cluster head forwards data to
head to achieve energy balance.[14] The two main objective the BS by merging the transmitted data into packets. When
of this protocol is to increase network lifetime and second to compared the results its survival time is increased and death
reduce the energy consumption. Energy consumption is rate of nodes is slower then Leach and Leach-C.
reduced by data aggregation process in which E. Distributed Fault Tolerant Clustering Algorithm
communication messages is lowered. It is probably first This algorithm addresses both issues of cluster formation
dynamic clustering protocol, still serving as base for various based on the remaining energy of the nodes and fault
schemes. In this the nodes themselves take the random tolerance when the CH dies. It undergoes two phases steady-
decision of selecting CH without any centralized control. state phase and set-up phase. [2] Firstly the bootstrapping
Each node gets an equal chance to become CH as the rounds process takes place. In which the BS provides unique ids to
keep on going. Leach can provide a better balance of energy the nodes and gateways. BS forms the schedule and divides
by random rotation of CH. It provides scalability and the process into multiple rounds and duration is fixed of each
enhances the network performance as a localized protocol. round. Gateway generates a hello message. The hello
[2] The main drawback of this algorithm can be taken as the message holds the three main parameters like residual
possibility of selecting the node of low energy as CH. It has energy, distance from base station and its id. The sensor
the probability to select every node as CH. The node with nodes that have no CH within their communication range
low energy when selected as CH may die more fastly come under uncovered set and generates help message for
because CH consumes more energy causing a fault in the backup. DFCA also presents a distributed run time recovery
network. of the faulty cluster members due to sudden failure of the

All rights reserved by www.ijsrd.com 374


Fault Tolerance and Clustering Algorithm in WSN-A Review
(IJSRD/Vol. 4/Issue 05/2016/093)

CH. When the CH dies then the instead of disconnecting the thus by this it covers shortcomings of FDCB algorithm. It
whole cluster it moves to the neighbouring CH and transfer works in four phases cluster formation, selecting cluster
its data to neighbour CH. A sensor node can join to only one head, to find faulty node and then provide fault free path for
gateway. data transmission. In this algorithm CH has all the
information regarding its area and the nodes connected.
F. Extended Fully Distributed Cluster-based Routing
When the connection between the CH and the Cluster-
Algorithm
member is lost due to mobility or fault. CH has whole
This technique presents a distributed architecture in which knowledge of that node and without any delay it
each and every cluster member maintains a table holding reestablishes the connection again.
parameters of quality of service. Cluster members, table and
failure are calculated locally at the nearest site. [1] Only IV. COMPARATIVE ANALYSIS OF CLUSTERING AND FT
suffers minor disturbance in the path during recovery and TECHNIQUES
Parameter Leach Leach-C Leach-KED CMATO DFCA EFDCB
Deal with Fault
Better Fault detection
Improve survival permanent tolerance and
Good at To inc life time performance as and fault
time then leach and medium energy
BS forms cluster management.
faults efficient
Fault LBCA and
Leach and Leach, Leach-C detection and Fault
Technique Extended FDCB
Clustering centralized and K-mean recovery done tolerance
used and local repair.
algorithm algorithm by cluster clustering
members. algorithms.
Data send to BS Multiple Takes some time
Randomly select Little slower Load on
so energy gets failure inc. in
Drawbacks low energy node than the neigbouring
consumed in this load on newly reestablishment
as CH algorithms used CH inc.
process CH of path
CH Probability/Rand Weight-based Average energy
Random Cost-based Cost-function
selection omly function of nodes
Cluster
Distributed Centralized Distributed Distributed Distributed Distributed
process
Node
Limited Limited NO NO Limited Yes
mobility
Cluster
1-hop 1-hop 1-hop Multi-hop Multi-hop Multi-hop
topology
Cluster
Variable Variable Fixed variable Variable Variable
count
Table 1: Comparative Analysis of Clustering and Ft Techniques
[2] Azharuddin, Md, Pratyay Kuila, and Prasanta K. Jana.
V. CONCLUSION "A distributed fault-tolerant clustering algorithm for
In this review paper we have gone through the concept of wireless sensor networks." In Advances in Computing,
wireless sensor networks .Various failure that can occur and Communications and Informatics (ICACCI), 2013
can cause damage to WSN network. To make the network International Conference on, pp. 997-1002. IEEE, 2013.
work in case of any fault that means capacity of the network [3] Yunjie, Jia, Liu Ming, Zhu Song, and Dong Pengtao. "A
to tackle fault situation is termed as fault tolerance. To clustering routing algorithm based on energy and
achieve FT, researches have come up with so many distance in WSN." In Computer Distributed Control and
techniques. Some clustering and FT techniques are analyzed Intelligent Environmental Monitoring (CDCIEM), 2012
and some of the good points of techniques are laid down International Conference on, pp. 9-12. IEEE, 2012.
with some drawbacks. Various parameters are laid down [4] Kumar, Rajesh, and Uttam Kumar. "A hierarchal cluster
like CH selection, Cluster Topology. From this we came to framework for wireless sensor network." In Advances
know about various research challenges that need to be in Computing and Communications (ICACC), 2012
focused like in DFCA the load on neighbouring cluster head International Conference on, pp. 46-50. IEEE, 2012.
increases. In future one can come up with a technique to [5] Alkhatib, Ahmad Abed Alhameed, and Gurvinder
reduce this load further. Singh Baicher. "Wireless sensor network architecture."
In International Conference on Computer networks and
Communication Systems (ICNCS 2012), vol. 35, pp.
REFERENCES
11-15. 2012.
[1] Kshirsagar, Ravindra V., and Ashish B. Jirapure. "A [6] Boyinbode, O., Hanh Le, Mbogho, A., Takizawa, M.,
fault tolerant approach to extend network life time of Poliah, R., "A Survey on Clustering Algorithms for
wireless sensor network." In Advances in Computing, Wireless Sensor Networks," in Network-Based
Communications and Informatics (ICACCI), 2015 Information Systems (NBiS), 2010 13th International
International Conference on, pp. 993-998. IEEE, 2015. Conference on , vol., no., pp.358-364, 14-16 Sept. 2010

All rights reserved by www.ijsrd.com 375


Fault Tolerance and Clustering Algorithm in WSN-A Review
(IJSRD/Vol. 4/Issue 05/2016/093)

[7] Fang, Weiwei, Feng Liu, Fangnan Yang, Lei Shu, and
Shojiro Nishio. "Energy-efficient cooperative
communication for data transmission in wireless sensor
networks." IEEE Transactions on Consumer Electronics
56, no. 4 (2010): 2185-2192.
[8] Lai, Yongxuan, and Hong Chen. "Energy-efficient
fault-tolerant mechanism for clustered wireless sensor
networks." In Computer communications and networks,
2007. ICCCN 2007. Proceedings of 16th international
conference on, pp. 272-277. IEEE, 2007.
[9] Gupta, Gaurav, and Mohamed Younis. "Fault-tolerant
clustering of wireless sensor networks." In Wireless
Communications and Networking, 2003. WCNC 2003.
2003 IEEE, vol. 3, pp. 1579-1584. IEEE, 2003.
[10] Muruganathan, Siva D., Daniel CF Ma, Rolly I. Bhasin,
and Abraham O. Fapojuwo. "A centralized energy-
efficient routing protocol for wireless sensor networks."
IEEE Communications Magazine 43, no. 3 (2005): S8-
13.
[11] Mamalis, Basilis, Damianos Gavalas, Charalampos
Konstantopoulos, and Grammati Pantziou. "Clustering
in wireless sensor networks." RFID and Sensor
Networks: Architectures, Protocols, Security and
Integrations, Y. Zhang, LT Yang, J. Chen, eds (2009):
324-353.
[12] Kakamanshadi, Gholamreza, Savita Gupta, and
Sukhwinder Singh. "A survey on fault tolerance
techniques in Wireless Sensor Networks." In Green
Computing and Internet of Things (ICGCIoT), 2015
International Conference on, pp. 168-173. IEEE, 2015.
[13] Raj, A. Samson Arun, Mrs K. Ramalakshmi, and Mrs
C. Priyadharsini. "A survey on classification of fault
tolerance techniques available in wireless sensor
network." In International Journal of Engineering
Research and Technology, vol. 3, no. 1 (January-2014).
ESRSA Publications, 2014.
[14] Heinzelman, Wendi B., Anantha P. Chandrakasan, and
Hari Balakrishnan. "An application-specific protocol
architecture for wireless microsensor networks." IEEE
Transactions on wireless communications 1, no. 4
(2002): 660-670.

All rights reserved by www.ijsrd.com 376