Content uploaded by Abdul Razaque
Author content
All content in this area was uploaded by Abdul Razaque on Oct 10, 2014
Content may be subject to copyright.
ISSN 2320-5407 International Journal of Advanced Research (2014), Volume 2, Issue 9, 721-731
721
Journal homepage: http://www.journalijar.com
INTERNATIONAL JOURNAL
OF ADVANCED RESEARCH
RESEARCH ARTICLE
Robust Sink Failure Avoidance Protocol for Wireless Sensor Networks
1 Abdul Razaque, 2 Khaled Elleithy
University of Bridgeport, CT-06604, USA
Wireless and Mobile communication (WMC Laboratory)
Manuscript Info Abstract
Manuscript History:
Received: 26 July 2014
Final Accepted: 29 August 2014
Published Online: September 2014
General terms:
Algorithms, Design, Performance
and theory.
Keywords:
WSNs, Sink failure avoidance
(SFA), Error detection, Error
recovery, Energy consumption.
*Corresponding Author
Abdul Razaque
Sink failure is one of the critical issues in wireless sensor networks (WSNs).
Prolonging the network lifetime of WSNs, the robust sink has a paramount
significance to provide an interface between deployed sensors successfully.
The sink failure highly affects the performance of several WSN applications.
Hence, the Sink must have the capability to recover from the failure state
immediately.
In this paper, we introduce a sink failure avoidance (SFA) protocol that
improves the network lifetime by using sink fault tolerance algorithms. SFA
helps determine the error detection and error recovery processes successfully.
To demonstrate the effectiveness of algorithms, we use network simulator-2
(ns2.35). Based on the simulation results, it is validated that our algorithms
highly support the error-detection and error-recovery process of the sink
failure successfully. Copy Right, IJAR, 2014. All rights reserved
Introduction
Wireless sensor networks typically involve the technology of small devices that comprise of tiny sensor nodes. Each
sensor node does its job as a unit to monitor the critical data in the environment[1],[2],[3]. WSNs are self-configuring
and self-healing networks comprising of homogenous and heterogeneous nodes. WSNs provide the promising
solutions for numerous applications such as target detection, intrusion detection, industrial automation, airport
surveillance systems, medical diagnosing systems, environmental monitoring, battlefield, etc. However, WSNs
experience the challenge due to sink-failure for several applications.
The sink is of paramount importance to collect the data from different sensor nodes and forwards to the base station.
The sink failure affects the network performance caused by lack of energy, limited storage capacity, computing
power and security in the susceptible environment[4],[5]. As, these problems are not sufficiently addressed to avoid
the sink failure in WSNs. Due to these restrictions, fault-affecting sensor nodes may lead to an unpredictable
condition in which a region of interest is disconnected from the WSN. As a result, the forwarded data cannot be
reached to sink node. A more complex situation in which a sink node fails is inappropriate selection of routing
protocols for a specific application[6].
The sink is data-gathering node from all the nodes in the WSNs and is also responsible for transmitting to the end
user. The sink node is selected to improve the data flows so that energy can be preserved[7]. Thus, generic two-
tiered model was created to improve the network topology to augment the sink and application nodes[8]. The
average bit-stream was used between sink and sensor nodes for prolonging the network lifetime. However, the
energy efficiency was not considered [9]. Probability and missing probability (PMP) model was used to resolve the
sink placement problem[10]. However, sink placement needs to be determined prior to selection of some points [11].
If sink functions correctly that achieves high probability of success for WSN applications.
On the other hand, the failed sink makes WSN as unusable. As a result, the quality of service provisioning is
highly affected. Thus, balancing and improving the performance of WSNs, we introduce a sink failure avoidance
protocol that improves the network lifetime by using sink fault tolerance algorithms that help determine the sink
ISSN 2320-5407 International Journal of Advanced Research (2014), Volume 2, Issue 9, 721-731
722
failure and sink recovery process. This process is initiated by using backup sink (BS) to replace principal sink (PS)
in case of failure. The research of art in this contribution depends on prolonging the network lifetime and improve
the QoS provisioning. The remainder of the paper is organized as follows: The section I, presents the problem
statement. Section II discusses the protocol that provides the sink with fault-tolerance capabilities. Section III gives
the simulation setup and analysis of the result and section IV, finally concludes the paper.
I. PROBLEM STATEMENT
The wireless sensor networks comprise of tiny nodes with limited energy resources. They are scattered in the
different regions of the network to collect the critical information from the physical environment[12]. These tiny
nodes process the collected critical data to sink node in order to forward to end node. The WSNs experience the
problem due to the sink failure. As a result, the sink failure causes the interruption and subsequently may
compromise the network resources. Even if a sink node encompasses of additional resources as compared with other
sensor nodes. However, it experiences the severe problem due to failures caused by the tough deployment
phenomena.
The prerequisite of fault tolerance is inevitable with WSNs, particularly if the application running on WSN is of
high significance. To make WSN functional, fault-tolerance capability fulfills its task despite faults[13]. There are
two approaches are introduced to handle permanent and temporary sink failures, which are check pointing with
active and passive replications. However, these both approaches fail to address the exact cause of sink failure and its
recovery.
II. PROTOCOL PROVIDING SINK WITH FAULT TOLERANCE CAPABILITIES
The goal of this protocol is to make sink be highly robust for handling the error occurrences. In resulting, the
fault-tolerance protocol prolongs the network lifetime by rending the sink as fault-tolerant. The protocol follows
three basic rules, which include fast failure detection, maintaining the reliable sink state, and finally resuming the
sporadic task. We represent the sink (base station) by that creates the interface between WSN and end users. The
sink involves two types: principal sink
and backup sink . The creates the relay for
. Therefore,
functions rather than
and collects the critical information from sensing nodes and forwards to the end nodes
depicted in Figure 1. The
is invoked to distribute the interested queries periodically on the request of users
during each unit waiting time . The initial behavior of
is defined in Algorithm-1.
Algorithm 1: Initial behavior of Principal Sink
1. Initialization of variables (: waiting time for principal sink, : sensed data table, : interested queries,
and : sensor node j)
2.
sending initialization request
3. Repeat until all nodes are initialized
4. Initializing (( and )
5. if {Reception (,
)║ Timeout then
6. Broadcasting ()
7. endif
8. If (Reception (, )) then
9. Add (
10. Compute () //Aggregating and computing the data
11. Send (,
) until (End of )
12. endif
Let us assume that
may fail due to either by unexpected physical interruptions or dearth of energy. Let
be a minimum energy required for
to work properly. Once an error is detected at
, then responsibilities of
are shifted to .
The protocol uses checkpoint process that helps determine the status of the principal node. Once principal node
saves its state and determines the checkpoint (level of energy), then it forwards to . Thus, this process occurs
in the following three steps.
It occurs after computing and data aggregation phase.
It occurs once inquires the check point request.
ISSN 2320-5407 International Journal of Advanced Research (2014), Volume 2, Issue 9, 721-731
723
After expiration of set-timer
When obtains the checkpoint of
, then it stores in the checkpoint table . The behavior of principal
sink and backup sink is elaborated in algorithm 2.
User-1
User-2 User-3
IEEE 802.15.4
IEEE 802.15.14
IEEE 802.15.4
Internet
Ps
Bs
INTERESTED REQUEST ROUTE
RESPONSE REQUEST ROUTE
User-4
IEEE 802.15.4
Figure 1. Fault tolerance robust architecture for a wireless sensor networks
Algorithm 2: New behavior of principal sink
1. Initialization of variables (: waiting time of principal sink, : waiting time of Backup sink to take
responsibility : sensed data table, : interested queries, and : sensor node k)
2. sending initialization request
3. Initializing ((, , )
4. Repeat until all nodes are initialized
5. if {Reception (,
)║ Timeout then
6. Broadcasting ()
7. endif
8. If (Reception (, )) then
9. Adding (,
10. processing () // aggregating and calculating the data
11. endif
12. Sending ( computed result,
)
13. Set ( )
14. Sending (Ḿ, // Measuring of to function instead of
15. Setting (,
16. if {Reception (, )║ Timeout then
17. set ( =Ḱ)
18. Sending (Ḿ, // Measuring of to function instead of
19. Setting (,
ISSN 2320-5407 International Journal of Advanced Research (2014), Volume 2, Issue 9, 721-731
724
20. endif
21. If Receiving message ( ―Hello‖,
) then
22. Responding message (―I am still alive‖)
23. Measuring ( )
24. Sending ( )
25. Until (End of )
26. endif
Furthermore, the protocol applies the sink error detection process that consists of the following steps.
The backup sink node checks each time units periodically and also sends a "hello" messages to
principal node to determine the status. Once does not receive a response after fixed
attempts, and then
is declared as failure sink.
Once backup sink inquires the minimum energy level of principal sink node, and if it finds an energy
level of principal sink node is less than set threshold value of minimum energy. And then
is
declared as failure sink node. This statement can be written as:
If , thus
is failure ‗Ɽ‘.
Once, principal sink node is declared as a failure, then message is broadcasted in the entire network in
order to restrict the nodes to stop sending messages to failure
.
If error is detected in principal sink node, then error recovery process is initiated. As a result, starts to store the
last state of
to avoid the energy consumption in order to improve QoS provisioning explained in
algorithm 3. Furthermore, Table 1 demonstrates the used parameters in the system.
TABLE 1: Showing used parameters with definition in the system model
PARAMETERS
SYSTEM DEFINITIONS
Bs
Backup sink node that replaces the principal sink
node in case of failure of principal sink node
MAC address of principal sink node
Minimum energy
ℓ
Initializing request
Principal sink node
Ɽ
Sink failure
Sensed data table
Threshold value that is set to measure the
minimum energy level of
sink node
Sink (base station)
sensor node j in WSN
Sensor node k in WSN
,, and
Different waiting times for principal node to
detect its failure state
Interested queries
Computed result
checkpoint
Stored checkpoint is used when detecting the
error state of the principal node
ISSN 2320-5407 International Journal of Advanced Research (2014), Volume 2, Issue 9, 721-731
725
checkpoint table is used when principal sink node
gets failure; then backup sink node stores the
current status of principal sink node in this table
Algorithm 3: Replacement of Principal Sink node
with Backup sink node
1. Initialization of variables (: waiting time for principal sink, : waiting time for principal sink
2. Initializing the waiting time and checkpoint table (,, Ḿ )
3. If receiving ( ℓ, PS) then
4. Sending
5. Repeat until message is delivered
6. endif
7. If occurs, then send
8. endif
9. If ( occurs, then send a message ("Hello",
)
10. endif
11. If receiving
, then store ( )
12. Until the end of the request
13. endif
14. If not receiving (―I am alive‖,
)║
║ ; then, error detection process starts
15. Broadcasting the message
16. Storing
17. Installing and Broadcasting (―I am
)
18. endif
III. SIMULATIONS AND RESULTS
We have simulated the performance of sink failure avoidance protocol using ns-2.35-RC7[14] with Ubuntu 13.10
operating system. We have created realistic scenario that reflects the real WSNs phenomenon. The nodes are
randomly placed with uniform fashion in the area of 500 X 500 square meters. The initial energy of each sensor
node is set 8 joules. The bandwidth of the node is 250 kb/sec, and maximum power consumption for each sensor
node is set 13.6 mW. In addition, the sensing and idle modes have 12.4 mW and 0.45 mW respectively. Each sensor
node has the capability to broadcast data from -16 dBm to 11 dBm power intensity.
The total simulation time is set to 35 minutes, and the pause time is set to 5 seconds for the initialization of phase
to warm the nodes at the start of the simulation. The obtained results demonstrate an average of 12 simulation runs.
The energy consumption pertaining to different radio modes and simulation parameters is summed up in Table 2.
Table 2. Summarized simulation parameters
Name of parameters
Description
Transmission Range
30 meters
Sensors
BT node sensors
Sensing Range of node
15 meters
Initial energy of the node
8 Joules
Bandwidth of node
250 Kb/Sec
Number of sensors
350
Number of sinks
2
Network size
500 X 500 m2
Packet transmission rate
25 Packets/Sec
Data Packet size
256 bytes
Simulation time
35 minutes
Initial pause time
5 Seconds
Transmitter energy
13.6 mW
Receiver energy
12 mW,
Power intensity
-16 dBm to 11 dBm
Location of principal sink
(0, 550)
Location of backup sink
(0, 530)
ISSN 2320-5407 International Journal of Advanced Research (2014), Volume 2, Issue 9, 721-731
726
The selected criteria for our protocol involve following metrics.
Initialization time
Energy consumption
Recovery time
A. Initialization Time
We determine the initialization time for sensor node, principal sink node and backup sink node depicted in Figure
2 and 3. In Figure 2, we use our sink failure avoidance protocol to detect the consumed time for initialization of
different number of sensor nodes. We have observed that when network size increases then initialization time
increases that prove the direct trade-off between initialization time and network size. Based on the result, we also
noticed very important point that the consumed time for first 50 nodes is slightly different from last 50 nodes.
050 100 150 200 250 300 350
400 450 500 550 600
NUMBER OF NODES IN NETWORK
Initialization Time (milliseconds)
SFA
0
Figure 2. Initialization time for different number of sensor nodes
050 100 150 200 250 300 350
200 250 300 350 400
NUMBER OF NODES IN NETWORK
Initialization Time (milliseconds)
Ps
0
SN1
Bs
ISSN 2320-5407 International Journal of Advanced Research (2014), Volume 2, Issue 9, 721-731
727
Figure 3. Initialization time for Principal sink node, backup sink node and sensor node at different topologies
The reason is this, the nodes have more energy for initializing the first 50 sensor nodes, but nodes run out their
energy for initializing the last 50 nodes.
However, time is not changed for each simulation runs. In Figure 3, we have calculated the initialization time for
sensor node, principal sink node and backup sink node at the different size of the network. We have observed that
network size increases, then initialization time for sensor node, principal sink node and backup sink node
increases in a different way. The principal sink node takes slightly higher initialization time as compared with
backup sink node and sensor node. The reason of the increase in initialization time of principal sink node is its
additional functionalities.
B. Energy Consumption
The sensor nodes in WSN are powered by limited battery resources. Thus, they require using the limited energy
budget. We have computed consumed energy for different components of sensor node e.g. CUP, LED, EEPROM
and Radio transceiver depicted from Figure 4 to 7. Each component of the sensor node consumes different energy
in network. We have observed that in Figure 4 and 5, principal sink node consumes more energy followed by
backup sink node, sensor node j and sensor node k for CPU and Radio transceiver components.
050 100 150 200 250 300 350
200 300 400 500 600
NUMBER OF NODES IN NETWORK
Consumed Energy for CPU [Millijoule]
Ps
0
SN2
Bs
SN1
Figure 4: Energy consumed for CPU on different sinks
The reason for the consumption of additional energy in case of principal sink node is to transmit additional data and
control packets. Whereas, backup sink node has responsibility to keep on monitoring the principal sink node on the
regular basis for replacement of principal sink node in case of its failure occurs that causes the energy consumption.
ISSN 2320-5407 International Journal of Advanced Research (2014), Volume 2, Issue 9, 721-731
728
050 100 150 200 250 300 350
200 250 300 350 400
Number of Nodes in Network
Consumed Energy for RADIO [Millijoule]
Ps
0
SN2
Bs
SN1
Figure 5: Energy consumed for RADIO on different sinks
050 100 150 200 250 300 350
30 60 90 120 150
Number of Nodes in Network
Consumed Energy for LED [Millijoule]
Ps
0
SN2
Bs
SN1
Figure 6: Energy consumed for LED on different sinks
Figure 6 demonstrates the consumed energy for light emitting diode (LED) for principal sink node, backup sink
node, sensor node j, and sensor node k. In LED, principal sink node consumes less energy as compared with other
nodes. We have observed from Figure 4 to 6 that more energy is consumed for Radio transceiver and CUP. In
Figure 7, consumed energy for electrically erasable programmable read-only memory (EEPROM) is measured only
for backup sink node because principal sink node and other nodes do not require EEPROM. The reason of
consuming the energy for EEPROM is to monitor the principal sink node periodically.
ISSN 2320-5407 International Journal of Advanced Research (2014), Volume 2, Issue 9, 721-731
729
050 100 150 200 250 300 350
30 60 90 120 150
Number of Nodes in Network
Consumed Energy for EEPROM [Millijoule]
0
Bs
CONSUMPTION ONLY FOR Bs
Figure 7: Energy consumed for EEPROM on different sinks
In Figure 8, we show the total energy consumed for all components of principal sink node, backup sink node,
sensor node j, and sensor node k. Based on the outcomes, we have observed that principal sink node consumes more
energy that is around 992.4 millijoules for CPU, but in our case, Radio does not consumes less energy in our case.
050 100 150 200 250 300 350
200 400 600 800 1000
Number of Nodes in Network
Total Consumed Energy for all Components [Millijoule]
Ps
0
SN2
Bs
SN1
Figure 8: Total energy consumed for different components of the sensor
C. Recovery Time
One of the key metrics is to determine the recovery time, when principal sink node got failure and replaced by
backup sink node. Therefore, simulation time and consumed energy are two significant constraints. We observe in
Figure 9, the recovery time that is the trade-off between the failure and recovery occurrences. We have observed that
principal sink node gets failure at 13.2 minutes, then backup sink node replaces the principal sink node for
improving the quality of service provisioning. In the case of replacement, much recovery is reduced. We further
ISSN 2320-5407 International Journal of Advanced Research (2014), Volume 2, Issue 9, 721-731
730
observe that the curve for principal sink node is balanced, and no information is retrieved because of the injection of
failure. The consumed energy for backup sensor node remains stabilized when it starts relaying rather than principal
sink node.
05 10 15 20 25 30 35
200 400 600 800 1000
Simulation time [minutes]
Consumed Energy [Millijoule]
Ps
0
Bs
FAILURE OF PS AT 13.2 MINUTES
Figure 9. Failure of the principal node and consumed energy using the different period
IV. CONCLUSION
In this paper, we have introduced sink failure avoidance protocol for the detection and recovery of the sink errors.
This contribution aims to determine the normal activity and energy consumption level of sink nodes using SFA. The
SFA protocol provides the robust fault tolerance functionality to improve the quality of service provisioning by
replacing failure primary sink node with backup sink node. To demonstrate the validity of SFA protocol, we have
used network simulator-2.
Based on the simulation results, we observed that CPU consumed more energy as compared with other
components of sensor nodes that are very interesting discovery. Furthermore, another significant discovery is to
determine the EEPROM that is only used for backup sink node because it only monitors the failure of principal sink
node. We also validated that SFA also reduces the recovery time and energy consumption.
REFERENCES
[1] A. Razaque and K. M. Elleithy, "Low Duty Cycle, Energy-Efficient and Mobility-Based
Boarder Node—MAC Hybrid Protocol for Wireless Sensor Networks," Journal of Signal
Processing Systems, pp. 1-20, 2014, 10.1007/s11265-014-0947-3.
[2] A. Razaque and K. M. Elleithy, "Energy-Efficient Boarder Node Medium Access Control
Protocol for Wireless Sensor Networks," Sensors, vol. 14, pp. 5074-5117, 2014, doi:
10.3390/s140305074.
[3] A. Razaque and K. Elleithy, "Automatic energy saving (AES) modelto boost ubiquitous
wireless sensor networks (WSNs)," arXiv preprint arXiv:1309.4504, 2013.
[4] L. Benini, G. Castelli, A. Macii, E. Macii, M. Poncino, and R. Scarsi, "A discrete-time
battery model for high-level power estimation," in Proceedings of the conference on
Design, automation and test in Europe, 2000, pp. 35-41.
[5] A. Razaque and K. Elleithy, "Efficient Search (RES) for One-Hop Destination over
Wireless Sensor Network," arXiv preprint arXiv:1310.1129, 2013.
[6] R. Sumathi and M. Srinivas, "A Survey of QoS Based Routing Protocols for Wireless
Sensor Networks," JIPS, vol. 8, pp. 589-602, 2012.
ISSN 2320-5407 International Journal of Advanced Research (2014), Volume 2, Issue 9, 721-731
731
[7] A. Bogdanov, E. Maneva, and S. Riesenfeld, "Power-aware base station positioning for
sensor networks," in INFOCOM 2004. Twenty-third AnnualJoint Conference of the
IEEE Computer and Communications Societies, 2004.
[8] J. Pan, L. Cai, Y. T. Hou, Y. Shi, and S. X. Shen, "Optimal base-station locations in two-
tiered wireless sensor networks," Mobile Computing, IEEE Transactions on, vol. 4, pp.
458-473, 2005.
[9] J. Li and P. Mohapatra, "Analytical modeling and mitigation techniques for the energy
hole problem in sensor networks," Pervasive and Mobile Computing, vol. 3, pp. 233-254,
2007.
[10] W. Wang, V. Srinivasan, K.-C. Chua, and B. Wang, "Energy-efficient coverage for target
detection in wireless sensor networks," in Proceedings of the 6th international conference
on Information processing in sensor networks, 2007, pp. 313-322.
[11] D. Luo, H. Zhao, Y. Bi, K. Lin, X. Zhang, Z. Yin, and P. Sun, "Multiple sink nodes'
deployment based on PMP in WSNs," in Pervasive computing and applications, 2007.
ICPCA 2007. 2nd international conference on, 2007, pp. 675-678.
[12] A. Razaque and M. Elleithy, "Least Distance Smart Neighboring Search (LDSNS) over
Wireless Sensor Network," in Proceedings of the IEEE Conference on European
Modeling Symposium (EMS), Manchester, UK, 2013, pp. 20-22, doi:
10.1109/EMS.2013.91
[13] L. Paradis and Q. Han, "A survey of fault management in wireless sensor networks,"
Journal of Network and Systems Management, vol. 15, pp. 171-190, 2007.
[14] T. Issariyakul and E. Hossain, Introduction to network simulator NS2: Springer, 2011.
BIOGRAPHY
Abdul Razaque is Editor-in-Chief for International Journal for Engineering and
Technology (IJET), Singapore and also associated with Computer Science and
Engineering Department, University of Bridgeport, USA. He holds fellowship form
Higher Education Commission (HEC) Pakistan, and Common Wealth, UK. He served as
Head of computer science department in Model colleges setup Islamabad, Pakistan from
2002 to 2009. He also led the projects as project Director for promoting the trend of
information technology (IT) in Pakistan funded by United Nation organization (UNO)
and World Bank during 2005 to 2008. He is currently active researcher of wireless and
Mobile communication (WMC) laboratory, UB, USA. Abdul Razaque has also been
working as Chair, Strategic Planning Committee for IEEE SAC Region-1. USA and Relational Officer for IEEE
SAC Region-1 for Europe, Africa and Middle-East. Abdul Razaque has chaired more than dozen of highly reputed
international conferences and also delivered his lectures as Keynote Speaker. His research interests include the
wireless sensor networks, design and development of learning environments, TCP/IP protocols, multimedia
applications and ambient intelligence.
Dr. Elleithy is the Associate Vice President of Graduate Studies and Research at the
University of Bridgeport. He is a professor of Computer Science and Engineering. He has
research interests are in the areas of wireless sensor networks, mobile communications,
network security, quantum computing, and formal approaches for design and verification.
He has published more than three hundred research papers in international journals and
conferences in his areas of expertise. Dr. Elleithy has more than 25 years of teaching
experience. His teaching evaluations are distinguished in all the universities he joined. He
supervised hundreds of senior projects, MS theses and Ph.D. dissertations. He supervised
several Ph.D. students. He developed and introduced many new undergraduate/graduate
courses. He also developed new teaching / research laboratories in his area of expertise.
Dr. Elleithy is the editor or co-editor for 12 books by Springer. He is a member of technical program committees
of many international conferences as recognition of his research qualifications. He served as a guest editor for
several International Journals. He was the chairman for the International Conference on Industrial Electronics,
Technology & Automation, IETA 2001, 19-21 December 2001, Cairo – Egypt. Also, he is the General Chair of the
2005-2013 International Joint Conferences on Computer, Information, and Systems Sciences, and Engineering
virtual conferences.
