International Journal of Engineering Trends and Technology (IJETT) – Volume 7 Number 3- Jan 2014
A Reliable Routing Approach For In-Network
Aggregation in Wireless Sensor Networks
Subhashini.S#1
T. K Parani*2
#
Student, M.E Communication Systems engineering, Anna University
DSCE Coimbatore, India
*Assistant professor, dept. Electronics &Communication Engineering
DSCE Coimbatore, India
Abstract— Wireless Sensor Networks are deployed densely in
close proximity to the phenomenon to be monitored accurately
in different applications. As these networks are highly
populated, the data generated will be enormous. Since
communication is the most expensive activity in terms of
energy, Data aggregation technique should be made use of
while transmitting data to the sink to save energy. With data
fusion the size and number of packets transmitted can be
reduced by aggregating the messages from sources in the
intermediate nodes, thus decreasing communication cost, data
traffic and energy consumption. For the network using data
aggregation, the routing of data to the sink is an important and
critical part in the processing, as the loss of aggregated data
packets will have immense impact for data inference at the sink.
Moreover, the routing technique used should be capable of
reliable data transmission in all situations. In this paper, a
novel data routing approach for in-network aggregation called
DRINA (Data Routing for In-Network Aggregation) is
proposed, that has the features such as reduced number of
messages for setting up the routing tree, maximized number of
overlapping routes, high aggregation rate and reliable data
transmission even in the nodes failure situation. DRINA is
compared with two other known methods: Information Fusionbased Role Assignment (InFRA) and Shortest Path Tree (SPT)
algorithms using Network Simulator and is proved to be
outperforming in all the evaluated scenarios. The future work
will incorporate method to reduce the network overhead and to
improve the aggregation rate.
Keywords--- Wireless Sensor Networks, data aggregation,
routing protocol
I. INTRODUCTION
Wireless Sensor Networks consist of large
number of sensors which are deployed in the environment
to monitor various phenomenons like pressure,
temperature, sound, motion, or vibration in different
locations. WSNs are widely used in many applications like
monitoring critical infrastructure, home land security,
environmental monitoring, communication, military and in
many other critical applications[2].
ISSN: 2231-5381
Typically sensor is a battery operated device , thus
are energy constrained. Node in a network performs
basically three operations a) Sensing the event b)
Processing the data and c) data communication. Energy
consumption in a sensor node mainly depends on the
amount of data gathered. So, the communication of data in
the network is the most expensive activity in terms of
energy consumed [7]. For this reason, all the protocols and
techniques used in the WSNs should by considering the
energy consumption in the sensors.
WSNs are data-driven, where large amount of data
generated has to be routed to the sink node, usually in a
multihop fashion. In this scenario, routing technique plays a
vital role in the data transmission process. A possible strategy
to optimize the routing task is to use the available processing
capacity provided by the intermediate sensor nodes along the
routing paths. This is known as data-centric routing or innetwork data aggregation [1]. The resource utilization has to be
minimized for the efficient data gathering in the networks. For
this data aggregation has to be performed on the collected data
in the nodes, which reduces the energy utilization in the nodes.
Data aggregation forwards only smaller number of data,
reducing the redundant data, leads to lower communication
cost, energy saving and thus improves the lifetime of the
network.
The nodes forward the aggregated data packets to the sink
node, where a lose in the packet cause severe impact in data
analysis in the monitoring center. Routing algorithm used plays
an important role in aggregated data forwarding and it should
be capable of providing guaranteed service even in the case of
node failure situations.
For WSN, data aggregation aware routing protocol used
should have some desirable characteristics such as: a
reduced number of messages for setting up a routing tree,
maximized number of overlapping routes, high aggregation
rate, and also a reliable data transmission[1]. Here, a novel
approach for data routing called Data Routing For InNetwork Aggregation (DRINA) is proposed. Our proposed
method can maximize the data aggregation along
communication route in reliable way, through a faulttolerant routing mechanism.
http://www.ijettjournal.org
Page 133
International Journal of Engineering Trends and Technology (IJETT) – Volume 7 Number 3- Jan 2014
II CONVENTIONAL PROTOCOLS
Various protocols are available for the data
aggregation while forwarding the packets. They are
mainly classified as tree-based approach, cluster-based
approach and structure-less approach.
A. Shortest Path Tree (SPT) Algorithm.
It is a tree-based routing approach. It usually
depends on hierarchical organization of the nodes. It
uses a very simple strategy to construct the tree
structure. Each source sends its information to the sink
along the shortest path between the two[7]. Where
these paths overlap for different sources, they are
combined to form the aggregation tree. It has static
routes[1].
B. Greedy Incremental Tree (GIT) Algorithm.
It is also a tree-based approach. It is based on
Direct Diffusion Approach. It establishes an energyefficient path and greedily attaches other sources onto
the established path. The information is routed using
the shortest path in the tree. Whenever a new branch is
created new aggregation point is also selected.
Routing tree cost is more[7].
C. Tiny AGgregation (TAG) Service.
packets for aggregation. It has mechanisms for
increasing the chance of packets meeting at the same
node (spatial aggregation) and at the same time
(temporal aggregation).It does not guarantee
aggregation of all packets, the cost of transmitting
packets with no aggregation increases in larger
networks[7].
III ENERGY EFFICIENT ROUTING: DRINA
The DRINA algorithm is also a cluster-based
approach. In our algorithm, for each new event, it is
performed the clustering of the nodes that detected
the same event as well as the election of the clusterhead. After that, routes are created by selecting nodes
in the shortest path, to the nearest node that is part of
an existing routing infrastructure, where this node
will be an aggregation point. Our DRINA routing
infrastructure tends to maximize the aggregation
points and to use fewer control packets to build the
routing tree. Also, differently from the InFRA
algorithm, DRINA does not flood a message to the
whole network whenever a new event occurs. The
main goal of our proposed the DRINA algorithm is to
build a routing tree with the shortest paths that
connect all source nodes to the sink while
maximizing data aggregation. The proposed
algorithm considers the following roles in the routing
infrastructure creation:
In the TAG algorithm, parents notify their
children about the waiting time for gathering all the
data before transmitting it so the sleeping schedule of
the nodes can be adjusted accordingly. It makes use of
shortest path to route the message[7]. It requires large
number of message exchange to construct and
maintain a tree.
·
D.
Information Fusion-based Role Assignment
(InFRA) Algorithm.
·
It is a cluster-based approach. The algorithm aims
at building the shortest path tree that maximizes
information fusion. Thus, once clusters are formed,
cluster-heads choose the shortest path to the sink node
that also maximizes information fusion by using the
aggregated coordinators distance. For each new event
that arises in the network, the information about the
event must be flooded throughout the network to
inform other nodes about its occurrence and to update
the
aggregated
coordinators-distance[1].
This
increases the communication cost of the algorithm
and, thus, limits its scalability.
E. Data-Aware AnyCast (DAA) Algorithm.
·
·
Collaborator: A node that detects an event and
reports the gathered data to a coordinator node.
Coordinator: A node that also detects an event
and is responsible for gathering all the gathered
data sent by collaborator nodes, aggregating
them and sending the result toward the sink node.
Sink: A node interested in receiving data from a
set of coordinator and collaborator nodes.
Relay: A node that forwards data toward the
sink.
IV MODULE DESCRIPTION
The DRINA algorithm can be divided into three
phases. In Phase 1, the hop tree from the sensor nodes
to the sink node is built. In this phase, the sink node
starts building the hop tree that will be used by
Coordinators for data forwarding purposes. Phase 2
consists of cluster formation and cluster-head election
among the nodes that detected the occurrence of a
new event in the network. Finally, Phase 3 is
responsible for both setting up a new route for the
reliable delivering of packets and updating the hop
tree.
It is a structure-less approach. Uses anycast to
forward packets to one-hop neighbors that have
ISSN: 2231-5381
http://www.ijettjournal.org
Page 134
International Journal of Engineering Trends and Technology (IJETT) – Volume 7 Number 3- Jan 2014
F. Hop Tree Building
Flowchart for this phase is given in Fig 1. In this
phase, the distance from the sink to each node is
computed in hops. This phase is started by the sink
node sending, by means of a flooding, the Hop
Configuration Message (HCM) to all network
nodes. The HCM message contains two fields: ID
and HTT, where ID is node identifier that started or
retransmitted the HCM message and HTT is the
distance, in hops, by which an HCM message has
passed.
HTT in the HCM message is less than the value of
HTT that it has stored and if the value of
FirstSending(FS) is true. If that condition is true then
the node updates the value of the NextHop variable
with the value of the field ID of message HCM, as
well as the value of the HTT variable, and the values
in the fields ID and HTT of the HCM message. The
node also relays the HCM message. Otherwise, if
that condition is false, which means that the node
already received the HCM by a shorter distance, then
the node discards the received HCM message. The
above steps are repeated until the whole network is
covered. The flow diagram is shown below. Before
the first event takes place, there is no established
route and the HTT variable stores the smallest
distance to the sink.
On the first event occurrence, HTT will still be
the smallest distance; however, a new route will be
established. After the first event, the HTT stores the
smaller of two values: the distance to the sink or the
distance to the closest already established route.
G. Cluster Formation and Cluster Head Election
Fig 1 Flowchart for Hop Tree Configuration in the network.
The HTT value is started with value 1 at the sink,
which forwards it to its neighbors (at the beginning,
all nodes set the HTT as infinity). Each node, upon
receiving the message HCM, verifies if the value of
ISSN: 2231-5381
Fig 2 Flowchart for cluster formation and leader election process
http://www.ijettjournal.org
Page 135
International Journal of Engineering Trends and Technology (IJETT) – Volume 7 Number 3- Jan 2014
Fig 2 shows the process of cluster formation and
the leader election from a cluster in the network.
When an event is detected by one or more nodes,
the leader election algorithm starts and sensing
nodes will be running for leadership (group
coordinator); this process is described in this phase.
For this election, all sensing nodes are eligible. If
this is the first event, the leader node will be the
one that is closest to the sink node. Otherwise, the
leader will be the node that is closest to an already
established route. In the case of a tie, i.e., two or
more concurrent nodes have the same distance in
hops to the sink (or to an established route), the
node with the smallest ID maintains eligibility[1].
At the end of the election algorithm only one
node in the group will be declared as the leader
(Coordinator). The remaining nodes that detected
the same event will be the Collaborators. The
Coordinator gathers the information collected by
the Collaborators and sends them to the sink. A key
advantage of this algorithm is that all of the
information gathered by the nodes sensing the
same event will be aggregated at a single node (the
Coordinator), which is more efficient than other
aggregation mechanisms (e.g., opportunistic
aggregation).
H. Route Formation and Hop Tree Update.
Fig 3 shows the process of route update in the
network. The elected group leader, starts establishing
the new route for the event dissemination. This process
is described in this phase. For that, the Coordinator
sends a route establishment message to its NextHop
node. When the NextHop node receives a route
establishment message, it retransmits the message to its
NextHop and starts the hop tree updating process.
These steps are repeated until either the sink is reached
or a node that is part of an already established route is
found. The routes are created by choosing the best
neighbor at each hop.
The process of data transmission is described in this
phase. While the node has data to transmit, it verifies
whether it has more than one descendant that relays its
data. If it is the case, it waits for a period of time and
aggregates all data received and sends the aggregated
data to its NextHop. Otherwise, it forwards the data to
its NextHop. For every packet transmission with
aggregated data, the Route Repair Mechanism is also
executed. A route repair mechanism is used to send
information in a reliable way. Sender nodes wait a
predefined time period to receive a packet delivery
confirmation. When the confirmation is not received by
the sender node, a new destination node is selected and
the message is retransmitted by that node[1].
ISSN: 2231-5381
Fig 3 Flowchart for Route Establishment.
http://www.ijettjournal.org
Page 136
International Journal of Engineering Trends and Technology (IJETT) – Volume 7 Number 3- Jan 2014
evaluated in various scenarios to find the algorithm’s
efficiency and its working capabilities in various situations.It
are simulated by using NS-2 simulator version 2.34. Various
performance parameters are evaluated.
Data Packet Delivery Rate: Number of packets that
reach the sink node. This metric indicates the quality of
the routing tree built by the algorithms—the lower the
packet delivery rate, the greater the aggregation rate of the
built tree.
Control Packet Overhead: Number of control messages
used to build the routing tree including the overhead to
both create the clusters and set up all the routing
parameters for each algorithm.
Efficiency: Packets per processed data. It is the rate
between the total packets transmitted (data and control
packets) and the number of data received by the sink.
Routing Tree Cost: Total number of edges in the routing
tree structure built by the algorithm gives the routing tree
cost required.
Loss of Aggregated Data: Number of aggregated data
packets lost during the routing. In this metric, if a packet
contains X aggregated packets and if this packet is lost, it
is accounted the loss of X packets.
Number of Transmissions: Sum of control overhead and
data transmissions, i.e., the total packets transmitted by
the nodes in the network.
J. Simulation Parameters
Table 1 simulation parameters
Fig 4 Flowchart for Reliable data routing in the network.
I. Route Repair Mechanism
The route used by the algorithm is unique for data
transmission, which maximizes the data aggregation along
with the overlapping routes. Any failure in one of its nodes
will cause disruption, preventing the delivery of several
gathered event data. In the conventional methods, flooding of
the message is used for identifying the failure nodes. In our
proposed method, ACK based repair mechanism is used to
identify the node failure. When a node sends aggregated data
packet to its NextHop, the receiver node should transmit an
ACK to its sender. If the receiver does not receive the ACK, it
has to find an alternate NextHop to forward the packet. Thus a
new route will be established excluding the repaired node.
This ensures the reliable data routing in the network.
VI PERFORMANCE ANALYSIS
To evaluate the performance of DRINA, it is compared
with some other known solutions such as InFRA and SPT.
This two were chosen as they have same goal as that of
DRINA
algorithm in data routing. Performance is
ISSN: 2231-5381
Simulation
Parameter
Simulator
Topology Size
Number of Nodes
Communication
Radius(m)
# of events
Event Duration
(hours)
Loss Probability
(%)
Simulation
Duration (hours)
Notification
Interval (sec)
Value
NS-2(v2.34)
700 m 700 m
500
80 m
3
4
0
4
60
VII RESULTS
Each of the discussed performance metrics of the
DRINA algorithm are evaluated under various situations
in the network for determining its efficiency. The
algorithms performance is compared with the known
http://www.ijettjournal.org
Page 137
International Journal of Engineering Trends and Technology (IJETT) – Volume 7 Number 3- Jan 2014
protocols SPT and InFRA under various criteria to out
show its capability in various scenarios. Following
parameters are varied for the performance evaluation and
comparisons are,
Network size.
Number of Events
Events Duration
Communication Failure
K. Impact Of Network Size
The performances of the assessed algorithms are
evaluated on the basis of number of nodes in the
network, to evaluate the algorithms scalability.
Fig 6 Overhead- Packets Vs Number of Nodes
Fig 5 Data packets –Packets Vs Number of Nodes
DRINA sends less than 75 percent of the data packets
sent by InFRA and about 60 percent of the data packets
sent by SPT. This result clearly indicates that DRINA
maintains the quality of the routing tree even when the
number of nodes increases. The data packets generated
increases with the increase in the number of nodes. The
graph shown in the Fig 5 indicates that DRINA has high
aggregation rate compared to that of the SPT and InFRA
algorithms. The data packets generated by the SPT are
more than the InFRA, which has as medium delivery rate
and aggregation rate. This shows that as the number of
overlapping routes increases the data packets generated
for transmission get decreased as the aggregation rate
increases.
DRINA is more scalable than the InFRA algorithm
since our algorithm needs 30 percent less control messages
to build the routing structure. On the other hand, the DRINA
algorithm requires, on average, 25 percent more control
messages than the SPT algorithm. However, the routing
trees built by SPT results in 30 percent less efficiency than
the trees built by DRINA algorithm. From the Fig 6, it is
clear that the overhead transmitted over the network get
increased with increase in number of nodes. The overhead
for SPT is less when the number of nodes in the network is
varied. The InFRA has maximum overhead in the network.
As, overhead decreases the performance of the system, it is
to be controlled.
From the graph in Fig 7, it is clear that the efficiency
of DRINA increases with the increase in the network size.
The packets per processed data are less for DRINA when
compared with that of SPT and InFRA algorithms. It proves
that DRINA has high aggregation points in the network.
Fig 7 Efficiency –Packets per processed data Vs Number of Nodes.
DRINA is more than 20 and 28 percent efficient than the
InFRA and SPT algorithms, respectively. This occurs
because the DRINA algorithm needs less control
ISSN: 2231-5381
http://www.ijettjournal.org
Page 138
International Journal of Engineering Trends and Technology (IJETT) – Volume 7 Number 3- Jan 2014
messages to build the routing tree when compared to
InFRA.
packets sent by the InFRA and SPT respectively.
This indicates that one of the main advantages of
the DRINA: by varying the number of events,
DRINA builds routing trees more likely to have
higher data aggregation rates.
Fig 8 Tree Cost- Edges Vs Number of Nodes
From the above result it is clear that the number of
edges required to build the routing tree is less when
compared to InFRA and SPT. So, the tree cost required is
also less. Moreover, the routing tree built by DRINA has a
better data aggregation quality than InFRA and SPT.
L.
Impact Of Number of Events
The number of events was varied to evaluate the
behavior of the proposed algorithms in networks with 1 to
6 events occurring simultaneously.
Fig 10 Overhead– Packets Vs Number of Events
The above graph indicates that DRINA needs
less than 50 percent of the control messages
requierd by the InFRA in the occurrence of events.
The overhead of the DRINA is medium when
compared to others.
Fig 11 Efficiency – Packets per processed data Vs Number
of Events
Fig 9 Data Packets – Packets Vs Number of Events
From the above graph it’s clear that DRINA
sends fewer packets than the InFRA and SPT
algorithms. For instance, DRINA sends
approximately 81 and 67 percent of the data
ISSN: 2231-5381
For more than one event, DRINA is more
efficient than SPT and InFRA.
http://www.ijettjournal.org
Page 139
International Journal of Engineering Trends and Technology (IJETT) – Volume 7 Number 3- Jan 2014
sent by InFRA and SPT, respectively. This indicates that
by varying the time of an event duration, DRINA obtains
a data aggregation rate greater than InFRA and SPT.
Fig 12 Tree Cost– Edges Vs Number of Events
The cost of the routing tree built by DRINA is less
than 10 percent smaller than in the InFRA algorithm, and
30 percent smaller than in the SPT. The number of relay
nodes required to transmit the data from source to sink
node is less for DRINA when compared to that of the SPT
and InFRA algorithms. Moreover, it generates better
routing tree for the data transmission, which reduces the
tree cost of the network.
J.
Impact Of the Event Duration
The event duration is varied from 1 to 5 hours to
evaluate the performance. The Fig 13 shows the data
packets transmitted with increase in the event duration for
the assessed algorithms. When the event duration is
increased, the data packets transmitted by the DRINA is
less when compared to SPT and InFRA. Thus, it indicates,
the aggregation rate of the DRINA is better than the other
algorithms.
Fig 14 Overhead– Packets Vs Event Duration
The graph in Fig 14 shows that DRINA requires less
control messages to create the routing structure than
InFRA but it requires more control messages than the SPT
algorithm. Although DRINA requires 35 percent more
control messages than SPT, SPT does not build a good
data aggregation routing tree, as shown in previous results.
It is clear that the overhead in the DRINA is medium
when compared to the other two algorithms, which is to
be controlled.
Fig 15 Efficiency – Packets per processed data Vs Event Duration
Fig 13 Data Packets – Packets Vs Event Duration
DRINA algorithm sends less data packets than the
other evaluated algorithms. More specifically, DRINA
sends more than 84 and 64 percent of the data packets
ISSN: 2231-5381
The above graph shows that DRINA is more efficient
than InFRA and SPT. This algorithm outperforms the
other evaluated algorithms even in scenarios of short-term
events while InFRA exceeds the SPT only in scenarios
where the event duration islonger (typically more than 2
hours).
http://www.ijettjournal.org
Page 140
International Journal of Engineering Trends and Technology (IJETT) – Volume 7 Number 3- Jan 2014
aggregated data are retransmitted. On the other hand, SPT
and InFRA protocols send less data packets when the
communication failures probability increases. This
happens because when a packet is lost due to
communication failures the packets are not retransmitted
and do not reach the sink. The Fig 18 shows the delivery
rate of packets for the assessed algorithms with varying
loss probability in the network. The graph shows that the
delivery rate of the DRINA is high compared to that of
SPT and InFRA algorithms as it make use of repair
mechanism in the network. The delivery rate of SPT of
very less compared to others as there is no efficient route
repair technique used in the algorithm.
Fig 16 Tree Cost– Edges Vs Event Duration
K.
Impact Of Communication Failures
Communication failure is considered for evaluating
the algorithms reliability. For this, the communication
failure probability parameter is varied from 0 to 20
percent. The cost of DRINA path repair mechanism is
also evaluated. The DRINA make use of route repair
mechanism for the reliable data transmission in the
network. The node failure in the network is identified by
sending the ACK packet to the node in the transmission
route. If found to be failure, then alternate route is used
for transmission to sink. Thus with the failure node,
reliable data transmission can be achieved in the network.
The Fig 17 shows that the data packets generated is better
for DRINA than the SPT and InFRA algorithms.
Fig 18 Delivery rate of packets – delivery rate Vs Loss Probability
With 20 percent communication failure, the delivery
rate of InFRA is less than 30 percent, while DRINA
delivers all aggregated data that have been sent. Therefore,
the delivery rate of the aggregated data packets in DRINA
is high when compared to other two algorithms.
Fig 17 Data Packets- Packets Vs Loss probability
In the DRINA algorithm, data packet transmission
increases when the probability of communication failure
increases. This is due to the fact that lost packets with
ISSN: 2231-5381
Fig 19 Efficiency – Packets per processed data Vs Loss Probability
http://www.ijettjournal.org
Page 141
International Journal of Engineering Trends and Technology (IJETT) – Volume 7 Number 3- Jan 2014
From the above graph it’s clear that even in the loss
probability scenario, the efficiency of the DRINA
outperforms the other two algorithms. In summary,
DRINA delivers aggregated data reliably with the best
performance when compared to SPT and InFRA.
VIII CONCLUSIONS
Data Routing for In-Network Aggregation (DRINA) is
efficient aggregation aware routing algorithms which play an
efficient role in the event based WSNs. The routing
mechanisms and the route repair mechanism followed by
this algorithm supports reliable aggregated data transmission
among the nodes. This algorithm is compared with other two
known solutions namely SPT and InFRA extensively in
various scenarios like varying the number of nodes, number
of events, events duration, and loss probability for the
communication cost, delivery efficiency, aggregation rate
and data delivery rate. With the help of route repair
mechanism and by increasing the number of aggregation
points the DRINA outperforms in all the evaluated scenarios.
It also has some key aspects required by WSNs aggregation
aware routing algorithms such as a reduced number of
messages for setting up a routing tree, maximized number of
overlapping routes, high aggregation rate, and reliable data
aggregation and transmission. Thus it is proved that the
implementation of the DRINA in the networks can improve
the lifetime of the network by reducing the energy
consumption by transmitting relatively less number of
packets to the sink.
ACKNOWLEDGMENT
I,SUBHASHINI S, student of M.E COMMUNICATION
SYSTEMS, Dept. of ECE, Dhanalakshmi Srinivasan
College of Engineering, Coimbatore. I would like to thank
Asst. Prof. Ms.T.K..PARANI for her constant encouragement
and support throughout the completion of the paper. I deeply
express my gratitude to all the ECE department staffs for their
valuable advice and co-operation.
[4] Anastasi G, Conti M, Francesco M, and
Passarella A , “Energy Conservation in
Wireless Sensor Networks: A Survey,” Ad
Hoc
Networks, vol.7,no. 3, pp. 537- 568,2009.
http://dx.doi.org/10.1016/j.adhoc.2008.06.003.
[5]
Chandrakasan A.P, Smith A.C, and
Heinzelman
W.B , “An Application-Specific Protocol
Architecture for Wireless Microsensor
Networks,”IEEE Trans. Wireless Comm., vol.
1,
no. 4, pp. 660-670,2002.
[6] Krishnamachari B, Estrin D, and Wicker S.B,
“The Impact of Data Aggregation in Wireless
Sensor Networks,” Proc. 22nd Int’l Conf.
Distributed Computing Systems (ICDCSW ’02),
pp. 575-578,2005
[7] Nakamura E.F, Loureiro A.A.F, and Frery A.C
“Information Fusion for Wireless Sensor
Networks: Methods, Models, and
Classifications,” ACM Computing Surveys,
vol.
39, no. 3, pp. 9-1/9-55,2007
[8] Romer K and Mattern F, “The Design Space of
Wireless Sensor Networks,”IEEE Wireless
Comm., vol. 11, no. 6, pp. 54-61,2004.
[9] Villas L.A, Boukerche A, Araujo R.B, and
Loureiro A.A (2009), “A Reliable and ACM
Int’l Conf. Modeling, Analysis and Simulation
of Wireless and Mobile Systems (MSWiM), pp.
245-252,
[10] Younis O, Krunz M, and Ramasubramanina S,
“Node Clustering in Wireless Sensor
Networks:
Recent Developments and Deployment
Challenges,” IEEE Network, vol. 20, no. 3,
pp. 20- 25,2006.
REFERENCES
[1]
Villas L.A, Boukerche A, Ramos H.S, De and
Loureiro A.A.F (2013)“DRINA:A Lightweight
Aggregation in Wireless Sensor Networks,”
IEEETrans.,on computers, vol.62 No.4, pp
676-689.,2013
[2] Al-Karaki J, Ul-Mustafa R, and Kamal A ,
“Data Aggregation in Wireless Sensor
Networks —Exact and Approximate
Algorithms,”Proc. High Performance
Switching and Routing Workshop
(HPSR ’04),pp. 241-245,2004.
[3] Akyildiz I.F, Su W, Sankarasubramaniam Y,
and Cyirci E, “Wireless Sensor Networks:
A Survey,” Computer Networks, vol.38,
no. 4, pp. 393-422,2002.
ISSN: 2231-5381
http://www.ijettjournal.org
Page 142