In network aggregation using efficient routing techniques for event driven sensor network

International Journal of Computer Networks & Communications (IJCNC) Vol.6, No.1, January 2014

IN-NETWORK AGGREGATION USING EFFICIENT
ROUTING TECHNIQUES FOR EVENT DRIVEN
SENSOR NETWORK
Smitha N. Pai1, K.C. Shet2and Mruthyunjaya H.S. 3
1

Dept. of CSE, M.I.T., Suratkal
Comp. Engg.,NITK, Suratkal
3
Dept. Of E and C., M.I.T., Manipal University, Manipal
2

ABSTRACT
Sensors used in applications such as agriculture, weather , etc., monitoring physical parameters like soil
moisture, temperature, humidity, will have to sustain their battery power for long intervals of time. In
order to accomplish this, parameter which assists in reducing the consumption of power from battery need
to be attended to. One of the factors affecting the consumption of energy is transmit and receive power.
This energy consumption can be reduced by avoiding unnecessary transmission and reception. Efficient
routing techniques and incorporating aggregation whenever possible can save considerable amount of
energy. Aggregation reduces repeated transmission of relative values and also reduces lot of computation
at the base station. In this paper, the benefits of aggregation over direct transmission in saving the amount
of energy consumed is discussed. Routing techniques which assist aggregation are incorporated. Aspects
like transmission of average value of sensed data around an area of the network, minimum value in the
whole of the network, triggering of event when there is low battery are assimilated.

KEYWORDS
In-network aggregation, agriculture, sensor network, routing, event handling

1. INTRODUCTION
Sensor devices are used to measure physical parameters like pressure, temperature, humidity etc.
When placed within the transmission range of each other, it forms a sensor network. It carries the
task of sensing, computation and forwarding. They have some limitations like computation,
memory and energy. Sensors deployed in applications like the agricultural field require that the
batteries be operating for one cropping season. Energy in the battery can be saved by reducing the
number of transmissions and receptions. Reduction of the packet size, or distance between the
nodes can also help in saving sufficient amount of energy. Efficient routing algorithms will have
to be incorporated to find paths which consume minimal energy during path establishment and
data transfer [1, 2].
The current paper is based on an on-going project COMMON_Sensewhere the water level is
monitored using sensor network [3]. Sensors obtain energy to operate using solar energy, power
grid lines or the battery. In applications like agriculture, the fields are away from the main land
and so power grid lines are difficult to obtain. Solar panels when placed in the agricultural field
will not only block the panels because of the large leaves, also there are chances of theft. Hence
battery cells are the next available options.

DOI : 10.5121/ijcnc.2014.6110

147


Sensors deployed in the agricultural field are less in number and they are placed relatively far
from each other. This reduces the cost factor on the farmers. They are basically used to monitor
the water level in the field. Some sensors are also used to measure humidity. Information
regarding the water level in the field if provided to the farmer could help them in regulating the
flow of water. This avoids salinization of soil when the water is in excess. It can also provide only
sufficient amount of water and hence improve on the yield of the crop. Over irrigating leads to
disturbing the hydrology of the water bed which is undesirable.

2. RELATED WORK
Sensors deployed in the agricultural field read the soil moisture and report this data information to
a common access point called the base station or the destination node. All the nodes in the
network form the source node. One node among them which is approximately placed at the centre
of the network is the base station. A path has to be established between all these nodes and the
base station. In order to achieve this routing algorithms are essential. Nodes use these paths to
send data continuously unless the path fails either due to atmospheric reasons, battery, electronic
failure etc. Energy is expended in finding new path.
Usage of sensors in precision agriculture is common[4-7]. The COMMON-Sense project uses
AODV protocol to establish the path between the source nodes up to the destination. In the
current work better routing techniques are added so that path once established is active for long
intervals of time. This avoids repeated path search and assist in saving energy.
In the field, few sensors are used and they are placed far away from each other. This can avoid
expending more money on sensors. As the distance to the base station is too far off, the data has
to be transmitted using the multi-hop technique.To avoid multiple transmissions of same data the
whole network is partitioned into clusters, and data within a cluster is aggregated and sent to its
cluster head.
Lot of work pertaining to routing, clustering and aggregation is carried out. A protocol based on
the residual energy and minimum transmission cost (minimum hop count) along a path is
projected [8]. Using minimum spanning tree a routing of the path is designed [9]. A genetic
algorithm based routing protocol where a part of the route is swapped and new routes are
established is addressed [10]. Anunequal clustering approach along with minimum hop inters
cluster communication to achieve minimum total energy consumption in the network is suggested
[11]. The network is partitioned into rings along with creating sectors within each ring. This
sector is changed dynamically to form cluster with different set of nodes and send data using
minimum hop [12].In the case of wheel based event triggered data aggregation, aggregation is
done along the spokes of the wheel [13]. Aggregation along the levelled ring structure is carried
out. Ring structure with aggregation along a ring is sent to the next layer of the ring is suggested
[14]. A fine grained aggregation where wedges are created in the network and data is aggregated
from the outer hop to the inner hop aggregating along the way to the base station [15]. Choice of
route should be such that it takes lesstime to reach the base station along with aggregation taking
place at the first level of the tree is projected. [16]. The aggregation and routing techniques
studied help in designing a path to the base station, such that it lives for long, once established.

3. DEPLOYMENT
Broadcasting of control signals unnecessarily to large number of neighbours during path search
can be avoided by choosing efficient topology for deployment. This is achieved by placing sensor
such that it is heard either by three or four of its neighbours during transmission. This is the
148


minimum number of nodes that can be placed close by so that the transmission is continued with
at least three or four paths available all the time. This also ensures minimum usage of sensors, if
the cost of the sensor network is expensive. These sensors form the coordinator nodes to send
data to the neighbours using multi hop. Ref. [17, 18] relates to placement of sensor in grid
topology. A comparison of triangular, square and hexagonal topology is carried out [17]. The
triangular topology though a reliable network with longer life, requires large number of nodes.
Square and hexagonal topologies require almost same number of nodes, with hexagonal requiring
a little less number of nodes at the expense of reliability. Both have almost the same life span.
Triangular topology sends data to the neighbouring six nodes, expending more energy than grid
with four neighbours and hexagonal with three neighbouring nodes. In the current work square
and hexagonal topology is used. Figure 1 shows the various topologies.

(a)Triangular

(b) Square

(c) Hexagonal

Figure. 1 Various topologies

4. SIMULATION PARAMETERS FOR AGRICULTURE
Parameters are designed using Tiny node. This mote is used in COMMON-Sense project. Table
1 shows the data sheet for Tiny node. The values marked in the table refers to the parameter
values used in this study. Frequency of operation is 868Mhz, receiver sensitivity of -104dBm with
transmit power of 5dBm, with antenna height of 1m and minimum transmission range of
200m.Current consumption during transmit, sleep and receive include 33mA, 1µA and 14mA.
These values are again used for the computation as showed in the table 2.
Table 1 Tinynode data sheet [20]

149


5. ROUTING
Nodes are randomly deployed in the rectangular field. Efficient routing techniques are essential
to enhance the life time of the sensor network. Some nodes are identified as coordinator nodes
which approximately lie in the centre of a square or hexagonal region and act as aggregator node.
These coordinator nodes are as planted as per the Figure 1. Packets are transferred between
coordinator nodes using multihop. Other nodes in the network are associated with the coordinator
nodes. The process of routing involves sending broadcast message during path search and
establishing a path and sending data along the established path.During the path discovery from
the source node to the sink node, it finds a path with maximum amount of residual energy and
minimum hop count.
Path with maximum amount of energy is computed by sending the broadcast message to it
neighbors. The neighbors send their position and energy information to its source node. The node
then establishes the path with the next nieghboring node having highest amount of energy. This
process repeats until it reaches the base station. Figure 2a shows the topology with energy
information in it. Node 7 is the source node and node 5 the destination. Figure 2b displays the
process of broadcasting from node 7. In Figure 2c the path is established upto the node 3 and
broadcast message from node 3 to its neighbors. This process is repeated till it reaches the
basestation. Figure 2f illustrates the path established up to the base station such that it is along
the path with neighbors having high energy. The same is presented for a topology of 25 nodes,
with node 0 as the source and node 12 as the sink in Figure 2g. Using the concept explained
above in this diagram, the path that is taken to reach the sink, is longer than what is expected.

(a)

(b)

(e)

(c)

(f)

(d)

(g)

Figure 2 Path establishment with nodes having maximum energy along its neighbors

Alternate strategy is to use the directional information for path finding. Figure 3shows node 0 is
the source and node 12 the sink. [23] address the routing based on directional information. With
vertical distance of X from the sink and Y the horizontal distance as depicted in Figure 3.1a the
minimum distance to the base station from the source is √(X2+Y2)
The path is correctly followed if from any node,
150


(Distance traversed so far from the source node + minimum distance from current traversed node
to destination) ≤ (horizontal distance (X) + vertical distance (Y) from the source node to the
destination).
(1)
At source node 0, using eq. (1) we have,
(0 + √(X2+Y2)) ≤ (X+Y)
(2)
If a path of length Y1 is traversed along vertical direction and X1 along horizontal direction then
the distance from the source to the destination is, X1+Y1+ minimum distance from current node to
destination. Using eq. (1) we should have,
X1+Y1+ √ ((X-X1)2 +(Y-Y1)2) ≤ (X+Y) for a valid path

(3)

In Figure 3.b, X1=0 and the criteria for valid path is satisfied with,
Y1+ √ (X2 +(Y-Y1)2) ≤ (X+Y)
In Figure 3.c, with X1=0 and the criteria for valid path is not satisfied as,
Y1 + √ (X2+Y2 2)) > (X+Y)

(4)

Path shown in Figure 3c is longer as it does not satisfy the eq. (1) and those paths are freed during
path discovery.
Figure 3.d shows that a path of length Y1+X11 is traversed at node 11. On broadcast from node 11
it reaches node 16. At node 16, from the control packet the knowledge of the distance Y1+X11 so
far traversed is obtained. Knowingthe node 11’s x and y coordinates; the distance X 12 between 11
and 16 is computed. X1= X11+X12. The total distance traversed from the source 5 to the current
node becomes X1+Y1. The distance between node 16 and node 12 is computed knowing the
coordinates of the current node and that of base station.
The path become invalid as (X1+Y1+ √(X122 + (Y- Y1)2))> (X + Y).
Paths with highest amount of residual energy along its route could result in a long lasting path.
Given the energy levels ei1,ei2,ei3…ein in each of the 1..n nodes along the path i, where 1st node
represents the source and nth node the destination node
n

the amount of total residual energy along path i is given by Ei=

 eik

(5)

k 1

If another path has energy levels ej1,ej2,ej3…ejn in each of the nodes 1..n with 1 the source node
and n the destination along the path j, the amount of residual energy along path j is given by E j=
n

 ejk

(6)

k 1

Path is so chosen such that, for all the paths p from1...m satisfying the minimum hop count, using
eq. (5) and eq. (6), the path with maximum residual energy is given by,
Max. {Ep}  path p from 1…m.
(7)
where Ep represents total residual energy along path p.
Figure 4a illustrates the topology with source node 7 and destination node 5, with residual energy
shown against each node. Path shown in Figure 4b is with neighbour having maximum energy.
Figure 4c is path with maximum total residual energy. (1+5+1+3+1=11J). The best path is one
with higher minimum energy along its path as shown in Figure 4d.
151


The minimum energy node is the first one to lose its energy. Hence the higher the amount of
minimum energy along the path, the longer the period the data gets transmitted. Hence a third
approach is used which has maximum amount of minimal energy along its path.

(a)

(b)

(c )

(d)

Figure3: Various Partial Paths Traversed (a) Optimal Path (b) Path with Minimum Hop Count (c) Path with
Higher Hop Count (d) Path with a Different Source and Higher Hop Count

(a)

(b)

(c)

(d)

Fig. 4 Routing methods

A path containing the maximum amount of minimum value of energy in the node is chosen as the
best path. Longer life of a path is based on the first node losing its energy which is the node with
minimum energy. Hence chose a path with higher value of minimum energy.
Given the residual energy levels of ei1,ei2,ei3…ein in each of the 1..n nodes along the path i, where
1st node represents the source and nth node the destination node, the minimum residual energy
level along the path i is given by
Emini=min{ ei1,ei2,ei3…ein }
(8)
152


If another path has energy levels ej1,ej2,ej3…ejn in each of the nodes 1..n with 1 the source node
and n the destination along the path j, the minimum residual energy level along the path j is given
by
Eminj=min{ ej1,ej2,ej3…ejn }
(9)
The best path is the one with min hop counts and highest value among the minimum energy
value. If m path exists then the path chosen is such that it satisfies max{ Emin1, Emin2,… Eminm }
(10)
If there exists many paths with same maximal minimum energy, then path with higher residual
total energy along the path is considered. Figure 5 shows the three paths to send the data from
node 0 to node 24. The path along 0510152021222324 is the maximum total
residual energy path with value (4+1+1+1+8+1+1+3+2=22J). The path along
0127813181924 and 0127813182324 are paths with
maximum- minimal residual energy among the various path from source to destination. Both the
path has minimum residual energy node of value 2J, but the path passing through the node 23,
has higher total residual energy among them and this path is chosen. Path containing less than the
minimum threshold value to sustain at least one data packet dispatch is dropped.
An algorithm is proposed to carry out maximum amount of minimum energy. Every node has a
routing table as shown in the Table 2. This routing table helps in establishing a path from any
source node to the destination. To achieve this control signals of the type in Figure 6 are
broadcast from the source node to its neighbors. The working of the basic protocol is explained in
my earlier paper [21]. The packet Type PATH_SEARCH. Each node in the routing table
maintains the maximum value of minimum energy of any path that has traversed through that
node.

Figure 5 Paths traced by various protocols

Figure 6. Path discovery with Minimum Energy Information
153


Figure 7 Packet format used for Path Establishment
Table 2: Routing Table

Information stored
Destination Address
Next Hop Node
Address
Previous Hop Node
Address
Time Stamp
Sequence Number
Hop Count
Route_establish_
flag
Address of
Neighbours*
Source Address
Minimum Energy

Size
occupied in
bytes
4

Functionality
For path discovery the forward route destination address

4

Next neighbour address towards destination

4

Previous neighbour address towards source

4
1
1

Time when the signal was received
The number of times path search was carried out
Hop count from the source
Set to zero during path discovery and set to one on path
establishment.
Address of the linked list containing the information of its
neighbours
Address of the node for which path discovery is initiated
Minimum Energy along the path (not including the current
node energy level.)

1
4
4
4

5.1. Routing Algorithm for maximum amount of minimum residual energy
a) Start the path search using the control signal with a new sequence number from the
source to the destination using the control signal of the type in Figure 6.
b) In each of the traversed node
i) Drop the control packet when it encounters a node which has energy less than the
path threshold energy level (Energy to transmit control signal and at least one
data signal).
ii) Make an entry in the routing table for each source destination combination for
each of the node traversed provided,
i. The route to be searched is a new one with a new sequence number
ii. With minimum hop count from those found so far between the
corresponding source destination combination
iii. Distance traversed so far from the source node + minimum distance from
current traversed node to destination < horizontal distance+ vertical
distance from the source node to the destination
iv. Having maximum minimal energy from a path traversed through the
same node.
v. Having maximum total energy.
vi. For each entry made in the routing table a reverse routing table entry is
made from current traversed node to its previous source node.
vii. Otherwise drop the control packet.
154


c) If the hop count is more than the maximal hop count that can be supported in this
network, drop the packet as this gives an indication that there is no route from this node
to the destination.
d) This process is repeated until it reaches the destination to drop the control packet.
e) Once the packet reaches the destination another control packet of the type in Figure 7 is
started in the reverse direction from the destination to the source.
f) The node traverses the path as specified in the reverse routing table to reach the
destination.
g) Once the control signal reaches the source, data transmission starts from the source to the
destination in the path as established during the reverse path establish.
Inference of algorithm 5.1:Path having maximum amount of energy along its path can sustain
for long intervals of time and avoids repeated transmission and reception during path search. In
this algorithm, the path containing maximum amount of minimal residual energy along with
maximum total residual energy and directinal information is considered for establishing the
path.This path once established will run for longer intervals of time.

6. AGGREGATION
Certain area of the agricultural land may not be even. This could result in deploying more sensors
around those area. In such cases, it may have related value with the neighboring sensors. If all the
associated data is sent to the base station at the same time large amount of energy could be
consumed because of the repeated transmission of correlated data. This results in draining of the
energy in the battery. To avoid this, in-network aggregation is carried out. Various aggregation
techniques are carried out. Data aggregation along the spokes of the wheel is proposed[24].

(a) Topology

(d) Establish path

(b) Closer view of node 21
Sending request

(e) Sensing nodes sending data
to the coordinator node

(c) Receive response

(f) Partitioning of network
into smaller area

Figure8. Various stages to establish a path between the sensing and its coordinator node.
155


In the current study, aggregation is carried out to find a single value like the minimum or the
maximum in the whole network. It is also used to find an average value in a particular area of the
network.
Figure 8 depicts the various stages for searching the coordinator node. Nodes circled with green
colour represent the coordinator nodes. These are placed as per the deployment strategies of
Figure 1. Other nodes are non-coordinator nodes which are sensing and sending their values to
the coordinator nodes. The coordinator node can sense, receive, compute and forward the data to
the next coordinator neighboring node.
Figure8a, picturises the deployment of the sensor nodes. Figure 8b, the close up of a small area
with one particular node with id 21 sending the sensing request information to the nearby
coordinating nodes. Node 0 is the basestation. Node 18 and node 1 are along the path established
to the base station.
The packet details are as per Figure 9a. Figure 8c shows the sensing reply packet received from
the four neighboring coordinator nodes to the node id 25. Reply packet format is as shown in
Figure 9b. Node id 25 finds the coordinator node based on the least distance towards the
coordinator node along with the maximum residual energy among the coordinator nodes and
sends back the message to one of the coordinating node as shown in Figure8d. The non
coordinator node’s information is updated in the coordinator node. This process is carried out
along the whole network. Figure 8e shows the non-coordinator nodes sensing data and
aggregating its value with its coordinator nodes along the path 1810. Figure8f shows the
partitioning of the network into sub areas with one coordinator node in each.
In the packet format of Figure 9, Packet type, is CORD_SEARCH packet type. Sequence no is
incremented, each time it is searching for a new coordinator. Source Address is the address of the
node trying to search for coordinator. In Figure10 the source address corresponds to the
coordinator node, with Coordinate Node x and y position specifiying the position of the
coordinator node. This information is sent to the non-corrodinaotr node from the coordinator
node. This packet size is smaller than that used to search for the next hop neighbor as in AODV
reducing the amount of energy consumed.

Figure 9 Packet Format to Search for the Coordinator Node

Figure 10 Packet format to establish the path.

Once the path to the coordinator is found, the next step is to accept data and send it to the base
station.
156


If all the nodes in the network send their data to the base station, there is likelihood of large
amount of energy consumption. In order to avoid this, two types of aggregation are discussed. In
the first type, the minimum data (soil moisture content) information in the whole of network is
sent to the base station as shown in Figure 11a and explained with the algorithm in section 6.1. In
the second case, Figure 11b the aggregated averaged data value of a particular area is sent to the
base station and explained with an algorithm in section 6.2

(a) Sending the minimum data of the whole network

(b) Average value around around a coordinator node

Figure 11 Routing for sending data values to the base station

6.1. Minimum data value in the whole network
To find minimum data in the whole network, consider the Figure 12 with coordinator node C1,
with non coordinator nodes as S11, S12, S13, with data values d11, d12, d13. The minimum data
among these, dmin1 = min{ d11, d12, d13} is updated in the routing table for the node C1 provided,
the data is the current data. Information pertaining to minimum data value node id and the
location information is updated in the routing table. Node C1 on sensing data d1 is compared with
dmin1. On updating the routing table with dmin12 = min { dmin1,d1} in node C1, send the data
along with the position information of minimum data value node to the next hop node C2 . Node
C2 updates its routing table to the value dmin2=min{ d21, d22}. Min{ dmin2, dmin12} is updated in
the routing table. Node C2 on sensing data d2 updates its routing table with the value which is the
minimum of all data at that node, i.e. min{ min{ dmin2, dmin12}, d2} and forwards this data to the
next neighboring node. This process is repeated until the messages reaches the base station,
sending along with it, the information of the node id with minimum data value, its time of sensing
the data and the position of the node.

Figure 12 Data aggregation at the coordinator node which is sent to base stations aggregating along its path.

Algorithm 6.1
a) Non coordinator node routing table is updated with the values obtained from the
application layer. These values include data from the sensing node, position of the node,
the node id, time when data is sensed and energy associated with the node.
157


b) The non–coordinator node data on sending data to the coordinator node, the
neighboring list information is updated in the coordinator nodes, provided it is the recent
data and this information packet is dropped.
c) Get the information of all the neighboring non –coordinator nodes and compare it with
the coordinating node and update in the coordinator nodes routing table with the value of
least data among all these.
d) When data is forwared to the next coordinator from the previous coordinator node, update
the previous coordinator nodes information in the current node.
e) Get the node with the minimum data information among its neighbors, (if neighbor
exists) with current information, assign it to the current nodes routing table.
f) If the current node, senses new data and if this data is less than the routing data, update
the routing data with the current sensed data.
g) If the data sensed is larger than the routing data, with routing table having current data
information, update the sensed data information with the routing table data information.
h) The data packet obtained from the previous node is dropped.
i) The process of comparing with the neighboring nodes and forwarding data is carried out
until it reaches the base station.
j) At the base station the aggregated information of the whole network i.e., the minimum
data value with its node id, position, energy and the time when the data was sensed along
with the aggregation type (0) is available.
Inference of algorithm 6.1: Aggregation results in sending minimal information to the base
station. This is due to the fact that large amount of data is partially computed in the coordinator
nodes. The minimum data in the whole network helps the operator of the network to regulate the
water supply to that particular area of the network.

6.2 Aggregated value at any specified coordinator node (or specific area of the field)
To find average data value in a particular area at the location of the the coordinator node C 1 in
the network , the average value davg1=avg{ d11, d12, d13, d1} is computed and sent to the next hop
node. This data is forwarded without updating the routing table information along the route
towards the base station. This procedure is depicted in the Figure 13.

Figure 13 Computation of average value of the data at the location where node C1 exists.

Algorithm 6.2
a) Non coordinator nodes obtain the sensed data informatin and updates those values in their
routing table. These values include data from the sensing node, position of the node, the
node id, time when data is sensed and energy associated with the node.

158


b) When the data reaches the coordinator node, recent data obtained from the noncoordinator node is updated in the neighboring node information list and the packets are
dropped.
c) The coordinator node averages the data from its entire non-coordinating node along with
its information and sends it to the next hop neighbor.
d) The process of forwarding is done till it reaches the base station where the information of
the coordinator node, with the averaged data, the time of aggregation, position of the
coordinator node and the aggregation type (1) is obtained.
Inference of algorithm 6.2: This algorithm gives the average soil moisuture content at certain
area of the network. This helps in analysing the water supply in different parts of the field. Water
distribution as used with sprinkler or any other method like using pump, which had to be
computed at the base station is partially carried out within the network.

6.3 Triggering of event in the case of failing battery supply or water level going
below | above the required threshold
Failing battery sends a notification to the the base station. If the battery level goes below the
threshold level ethresh, then a warning is sent to the base station carrying with it the information
pertaining to the energy level and the location of the sensor. This could assist the manager to
change the battery if it is a crucial node. If e11, e12, e13 are the energy values of the child nodes
and e10, that of the coordinator node, any node whose energy level is below the threshold will
report to the next coordinator node and the process is repeated until it reaches the base station.
The node whose energy level is below threshold along with its position is obtained at the base
station. This process is shown in the Figure 14. The same concept when used with the water level
indicator is shown in Figure 15.

Figure 14 Reporting the energy drained level to the base station

Figure 15 Water threshold indicator

159


Algorithm 6.3
a) Each non coordinator node node provides information of the data sensed along with its
energy information to the coordinator node.
b) If the non-coordinator data/energy is below the threshold, information is passed to the
coordinator node and the packet is dropped.
c) An event is triggered if the non-coordinaotr node or coordinator node is below/above the
threshold either for the energy or water.
d) If the path is not established to the base station, a new path is established to the base
station from the coordinator node and data is immediately sent without waiting for the
nodes along the path to sense their data.
e) Triggered information type whether it is battery(3) or water(4|5), remaining energy value,
data value, position and the interval when the reading was taken is intimated to the base
station.
Inference for algorithm 6.3:
a) Battery Depletion: This algorithm generates an event when the energy level goes below
the threshold level of handling even a single data transmission. If they are noncoordinator sensing nodes, an immediate message is passed on to the base station
without waiting for the coordinator node to initiate the transmission.
If it is a
coordinator node the battery low message is passed on as soon as the data is sensed with
higher priority. This could give an indication to the operator at the base station, the
location of the node whose battery is almost drainied. This gives an opportunity for the
operator to change the battery if the location of sensing is crucial or the self-organising
network will find an alternate path.
b) Water Level requirement: All nodes need not send message to the base station on timely
basis. The requirement of notifying the water level is when it is in excess or shortage of
water. This procedure can put all the nodes to sleep accept the ones responding to events.
The disadvantage being such events occuring rarely could make the operator unaware if
the network is dead or alive.

7. SIMULATION RESULTS
The simulation is carried out using ns2.34. Simulation parameters utilized in this work are as per
the Table 3 complying with Table 1. The comparison with minimum hop with total residual
energy is carried out in the earlier work [22]. The designing of the parameters for simulation is as
generated in the Table 3 is also addressed in this reference paper. In the current work, the time
interval for transmission of data is once in every 300 seconds. The simulation is executed for a
period of 6900secs. The source is node 0 and destination node 24 for topology as in Figure 5.
Table 4 shows the comparison of two proposed routing protocols. In the first protocol the
maximum amount of minimum residual energy along with total residual energy along with the
directional information is used for computing the route. In the second proposed protocol only the
maximum total residual energy is taken into account. It is observed that though many
transmissions take place during path establishment in the first case, the path established by this
method is alive for a longer interval of time with less amount of energy consumed. The extra
energy consumed in the second protocol was to find if any other path exists to reach the
destination. In this set up there is not much of a difference in the energy level between various
node energy level, hence only one extra transmission has taken place, else network could run
longer for the established path.

160

Table 3 Simulation parameters

Radio Parameters
Radio frequency
868 MHz
Antenna Height
Antenna Type
Transmit Power
Receive Power
Carrier Sense
Threshold
Capture Threshold
Gain of
transmitting and
receiving antenna

1m (min. reqd.
height 0.0819m)
Omnidirectional –
Quarter wave
3.16 mW =5dBm
−104dBm@
5dBm=3.98e-14W
−104dBm@ 5dBm

Simulation parameters
Date acquisition
Varying intervals of
interval
150/300/600seconds.
Nodes
21 coordinator nodes, 54
non- coordinator nodes
Topology
Square/Hexagonal
MAC
Queue

802.11
Drop Tail

Queue size

50

10 dB

Protocol

1

Transmission range

Proposed aggregation
algorithms
528m

Simulation period

Sensor parameters (Tiny Node)
Transmit Power
0.099W=19.95dBm
Receive Power

0.042W=16.23dBm

Sleep Power

0.000003W=
-25.2 dBm
0.006W =7.78dBm

Idle Power

1067110 seconds nearly
13 hrs.used for
aggregation protocol
Battery (Alkaline battery of 1.5V)
Battery supply
3V with 2 AA sized
alkaline battery
Power consumption 0.0705W for 23.5mA
discharge current
Energy
20304J for 80 hrs. of
consumption
active operation

Table 4: Comparison of routing protocol for consumption of energy and delivery ratio

Square
Topology with 25 nodes.

With Max-min energy+ With maximum
maximum total residual residual energy
energy + directional
information
23/24
95.8%
22/24
91.6%

Case I
Packet delivery ratio
24
data
Total average energy 7.9242
packet
consumed (J)
sent

Case II
Packet delivery ratio
1/1
One data
Total average energy 0.00204
packet
consumed (J)
sent

7.92043 –I!
0.00206 –T*
0.00178 –R#
100%

12.17217

0.00095 –I
0.00050 –T
0.00059 –R

0.001707

1/1

total

12.16840–I
0.001912–T
0.001853– R
100%
0.001027–I
0.000312 –T
0.000367 – R

All energy measurements are in Joules.
Split up of total energy is shown as I!-Idle energy T*-Transmit energy R#-Receive energy

161


Inference from the result:The path search process consumes more energy using the maximum
amount of minimum residual energy along with directional information and total energy residual
path than using only the total maximum residual path. But path once established it runs longer as
the path has higher minimal energy level. This will make the first node along the path to die at a
much later time than with just total maximum residual energy.
To carry out aggregation coordinator nodes are deployed using grid topology. Other noncoordinator nodes are dispersed randomly. Figure 16 shows the topology of the two types of
deployment used.
Ns2 does not support data handling directly. In order to incorporate handling of data, the
simulator is extended both in the application layer and the UDP layer. Providing data value for
all the 75 nodes is cumbersome, so inputs are stored in the files. Each file has information
pertaining each of the individual sensors, the data sensed by the sensor, along with the type of
aggregation required (average=1 or whole network=0) and time instance relative to current time
when the next type of aggregation has to be carried out.

(a) Hexagonal topology

(b) Square topology

Figure 16 Topology showing the region for aggregation in the block area.

If the input data is changing rapidly as in the case of rainfall or during irrigation the rate at which
data is read is fast (every 2.5 minutes=150s). If there is no water supply to the field the readings
are taken once in 10 minutes (600s). The network does not read data at night times. All the nodes
sense and send data to their neighbour node. It is multisource single destination concept. During
path search if a path is already established, nodes along the path when they sense the data, utilize
the existing path, instead of finding a new one.
A sample input reading of one of the sensor is shown in Table 5 and the output information of the
whole network stored at the base station is shown in Table 6.
Table 5 Example of the input file for one of the nodes –node id 9

Data value in cm Aggregation type
Relative time interval in seconds
(1 for average 0 for whole network)
2.90
0
0
2.92
0
600
2.93
0
600
2.95
0
600
2.104
1
300
2.208
0
300

162

Table 6 Sample output at the base station

Time at Time
Id of the Sensor
Position of Position of Agg.
Energy in
base
when
node
data
X
Y
Type* Joules
station
data was
sending reading coordinate coordinate
in
read in
the data in cm.
in meters
in meters
seconds seconds
2.06204 1.002000
51
1.060
842.0
1080.0
0
0.281649
602.061 601.00200
21
1.800
175.0
1000.0
0
20303.980
1202.06 1202.0000
18
2.010
0.0
1056.0
1
20303.975
1802.06 1802.0600
1
2.004
528.0
1056.0
0
20303.953
2102.06 2102.0600
9
2.104
0
528.0
1
20303.980
2402.06 24001.002
51
2.000
842.0
1080.0
2
0.025904
Aggregation type* 0-Average, 1- Whole network, 2- Low battery, 3- Low water level, 4- High
water level
The data received at the base station is stored in a file, the information conveyed to the base
station is the type of aggregation that was carried out (average or whole network) or the battery
discharge information or water access or shortage triggered information along with the time the
aggregation /event occurred and the position of the node which is conveying the information.
In Table 7 in column I, all the data are sent via the coordinator node, and this has higher amount
of energy consumption (11.7998J). In column II, the data is aggregated at each of the coordinator
node and packets are dropped after it sends the information to the next node and hence the
amount of energy consumed is the minimum (6.743J). The last column displays the amount of
energy consumed in associating the non-coordinator node to the coordinator node with energy
consumption of 0.01454J. The first 2 columns are run for a time period of nearly 13 hours
(1067110 seconds). The last column is for 30seconds when the non-coordinator node associates
with the coordinator node.
Inference as obtained from the results:It is observed that dropping the packets after it is sent to
the next hop neighbor or coordinator node saves considerable amount of data, instead of all the
nodes sending data to the base station using their own path
Table 7 Relative comparison of consumption of energy at various levels of aggregation

Column I

Column II

Column III

All
readings
are
in
Joules
Case I
Square
topology

Finding coordinator Finding coordinator
node
and node, aggregating
transmitting all data
and
selectively
transmitting
11.7998 0.0000 – I 6.743
0.0000– I
(Total
1.5902 – T (Total
0.9644– T
energy) 10.209 – R energy) 5.7788– R

Case II
Hexagonal
topology

11.0564
(Total
energy)

0.0000 – I
1.7376 – T
9.3188 – R

6.409
(Total
energy)

Finding coordinator

0.01454
(Total
energy)

0.0000– I 0.013159
0.9638– T (Total
5.4459– R energy)

0.00000 – I
0.00207 – T
0.01247 – R
0.00000– I
0.00213 – T
0.01102 – R

163


7. DISCUSSION AND CONCLUSION
Sensors are deployed in the agricultural field to monitor the moisture, humidity and temperature.
It is essential that the batteries have sufficient energy to transmit data for the entire cropping
period. To accomplish this, efficient routing and aggregation techniques are essential so as to
extend the life of the battery.
In this paper, different ways of establishing the path is studied. The path passing through nodes
should be such that that they have maximum amount of energy among all the different paths
reaching the base station from the source node. The path should be such that the path once
established is sustained for long intervals of time. The advantage of having such an approach is
that, the number of times the new path is searched gets reduced. This results in saving energy
which would have been expended because of unnecessary transmission and reception of control
signals. One of the approaches is to find the neighbour with maximum residual energy. The next
approach is the path which has the maximum total residual energy. Other approach is to follow
the path with maximum amount of minimum residual energy along the path. Directional
information also reduces a lot of unnecessary broadcasting of the control signals. All these
approaches are used to find efficient path from the source to the destination.
Aggregation techniques are used to avoid transmission of related data by aggregating them at the
coordinator node before transmission. This saves the processing time and computation at the base
station. Two types of aggregation are discussed; one is to find minimum data across the whole
network. Other is to find average value along a certain area of the network. Events are also
triggered when the battery level goes down or when the water supply exceeds the required
threshold or below desired value. The operator at the base station can take necessary steps to
handle the situation. Non-coordinator nodes are associated with the coordinator nodes based on
the minimum distance and maximum residual energy among the neighbouring coordinator node.
Higher the energy at the coordinator node, more number of nodes are associated with the
coordinator node. This balances the network which has uneven distribution of energy.
This work can be enhanced in the future by aggregating the data values based on closely related
information rather than the distance and energy information. This paper emphasizes the need of
finding efficient routing and aggregation techniques for the network to live long without wasting
energy on unnecessary transmission and reception. Having a proper decision support system
could help the farmers to take necessary action to monitor the way water is managed and enhance
the yield.

REFERENCES
[1]
[2]
[3]
[4]
[5]
[6]

Akyildiz, I.F., Su, W., Sankarasubramaniam, Y., Cayirci, E., (2002) “Wireless sensor networks: a
survey”, Journal of Computer Networks, Vol. 34, No. 4, pp. 393-422.
Al-Karaki, J.N, Kamal, A. E., (2004), “Routing techniques in wireless sensor networks: a survey”,
Journal of IEEE Wireless Communications, Vol.11, No. 6, pp.6-28
Panchard, J., (2009), PhD Thesis, “Wireless Sensor Networks for Marginal Farming in India “
Baggio, A. (2005), “Wireless Sensor Networks in Precision Agriculture”, Proc.Of ACM Workshop on
Real-World Wireless Sensor Networks, Sweden.
Corke, P., Sikka, P., et al. (2007), “Transforming Agriculture through Pervasive Wireless Sensor
Networks, IEEE Computer SocietyPervasive Computing”, Vol. 6, No. 2
Panchard, J., Rao, S., Prabhakar, T.V., Jamadagni, H.S., Hubaux, J., (2006) “COMMON-Sense Net:
Improved Water Management for Resource-Poor Farmers via Sensor Networks”, Int. Conf. on
Information and Communication Technologies and Development.
164

[7]
[8]
[9]

[10]
[11]

[12]
[13]
[14]
[15]
[16]

[17]
[18]
[19]
[20]
[21]

[22]

[23]

[24]

McCulloch John., et al., (2008), “Wireless Sensor Network Deployment for Water Use Efficiency in
Irrigation”, Proc. of the workshop on RealWSN’, New York.
Mohanoor A. B., Radhakrishnan S. and Sarangan V., (2009), “Online energy aware routing in
wireless networks,” Journal of Ad Hoc NetworksVol.7, pp.918–931.
Jia L., Noubir G., Rajaraman R. and Sundaram R.,(2006), ”GIST: group-independent spanning tree
for data aggregation in dense sensor networks,” Proc. of the 2nd IEEE international conference on
Distributed Computing in Sensor Systems, San Francisco, USA, pp. 282-304..
Bari A. et al., (2009), “A genetic algorithm based approach for energy efficient routing in two-tiered
sensor networks,” Journal of Ad Hoc NetworksVol.7, pp.665–676.
Liu T., Li Q. and Liang P. (2012),” An energy-balancing clustering approach for gradient-based
routing in wireless sensor networks,” Journal of Computer and Communication, Vol. 35, pp. 21502161.
Zhu Y. et al., (2008), “A Scalable Correlation Aware Aggregation Strategy for Wireless Sensor
Networks,” Journal of Information Fusion, Vol. 9, No.3, pp. 354-369.
Sutagundar,A. V, Manvi,S. S., (2012), “Wheel Based Event Triggered Data Aggregation andRouting
in Wireless Sensor Networks: Agent BasedApproach,”Journal of Wireless Personal Communication.
Kun-Hsien Lu et al., (2010), “Hierarchical Ring-Based Data Gathering for DenseWireless Sensor
Networks”, Journal of Wireless Personal Communication.
Maznah Kamat et al (2011), “Efficient In-network Aggregation for Wireless SensorNetworks with
fine grain Location-based Cluster,”MoMM2011, pp.66-71,2011
Misra S. andThomasinous P. D., (2010), “A simple, least-time, and energy-efficient routing protocol
with one-level data aggregation for wireless sensor networks”, The Journal of Systems and Software,
Vol.83, pp. 852–860.
Neng-ChungWang etal. (2010), “An Efficient Data Aggregation Scheme for Grid-BasedWireless
Sensor Networks”, IWCMC’10, June 28– July 2, Caen, France, pp.1213-1217.
Hsuan Penga, Yen-WenChen, (2013),” Energy consumption bounds analysis and its applications for
grid based wireless sensor networks”, Journal of Network and Computer Applications, pp. 444–451.
Pai, S. N., .Shet K.C, Mruthyunjaya H.S., (2011), “Efficient Deployment Strategies for sensors placed
in a sparse network”, International Journal of Micro and Nano Systems, Vol. 2, No.1 pp.11-14.
Tiny
node
584
Fact
sheet,
(2011),http://www.tinynode.com/?q=system/files/TN584_
Fact_Sheet_v_1_1.pdf version 1.1
Pai S.N., Shet K.C., Mruthyunjaya H.S., (2012),”Energy Aware Path Establishment and Maintenance
for Sensor Network with Parameters as used in Agricultural Field”, International Journal of Computer
Applications, Vol. 57, No.10.
Pai S.N., Shet K.C., Mruthyunjaya H.S., (2012),”Energy Aware Path Search for Sensor with
parameters as used in Agricultural Field”, International Journal of Adhoc, Sensors and Ubiquitous
Computing, Vol.3, No.5.
Weng C.E. and Lai T. W.,(2013),”An Energy-Efficient Routing Algorithm based on relative
identification and direction for Wireless Sensor Networks,” Journal of Wireless Personal
Communication Vol.69, No.1, pp. 253-268.
Sutagundar A. V. and Manvi• S. S.,(2013) ”Wheel Based Event Triggered Data Aggregation and
Routing in Wireless Sensor Networks: Agent Based Approach,” Journal of Wireless Personal
Communication, Vol. 71. No.1, pp. 491-517.

Authors
Smitha N. Pai is an Assistant Professor in theDepartment of CSE at MIT Manipal. She
obtained her M.Tech. in CSE from ManipalUniversity in 2002 and B.E in E&C
fromMysore University. She is currently pursuing her PhD. from Manipal University. Her
main interest lies in wireless sensor, adhoc networks.

Dr.K.Chandrashekar Shet obtained his B.E, M.Sc. (Engg) andPh.D degrees from
Mysore University, Sambalpur Universityand IIT Bombay in the years 1972, 1979 and
1987 respectively.He is working in NITK Surthakal since 1980 and presently he is the
professor in the Dept. Of Computer Sc. & Engg, and Dean(Faculty welfare).He has
165


published around 250 papers, in National, International,journals/conferences. Besides, he has published
three books, on Micro-processors, Software Engg. & quality Assurance. He has produced 10 Ph.D
professionals and currently six are pursuing research under his guidance leading to Ph.D.
Dr. H. S. Mruthyunjaya has completed his bachelor degree in Electronics and
Communication Engineering from Mysore University in 1988 and obtained his masters
degree in Electronics and Control Systems Engineering from Birla Institute of
Technology and Science, Pilani in 1994. He has a Ph.D in Electronics and
Communication Engineering conferred by Manipal University for his thesis entitled
‘Performance Enhancement of Optical Communication Systems and Networks using
Error Control Techniques’. He is currently serving as a Professor in the Department of Electronics and
Communication Engineering, Manipal Institute of Technology, Manipal, India where he joined as a
Lecturer in the year 1998. He has done research on countering non-linear effects and other noises in
WDM all-optical networks by employing error control coding techniques. His areas of major interests are
the Optical Fiber Communication systems, Fiber Optics, Photonic Crystal Fibers, WDM networks and
systems, Electromagnetic theory & General areas of Digital Communication Systems. He has authored or
co-authored over Forty three technical papers in refereed journals and International conference
proceedings. He is a Fellow of the Institution of Engineers (India) and member of Indian Society for
Technical Education.

166

In network aggregation using efficient routing techniques for event driven sensor network

Recommended

More Related Content

What's hot (20)

Similar to In network aggregation using efficient routing techniques for event driven sensor network (20)

More from IJCNCJournal (20)

Recently uploaded (20)

In network aggregation using efficient routing techniques for event driven sensor network