C018141418

IOSR Journal of Computer Engineering (IOSR-JCE)
e-ISSN: 2278-0661,p-ISSN: 2278-8727, Volume 18, Issue 1, Ver. IV(Jan – Feb. 2016), PP 14-18
www.iosrjournals.org
DOI: 10.9790/0661-18141418 www.iosrjournals.org 14 | Page
Energy Efficient Query Optimization in WSN usingThreeLevel
Modelling
C.Komalavalli1
,VijayBhushanAggarwal2
1
Department of Information Technology, Jagan Institute of Management Studies, Rohini, New Delhi, India
2
Department of Information Technology, Jagan Institute of Management Studies, Rohini, New Delhi, India
Abstract:With advancement in microprocessor technology, wireless sensor networks is playing vital role in
different types of applications such as health care monitoring, fire detection etc. In these applications, large
amount of data generated by the sensors are communicated to the base station, which will provide final result
for the query. Since the sensor nodes are energy constraint devices, energy efficient query methods and storage
of the data is required. Clustering is playing an important role in enhancement of the energy level of the sensor
and scalability of the network. Since communication depletes the energy level of the sensor, further selection of
cluster head is important for reducing the communication cost. In this paper, we are introducing the model
along with algorithm for the cluster head selection and efficient query processing in WSN.
Keywords:WSN, Clustering, Base Station, Cluster Head, Tiny DB
I. Introduction
Wireless sensor network is a network which consists of large number of sensor nodes distributed in a
geographical area for monitoring temperature, humidity, sound, pressure etc.[1]. The distance between nodes is
small compared to wired network. They are able to sense the phenomenon for which they deployed process and
send the data to the centralized server known as a gateway. Sensor nodes are battery powered and energy
getsdepleted in the nodes in due course of time. It is impractical to charge or replace the battery in the sensor
nodes of the WSN. Since the communication from nodes is more energy consuming than the computation at the
nodes, it is very important to minimize the communication at the nodes.
Users are sending the queries to the base station which are then transmitted to the sensor nodes. Every
time, the sensors are collectingrequired data which then communicated to the base station. For managing these
data, WSN is viewed as distributed database and sensors are considered as datasources. They are storing the
datain the rows of a relation, same like traditional database systems.WSN can be programmed for data sensing
and collection. Since the energy consumption is one of the major issues, data collection process from the nodes
to the base station should be optimized so as to minimize the energy consumption. [2]
Fig:1Wireless Sensor Network Architecture[1]
The limited computing power, life and energy of sensor nodes are to be considered for the efficient
query processing.
Since energy depletion is mainly due to communication between nodes, various reduction techniques
are existing for resolving this issue. Data aggregation techniques aggregates the data in the intermediate nodes

Energy Efficient Query Optimization in WSN using Three Level Modelling
and packet merging combines the small pockets into big one without considering the semantics and correlation
between packets. Data compression techniques involve data encoding and decoding at the sensor and sink node
respectively.
Sensors can be placed anywhere, data should be collected and sent to the base station. In order to
minimize the communication cost of query processing, various optimization techniques for WSN at different
levels have been discussed in the literature.[1]
Our consideration in this paper focuses on network level and it involves query dissemination, data
aggregation, network topology and programming. Queries can be compared according to the query attributes or
query conditions.
To reduce the overall cost of query processing, we are considering threelevels in the network: Base
station, cluster head and sensor node. Clustering means group of nodes workingtogether under one leader known
as cluster head. Cluster heads [3] also one of the sensor nodes and can be selected using different algorithms.
Sensors will be sending data to the cluster head which in turn will communicate to the base station, thus
reducing the overall communication cost. Cluster Head also maintains the metadata about the sensors in its own
cluster and physical location within the network.[4]. The key attributes such as number of clusters, nodes in
each cluster, intra cluster communication, nodes role, cluster head selection and multiple level cluster approach
are to be considered for the designing the cluster architecture. [5] [6]
The rest of the paper is organized as follows. Section 2 includes related work. Section 3 describes
problem formulation. Section 4 describes the proposed architecture and the algorithm optimizing
communication cost. Finally Section 5gives conclusions and future scope of this work. Section 6 gives
references.
II. Related Work
The sensor databases are different fromour traditional databases. The latter are mainly focussing on
collected data rather than extracting the data from the network. But sensor networks are focussing of both the
extraction and collection of the data from the network.
The two approaches, data ware house and distributed approach are existing for the efficient data storage
and query processing in WSN. In data warehouse approach, the sensors are collecting the data and sending the
data to the base station periodically. Only base station is responsible of processing the entire data. This approach
leads to the wastage of resources and creating bottleneck because of large volumes of data from sensors to base
station.[2]
In the second approach, the node itself is treated as adata source and they are not sending the data
periodically to the server, thus it forms a distributed approach.[4]. The queries are injected into the base station
and disseminated into network according to the routing protocol. The sensor nodes send their data to their parent
node called cluster head and cluster head aggregates the data from different nodes and sends that to the base
station. This approach is called network processing and it reduces the amount of transmitted data and latency.[3]
There are so many clustering protocols which are proposed in the literature such as LEACH, HEED,
TEEN, PEGASIS, and EEHC for efficient processing of the queries.
Low Energy Adaptive Clustering Hierarchy (LEACH) algorithmwas proposed by Heinzelman et.al [7]
in which the cluster formation is in a distributed manner. In this algorithm, any sensor node can be the cluster
head randomly and it broadcasts its status to other nodes. Thus a node with low energy can be selected and if
that node dies, that particular cluster will become non-functional leading to inaccurate result. This protocol is
not focussing on data recovery techniques at the time of node failure.
Power-Efficient Gathering in Sensor Information Systems (PEGASIS) was proposed by S. Lindsey et
al. [8] in which each node receives and transmits only to its close neighbours. Every node becomes a leader and
transmitting the data to the base station. This approach distributes the load evenly among the sensor nodes.
Manjeshwar et.al [9] introduced an algorithm called Threshold sensitive Energy Efficient sensor
Network protocol (TEEN) which is designed for distributed reactive network applications. It is based upon hard
and soft threshold values. If these values are not received by the sensor nodes, then the data cannot be retrieved
by the network.
Hybrid Energy Efficient Distributed Clustering (HEED) was proposed by Younis et al. [10] for real
time applications. It is based upon residual energy and intra-cluster communication cost. Unlike LEACH, only
sensors having higher energy can be selected as cluster head. For finding the communication cost, both the intra
and inter cluster communication cost is used but it is not considering the data recovery of the network.
Energy Efficient Hierarchical clustering (EEHC) [6] is a randomized, distributed clustering algorithm
and that has been divided into two stages. In the first stage, each node itself announces as cluster head and
becomes the volunteer on the basis of the probability of the node. In the next stage, clustering hierarchy is
formed and aggregated data will be sent using the hierarchy level. The node which is nearest to the base station
becomes the relay for the other cluster heads.

III. Problem Formulation
Clustering is the most convenient way of managing the communication overhead and data aggregation
in wireless sensor networks. Even though large numbers of algorithms are present but energy consumption
during cluster formation is very high. So working condition of the cluster head is playing an important role
when the nodes are dying. In this paper, we propose an algorithm for the checking the working condition of
cluster head and selection of cluster head according to energy level.
IV. Proposed Architecture
Assumptions:
 Sensor nodes are static in nature.
 Sensors broadcast their energy level to the bases station periodically.
4.1 System Model
The network consists of many clusters and each cluster consists of several sensor nodes. The clusters
are denoted by c1, c2, and soon. Every sensor has a unique id and will be denoted by cluster number and sensor
number. If the nodes are in first cluster, then id will be c101, c102, and soon. In every cluster, one node will be
considered as cluster head and sensor id will be assigned as cluster head id. We are maintaining these details in
the base station. Only cluster head can communicate to the base station and send the aggregated data to the base
station.
In the base station we are managing two tables namely cluster head details and sensors details.
Table I:Schema for the Cluster Head details
Cluster Head ID Stores the node ID which is currently Cluster Head
Position Latitude and Longitude(Lat 40°26'North; Long3°42'West)
Number of Nodes Number of nodes in that particular Cluster
Status Working condition of Cluster head (Working or Not working)
Table 2: Schema for the Sensors Details
Cluster Head ID Stores the node ID which is currently Cluster Head
Sensor ID It stores the ID of the sensors
Criteria Phenomenon such as Temperature, Sound, Pressure etc
Data Unit Celsius,Decibels, Pascal etc.
Energy Level Energy Level of the Sensor in Joules
4.2 Query Execution
Users will forward the query to the base station. Every query will have unique id.For efficient query
processing, Meta data is playing a vital role. Meta data is managed in two levels: 1) Base station and2) Cluster
Head.
There are two phases for executing the query. First phase is called setup phase and second phase is
called execution phase.
4.3 Setup Phase
In the base station, we are managing the data about cluster head details and details about the sensor
nodes also. All sensor nodes broadcast their energy level to the base station and keep updating the sensor table
periodically.
Initially, the base station checks whether the result of the query is available in the memorycache. If the
result is present, it retrieves and returns the result to the user. If the result is not there, base station checks for
cluster head working condition. If all cluster heads are working, then query will be disseminated to all the
cluster heads of the network.
If the status of any cluster head is “not working”, then base station checks the energy levels of all
sensor nodes in that particular cluster. The node which is having maximum energy will be selected as cluster
head and this will be communicated to all the sensors in that particular cluster. The id of this sensor will be
assigned as cluster head id and updated the cluster head table.
Setup cost in the base station can be calculated as:
SC(Bs) = h ∗ v𝑛
𝑖=1 (1)
Where n represents the number of clusters, h is number of hops between cluster head and base station
and v represents the size of the data. The cost has to be calculated only once when the cluster head is not
working in the particular cluster. Then it remains same for further query processing.

4.4Query Execution Phase
In the cluster head, sensor table(Table2) is also maintained and it gets updated alongwith the base station.
In this phase, query will be forwarded from the cluster head to the child nodes using modified Bellman
Ford algorithm.[4]
This algorithm was proposed by the authors [4] for getting the shortest path. The algorithmhas been
modified here according to the working condition of the cluster head. Cluster head checks for the sensor table
and find the sensor type matching with the query. Then query will be disseminated to that particular nodes and
data will be retrieved from the nodes. Cluster head collects, aggregates the data and send the result to the base
station.
Modified Bellman Algorithm [4]
Step 1:
Initialization
For each vertex v in vertices
If v is source then
v.distance=0;
Else
v.distance=infinity;
v.predecessor=null;
Step 2:
For i to size [vertices]-1
For each edge UV in edges
If node flag=1 and v.flag=1Then// working condition of cluster head is checking here
U=uv.source
V=uv.destination
If u.distance + uv.weight<v.distance
v.distance=u.distance=uv.weight
v.predecessor=u
Next edge
Next i
// for negative cycle
For each UV in edges
U=uv.source
V=uv.destination
If u.distance + uv.weight < v.distance
Then error “graph contains negative weight cycle”
Loc=0
V=sink
While v! =source
Route [loc++] =v
If (v.predecessor == source)
then query processing will be done
Delete_edge (v, v.predecessor)
V=v.predecessor
Since within a cluster Ci, nodes are represented as CiS1, CiS2,…..CiSj,….,.CiSmi,thereforeQuery Execution Cost
(QxE(Ci)) for cluster Ci, can be calculated as:
QxE(Ci) = CiSj𝑚𝑖
𝑗 =1
𝑚𝑖
𝑗=1 where Sj= projection (d) * size (d) (2)
whered is phenomenon such as temperature, sound, pressure,etc.(matching criteria).Thus, total Query Execution
Cost for all the clusters C1,C2, ….Cn can be calculated as:
Total QxE= QxE(Ci)𝑛
𝑖=1 (3)
Once the base station receives the aggregated data, it can process and send the results back to the user.
At the same time, data is stored in the base station for future use also.
This model reduces the setup cost by checking the working condition of the cluster head which is
maintained in the base station. The status and energy level of all sensorsare maintained in the base station. In
case all cluster heads are in working condition, the processing proceeds to the next phase. If any cluster head is
not working, then another sensor in that cluster having maximum energy level is selected as new cluster head.
This maximizes the life of the newly selected cluster head.

In each cluster, large numbers of sensors are there, however each sensor can sense only specific criteria
(temperature, pressure, sound etc.) and this information is maintained in the cluster head. Therefore upon
receiving the query, cluster head sends the particular query to those limited number of sensors which can answer
that particular query. Hence the communication cost during the execution phase is likely to be reduced to the
greater extent.
V. Conclusion
In this paper, we purposed the model for selection of cluster heads depending upon their working
condition and energy level. This ensures longer life for each cluster head, since sensor with maximum energy is
selected as new cluster head. This reduces the setup cost in the initial phase of the query.
In the query execution phase, communication cost is expected to be reduced since the query is being
sent only to the nodes which are meeting the required criteria, as detailed above. In future work, this model and
the algorithm will be implemented using MATLABsimulator. Further comparisons will also be carried out with
other models for the same purpose given in the literature.
References
[1]. Ousmane Diallo, Joel J. P. C. Rodrigues, Senior Member, IEEE, Mbaye Sene, and Jaime Lloret, Senior Member –“Distributed
Database Management Techniques for Wireless Sensor Networks”,IEEE Transcations on Parallel and Distributed Systems,IEEE
2013
[2]. MeenakshiDiwakar, Sushil Kumar – “ An Energy Efficient Level Based Clustering Routing Protocol For Wireless Sensor
Networks”, International Journal Of Advanced Sensor Network Systems, Vol 2, No 2, April 2012
[3]. Devesh Pratap Singh, R. H. Goudar , Bhasker Pant , Sreenivasa Rao-“ Cluster head selection by randomness with data recovery in
WSN” , Aug 2014, Springer
[4]. C. Komalavalli, Chetna Laroiya Energy Efficient Protocol in Wireless Sensor Network: Reactive On-Demand Routing, IITM
Journal for Management & IT, Vol 2 Issue 1,2010
[5]. SrieVidhya Janani. E, Ganeshkumar.P, Vasantha Suganthi.G, Sultan.M, Kaleeswaran. D,” A Survey on Algorithms for Cluster Head
Selection in WSN”, International Journal of Advanced Research in Computer Engineering & Technology (IJARCET) Volume 2, No
5, May 2013
[6]. S. Bandyopadhyay, E. Coyle, An energy efficient hierarchical clustering algorithm for Wireless sensor networks, in: Proceedings of
the 22nd Annual Joint Conference of the IEEE Computer and Communications Societies (INFOCOM 2003), San Francisco,
California, 2003
[7]. W. Heinzelman, A. Chandrakasan and H. Balakrishnan, “Energy- Efficient Communication Protocol for Wireless Micro sensor
Networks” Proceedings of the 33rd Hawaii InternationalConference on System Sciences (HICSS ‘00), 2000, pp. Vol.2:10.
[8]. S. Lindsey, C. Raghavendra, “PEGASIS: power-efficient gathering in sensor information Systems,” in: Proceedings of 2002 IEEE
Aerospace Conference, Big Sky, Montana, USA,vol.3, 2002, pp.1125-1130.
[9]. A. Manjeshwar, D. Agrawal, “TEEN: a routing protocol for enhanced efficiency in wireless sensor networks,” in: Proceedings of
15th International Parallel and Distributed Processing Symposium (IPDPS’01), San Francisco, CA, 2001, pp. 2009-2015.
[10]. O. Younis and S. Fahmy, “HEED: “A Hybrid Energy-Efficient Distributed ClusteringApproach for Ad Hoc Sensor Networks,” IEEE
Transactions on Mobile Computing, 2004,Vol. 3, No. 4.

C018141418

More Related Content

What's hot (17)

Viewers also liked (20)

Similar to C018141418 (20)

More from IOSR Journals (20)

Recently uploaded (20)

C018141418