Docstoc

ENERGY EFFICIENT HIERARCHIAL DATA AGGREGATION IN

Document Sample
ENERGY EFFICIENT HIERARCHIAL DATA AGGREGATION IN Powered By Docstoc
					 National Conference on Role of Cloud Computing Environment in Green Communication 2012
                                                                                                                     343


             ENERGY EFFICIENT HIERARCHIAL DATA AGGREGATION IN

                                   WIRELESS SENSOR NETWORKS

  C.Sahaya Anbarasu,                                      Mrs.A.AntonyViswasa Rani M.E.,(PhD)
  M.E Computer Science & Engineering,                     Assistant Professor,
  Sun College Of Engineering & Technology,                Sun College of Engineering and Technology,
  Nagercoil, Tamil Nadu.                                  Nagercoil, Tamil Nadu.
  mailto:sahagce@gmail.com                                mailto: ranionse@yahoo.com

ABSTRACT

                Wireless Sensor Network has the
potentiality to connect the physical world with the
virtual world by forming a network of sensor nodes.        1. INTRODUCTION
As sensor nodes are battery driven, an efficient
utilization of power is essential in order to use          1.1 Data Aggregation
networks for long duration hence it is needed to
reduce data traffic inside sensor networks, reduce                            It is widely accepted that the
amount of data that need to send to base station.          energy consumed in one bit of data transfer can be
Sensor nodes need less power for processing as             used to perform a large number of arithmetic
compared to transmitting data. Energy saving in these      operations [10] in the sensor processor. Moreover in
approaches can be obtained by cluster formation,           a densely deployed sensor network the physical
cluster-head election, data aggregation at the cluster-    environment would produce very similar data in near-
head nodes to reduce data redundancy and thus save         by sensor nodes and transmitting such data is more or
energy. For this purpose, a data aggregation tree,         less redundant. Therefore, all these facts encourage
rooted at the sink, is constructed in the network.         using some kind of grouping of nodes such that data
Node clustering can be employed to further balance         from sensor nodes of a group can be combined or
load among sensor nodes and prolong the network            compressed together in an intelligent way and
lifetime. The main goal of data aggregation                transmit only compact data. This can not only reduce
algorithms is to gather and aggregate data in an           the global data to be transmitted and within each
energy efficient manner so that redundant data are         individual group, but reduces the traffic and hence
avoided and network lifetime is enhanced. It is            contention in a wireless sensor network. This process
preferable to do an energy efficient hierarchical          of grouping of sensor nodes in a densely deployed
cluster-based data aggregation scheme which is a           large-scale sensor network is known as clustering [1].
bottom-up clustering approach by grouping similar          The intelligent way of combining and compressing
nodes together before the cluster head (CH) is             the data belonging to a single cluster is known as data
selected inside the network and reduction of packet        aggregation.
size. To achieve the goal, an energy efficient HAC
                                                                          Routing protocols [4] providing an
(Hierarchical Agglomerative Clustering) algorithm
                                                           optimal data transmission route from sensor nodes to
for WSNs is to be proposed . HAC uses simple
                                                           sink to save energy of nodes in the network. Data
numerical methods to make clustering decisions. In
                                                           aggregation plays an important role in energy
addition, HAC provides flexibility with respect to
                                                           conservation of sensor network. Data aggregation
input data type (e.g., location data or connectivity
                                                           methods are used not only for finding an optimal path
information) and weight assignment to different
                                                           from source to destination but also to eliminate the
factors (e.g., connections or power strength).
                                                           redundancy of data, since transmitting huge volume
Keywords : Redundancy, cluster , head rotation ,           of raw data is an energy intensive operation, and thus
data fusion , data aggregation .                           minimizing the number of data transmission. Also
                                                           multiple sensors may sense the same phenomenon,
                                                           although from different view and if this data can be
                                                           reconciled into a more meaningful form as it passes



Department of CSE, Sun College of Engineering and Technology
 National Conference on Role of Cloud Computing Environment in Green Communication 2012
                                                                                                                   344


through the network, it becomes more useful to an       algorithms adopt the head node on the basis of the
application.                                            distance (how far the Base-station is located from the
                                                        head node) and its energy level. After the cluster
                                                        head arrangement phase, algorithms constructs a
                                                        routing tree over the set of head nodes but only the
                                                        higher residual energy nodes can communicate with
                                                        the Base-station by single-hop communication. The
                                                        remainder of the paper is prepared as follows: In
                                                        Section 2, some related work is presented. In Section
1.2 Need for Data Aggregation
                                                        3, the network radio model for energy calculations
                                                        and problem statement has been discussed. In Section
                                                        4, the details of centralized algorithms for hierarchial
                                                        data aggregation have been provided. We present
                                                        simulation results and discussion in Section 5.
                                                        Section 6 concludes the paper.

                                                        2. RELATED WORK

                                                        Heinzelman et al. [1] propose LEACH, a substitute
                                                        clustering based algorithm. In order to save energy,
                                                        LEACH deals with the heterogeneous energy
                                                        condition is the node with higher energy should have
                                                        larger probability of becoming the cluster head. Each
                                                        sensor node must have an approximation of the total
                                                        energy of all nodes in the network to compute the
                                                        probability of becoming a cluster head but it can not
Fig :1 Data aggregation model and Non data
                                                        make decision of becoming a cluster head only by its
aggregation model
                                                        local information, so the scalability of this scheme
 The compulsory requirement for data aggregation in     will be influenced. Sh. Lee et al. suggest a new
wireless sensor network can be visualized from the      clustering algorithm CODA [8] in order to mitigate
above fig : 1.The basic idea of anytime and anywhere    the unbalance of energy depletion caused by different
computing leads to the new field called mobile          distance from the sink. CODA divides the whole
computing. The advances in the wireless technology      network into a small number of groups based on the
are also one of the major stimuli for the growth of     distance from the base station and the strategy of
mobile computing. But here in this ubiquitous           routing and each group has its own number of cluster
computing environment we can’t follow the normal        members and member nodes. The farther the distance
architecture and protocols which have been used in      from the base station, the more clusters are formed in
the fixed network due to it’s battery powered devices   case of single hop with clustering. It shows better
involved in the computing and transmission of the       performance than applying the same probability to
data. The advancement in these miniature computing      the whole network in terms of the network lifetime
model and wireless transmission techniques lead to      and the dissipated energy.
the development of the wireless sensor networks.
                                                        In [7] authors report an algorithm based on chain,
Sensor networks are needed in the applications like
                                                        which uses greedy algorithm to form data chain. Each
environment monitoring, industrial control units,
                                                        node, aggregates data from downstream node and
military applications and in the context aware
                                                        sends it to upstream node along the chain and
computing environments.
                                                        communicates only with a close neighbor and takes
In this paper, we propose an energy efficient           turns transmitting to the base station, thus reducing
hierarchical data collecting algorithms for             the amount of energy spent per round.
heterogeneous sensor networks. Algorithms include
                                                        In [9], the authors discuss a HEED clustering
two phases: the cluster head arrangement phase and
                                                        algorithm which periodically selects cluster head
the routing phase. For the cluster head arrangement,
                                                        based on the node residual energy and node degree



Department of CSE, Sun College of Engineering and Technology
 National Conference on Role of Cloud Computing Environment in Green Communication 2012
                                                                                                                    345


and a secondary parameter, such as node proximity                 from node A to node B is the same as that of
to its neighbors or node degree. The clustering                   transmission from node B to node A.
process terminates in O(1) iterations and it also        Each sensor nodes can operate either in sensing mode
achieves fairly uniform cluster head distribution        to monitor the environment parameters and transmit
across the network and selection of the secondary        to the base station or cluster head mode to gather
clustering parameter can balance load among cluster      data, compress it and forward to the BS.
heads.                                                    ETx = Eelec * l + Efs * l * d2 , d < d0

In [10] the authors introduce a cluster head election     ETx = Eelec * l + Emp * l * d4 , d ≥ d0
method using fuzz logic to overcome the defects of
LEACH. They inquired that the network lifetime can       and for receiving this message respectively is:
be prolonged by using fuzz variables in homogeneous
network system, which is different from the               ERX =Eelec * l
heterogeneous energy consideration. In [3] the
authors propose an EDGA algorithm to achieve good        where Eelec is the energy spent to operate the
performance in terms of lifetime by minimizing           transceiver circuit, Efs and Emp are the energy
energy consumption for in-network communications         expenditure of transmitting one bit data to achieve an
and balancing the energy load. It is based on            acceptable bit error rate and is dependent on the
weighted election probabilities of each node to          distance of transmission in the case of free space
become a cluster head, which can better handle the       model and multipath fading model . If the
heterogeneous energy capacities and adopt a simple       transmission distance is less than a threshold d0, the
but efficient method to solve the area coverage          free space model is applied; otherwise, the multipath
problem in a cluster range.                              model is used. The threshold d0 is calculated as d0 =
                                                         (Efs)1/2 / Emp
Recently, in [2,4], authors suggested the impact of
heterogeneity of nodes in terms of their energy that     3.2 Data Aggregation - Issues
are hierarchically clustered in WSNs and initiate an
                                                                    There are some issues involved with the
energy efficient heterogeneous clustered method for
                                                         process of clustering in a wireless sensor network.
WSNs based on weighted election probabilities of
                                                         First issue is, how many clusters [5] should be
each node to become a cluster head according to the
                                                         formed that could optimize some performance
residual energy in each node. For this they suppose a
                                                         parameter. Second could be how many nodes should
percentage of the population of sensor nodes is
                                                         be taken into a single cluster. Third important issue is
equipped with the additional energy resources.
                                                         the selection procedure of cluster-head in a cluster.
3. MODEL           FOR WIRELESS            SENSOR        Another issue that has been focused in many research
NETWORKS                                                 papers is to introduce heterogeneity in the network. It
                                                         means that user can put some more powerful nodes,
3.1 Assumptions                                          in terms of energy, in the network which can act as a
                                                         cluster-head and other simple node work as cluster-
A Fixed network which includes in mobile sensor          member only. Considering the above issues, many
nodes and base station is considered in our study with   protocols have been proposed which deals with each
the following assumptions .                              individual issue.

    1.   The network is considered homogeneous           4. ALGORITHM FOR HIERARCHIAL DATA
         and all of the sensor nodes have the same       AGGREGATION (HDA)
         initial energy.
    2.   Each sensor node knows its own                  4.1 Explanation of Proposed Algorithm
         geographical position.
    3.    All nodes measure the environmental            The existing energy efficiency model for the sensor
         parameters at a fixed rate and send it          network shows considerable improvement in one or
         periodically to the receiver nodes.             more objectives to suite the specific application, still
    4.    The radio channel is symmetric such that       there needs a lot of work to be done on energy
         energy consumption of data transmission         efficient model in terms of low clustering overhead,
                                                         distributed cluster heads, continuous packet delivery,


Department of CSE, Sun College of Engineering and Technology
 National Conference on Role of Cloud Computing Environment in Green Communication 2012
                                                                                                                       346


reduced data fusion cost. In this project a new hybrid
protocol model is to be proposed which considers all
these factors in the routing mechanism for the
wireless sensor network. The following are the steps
involved for adopting the proposed hybrid model.
                                                         By incorporating small changes in each step, we hope
                                                         this hybrid model will improve the efficiency of
                                                         routing protocol for Wireless sensor networks.

                                                         4.2 Design Modules
                                                          Module 1: Formation of network with the required
                                                         specifications..

                                                         Module 2:      Energy estimation and Cluster Head
                                                         formation.

                                                         Module 3: Transmission of aggregated data to the
                                                         sink.

                                                         4.3 Specification of Modules

                                                                    In Module 1, a well defined network is to be
                                                         formed considering the following parameters like the
                                                         number of nodes to be deployed, nodes localization
                                                         i.e, their positions , the node’s intial energy levels etc.
  1.Formation of network with suitable number of         Further the sink node floods the entire network
  sensor nodes and a single sink.                        requesting their current status , thereby the sink node
  2.The network topology is composed of                  gains the global knowledge about the whole system.
  tranceiving nodes placed between the sink and          This acquirement of global knowledge avoids the
  sensor nodes.                                          sink’s request to the unable nodes to gather the data
  2.The sink node sends a request message to all         to be sensed. This strategy avoids the energy
  sensor nodes inorder to acquire their                  wastage for requesting those unable nodes to face the
  configurations .                                       overhead for sensing the data.
  3.The sensor nodes response to the sink and hence
  the sink node attains the global knowledge about                 In Module 2, the energy spent for reception
  the system especially the energy levels.               , transmission , computation etc, during the network
  4.The lower level nodes that are capable of sensing    initialization phase is to be found and hence the
  the data are allowed to perform data gathering.        residual energy level at all the nodes are determined .
  5.After the sensor nodes complete data sensing,        The sink node establishes a threshold energy level
  clustering occurs.                                     that must be sufficiently present in a sensor node so
  6.The sensor node with maximum residual energy         that it is capable of sensing the particular event
  is elected to be the cluster head.                     Hence it is a must to distinguish the nodes into two
  7.If single cluster members sense a data,then data     types namely the nodes which have residual energy
  fusion is employed to obtain the best data.            greater than the threshold are called active ones and
  8.If two or more cluster members sense a data,then     the nodes which have residual energy lesser than the
  data aggregation is employed at the cluster heads      threshold are called dormant ones.
  to obtain the best data.
  9.The cluster heads send the fine data to the next               Depending upon the application and the
  level of tranceiving nodes.                            environment , the amount of energy spent by a sensor
  10. At each transceiver level, the node with           node is to be dealt in two scenarios. Under scenario
  maximum energy is selected to be the routing path      1,all the active nodes may spend equal amount of
  inorder to avoid packet loss.                          energy for sensing a specific data and hence there is
  11. Cluster head rotation is effected for each         no need for reclustering the system in the next cycle.
  subsequent cycles of data sensing based on the
  existing residual energy levels available at each
  clusters at that instance.
Department of CSE, Sun College of Engineering and Technology
 National Conference on Role of Cloud Computing Environment in Green Communication 2012
                                                                                                                     347


Under scenario 2, all the active nodes may spend
varying amount of energy for sensing a specific data
and hence it is essential for reclustering the system in
the next cycle.

          Now comes the central region of the project,
i.e, selection of suitable clusters and the respective
cluster heads . From the available set of active nodes
, given number of clusters to be formed based on K-
means algorithm using the associated distance
metric based on the Euclidean distance formula. For
                                                           Fig: 2 Hierarchical network formation
each cluster formed , a head node called cluster head
is to be identified which encompasses the maximum
of the residual energy of all the nodes in each of
the respective clusters.

            In Module 3, based on the environment ,
the type of data and the type of constraints specified
in the query the active nodes start gathering of data.
The received data at all the children nodes are fused
with the data available at their parent nodes. This
fusion process continues till the sink node is reached.
Further data correlation is done at all the parent nodes
for matching the attributes given in the query string.

         The sensor nodes that act as the interface
nodes between the cluster heads and the sink node are
called gateway nodes where the aggregation process         Fig 3 : Hierarchical cluster formation
is done using aggregator functions and the final
                                                           6. CONCLUSION
refined data is received by the sink.
                                                                            One of the most important
5. SIMULATION RESULTS
                                                           constraints in wireless sensor networks is the energy
We have implemented our proposed protocol in NS-           consumption. Aggregation algorithms have a
2(ver. 2.31) [12]. We considered a 20 node random          considerable role in decreasing the energy
network deployed in an area of 360 X 360 m within          consumption due to the reduction of the transmitted
which suitable number of sensing clusters are              data volume.          Aggregation reduces power
formed along with their respective cluster heads and       consumption on avoiding the communication directly
12 tranceiving nodes are deployed in a tree based          between sink and sensor nodes. The idea is to
hierarchical manner for routing the sensed data to the     combine the data coming from different sources,
sink. The type of MAC used is 802.11 The only Sink         eliminating redundancy, minimizing the number of
node is assumed to be situated 100 meters away from        transmissions and thus saving energy. In this work,
the above specified area. At the same time, we             an energy efficient hierarchical cluster based
considered specified area that is divided into 90 X 90     algorithm to construct the aggregation tree is
m square area called cluster and each cluster is           presented. The algorithm considers both energy and
divided into 30 X 30 m area called virtual grid.           distance to construct the aggregation tree.
Obviously, the first set of cluster heads are taken        Furthermore, an energy-driven method to rotate
randomly. The initial energy of all the nodes assumed      cluster-head instead of time-driven cluster-head
as 5 joules. The radio range is varies from 30m to         rotation is also adopted to balance the energy
130 m,each data packet has 64 bytes, and the others        consumption. It is expected that the simulation results
are 32 bytes long. The screenshots of hierarchical         show better performance than the existing algorithms
network formation and cluster formation are shown          and also, the algorithm decreases the number of
in fig 2 and fig 3 respectively.                           failure nodes and provides higher network lifetime



Department of CSE, Sun College of Engineering and Technology
 National Conference on Role of Cloud Computing Environment in Green Communication 2012
                                                                                                                 348


and better coverage. It is believed that this algorithm   [6] An Energy Efficient Data Gathering Scheme in
can offer significant improvement on the                  WSN       Using    Spannig    Tree    -   Kaushik
performance and energy-efficiency of mobile sensor        Chakrabarty,Abhrajit Sengupta - IEMCON 2011
networks if further research on mobility model is         organised by IEM in collaboration with IEEE on 5th
carried out. There are several interesting and            & 6th of Jan,2011.
challenging questions that yet remain to be answered
here. For example, modeling and quantifying the           [7] Efficient Cluster Head Selection Scheme for Data
savings in data aggregation that result through the       Aggregation in Wireless Sensor Network -Kiran
optimizations at the topology management and              Maraiya Kamal Kant Nitin Gupta -International
scalability phases pose a number of challenges.           Journal of Computer Applications (0975 – 8887)
                                                          Volume 23– No.9, June 2011
REFERENCES
                                                          [8] An unequal cluster-based routing protocol in
[1 ] Using hierarchical agglomerative clustering in       wireless sensor networks - Guihai Chen · Chengfa Li
wireless sensor networks:An energy-efficient and           · Mao Ye · JieWu - C Springer Science + Business
flexible approach -Chung-Horng Lung , Chenjuan             Media, LLC 2007
Zhou - Elsevier -Ad Hoc Networks 8 (2010) 328–344
                                                          [9] DATA GATHERING ALGORITHMS FOR
[2] A QoS Based Routing Protocol for Wireless             WIRELESS SENSOR NETWORKS: A SURVEY -
Sensor Networks - Mirela Fonoage, Mihaela Cardei,         K.Ramanan and E.Baburaj - International Journal of
and Arny Ambrose -IEEE TRANSACTIONS ON                    Ad hoc, Sensor & Ubiquitous Computing (IJASUC)
MOBILE COMPUTING, VOL. 10, NO. 5, MAY                     Vol.1, No.4, December 2010
2011
                                                           [10] An Adaptive Traffic Aware Data Aggregation
[3] A Sensor Network Data Aggregation Technique -         Technique for Wireless Sensor Networks - M.Y.
Mohamed Watfa, William Daher and                          Mohamed Yacoab ,V. Sundaram - American Journal
Hisham Al -International Journal of Computer              of Scientific Research ISSN 1450-223X Issue 10
Theory and Engineering, Vol. 1, No.1, April 2009          (2010), pp. 64-77

[4] An Multi-hop Cluster Based Routing Protocol for       [11] http://iitkgp.vlab.co.in/
Wireless Sensor Networks - Qi Yang, Yuxiang
Zhuang, Hui Li - Journal of Convergence                   [12] http://www.isi.edu/nsnam/ns/
Information Technology, Vol 6, No. 3. March 2011

[5] Wireless Sensor Network: A Review on Data
Aggregation - Kiran Maraiya, Kamal Kant,
NitinGupta - International Journal of Scientific &
Engineering Research Volume 2, Issue 4, April -2011




Department of CSE, Sun College of Engineering and Technology

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:6
posted:7/26/2012
language:English
pages:6