Available online at www.sciencedirect.

com

ScienceDirect
IERI Procedia 4 (2013) 288 – 294

2013 Inteernational Conference

C o Electronnic Engineerring and Coomputer Scieence
on

A Noveel Data Gathering

G g Methood for Laarge Wirreless Seensor
Netwoorks
D
Djuraev Maamurjon, Byoungchu
B ul Ahn*
Deparrtment of Computter Engineering, Yeungnam
Y University,Gyeongsan,Gyeonbuk 712-7449, South Korea

Absttract

Wireeless sensor netw works are appliied to collect deesired data usinng wireless radio
o communicatioons but they have limited
life because
b of limitted power usingg batteries. Enerrgy efficient daata collection method
m is very im
mportant to opeerate wireless
sensoor networks forr a long time. Thhis paper propooses a novel datta collection meethod for large wireles
w sensor nnetworks and
ubiquuities applicatioons. A typical data
d collection tourt using a moobile sink starts from a sensor node
n nearby a bbase station,
traveels all nodes of wireless
w sensorr network by vissiting each nodde, returns to thee base sation annd uploads colleected data to
the base
b station. Thiis method is tim
me inefficient annd each node sppends battery power quickly. A novel data coollection
methhod is proposed to increase eneergy efficiency using a mobile sink. The simu ulation result shhows that the prroposed data
colleection method decreases
d the ennergy consumpttion level by proolonging the liffetime of wireleess sensor networks to 15
perceents compared withw LEACH.

© 20013.
©2013 ThePublished d by Elsevier
Authors. Published B.V.B.V. Open access under CC BY-NC-ND license.
by Elsevier
Seleection
Selection andand
peerpee r review
review undeer responsibilit
under responsibility ty ofEngineering
of Information Informattion Engineer
Research ing
Institute Research Institute

Keyw
words: Mobile datta gathering, K-m
means algorithm, clustered
c data gatthering

1. In
ntroduction

G
Generally wireeless sensor neetworks (WSN Ns) are deployed as sensorr nodes and a base station ((sink). Their
appllication areas are to monitoor environmennts or inaccesssible regions, such as battleefields, jungless, or forests.
Nowwadays, the coost of sensor nodes
n has beenn reduced by the
t advances of semiconduuctor technologgy and mass
prodduction. Also nodes of WSN Ns are equippped with expennsive sensors such as videoo, audio, GPS and so on.

2212-6678 © 2013 The Authors. Published by Elsevier B.V. Open access under CC BY-NC-ND license.
Selection and peer review under responsibility of Information Engineering Research Institute
doi:10.1016/j.ieri.2013.11.041
 Djuraev Mamurjon and Byoungchul Ahn / IERI Procedia 4 (2013) 288 – 294 289

*
Sensor nodes on WSNs are operating with limited battery-power. The battery-power is one of important
parts for wireless sensor networks since it is difficult to replace batteries with new ones in large-scale wireless
networks Generally, at sensor networks, gathered data is sent to a fixed base station [1][2]. Since data
transmission and reception consume most of node energy, middle nodes on the data relaying paths consume
more energy when they forward packets to the base station. Especially, a node near to the base station should
relay packets from other nodes and deplete its own energy earlier than other nodes. This is the crucial reason
to shorten the network life time.
Recently, there are some researches for a mobile sink, which doesn’t reside at a fixed location but moves
around whole network area to collect data. Shah has proposed a mobile observer, data Mobile Ubiquitous
local area network Extensions(MULEs) to collect data from sensors directly, save data its own memory, and
upload data to the base station[3]. It is adapted for wide area to move around and monitor the area. For
applications of battle field, traffic monitoring, wild animal research, and pollution control, data gathering
point is not fixed but changed depending on the situation. For these cases, a receiver sensor could be attached
to a mobile robot, a vehicle, an animal, or a human. And it requests data and collects data from different types
of mobile infrastructure.
Since a mobile sink is an external device to WSNs, the power consumption of for the movement and
operation of the mobile sink doesn’t affect the lifetime of WSNs. Periodically the mobile sink returns to the
base station to upload collected data to process data and recharge its battery-power to make next travel. The
mobile sink gets data from sensor nodes at some predefined locations. This data collection scheme is called
MDG (mobile data gathering) and gathering position is called polling point [4]. The mobile sink collects just
once from each node by traversing sensor networks, and this trip is called round. The major advantage of the
mobile sink method is to reduce the energy consumption of nodes in WSNs is saved since it can approach
near sensors to collect data directly or through short hops.
This paper proposes an energy efficient data collection method using a mobile sink and clustering sensors
for large wireless sensor network. Before collecting data, a mobile sink clusters sensor network using K-
means [18] algorithm and calculates traveling. The greedy algorithm is used to calculate traveling paths of the
mobile sink. When the mobile sink arrives to a cluster, it broadcasts message to cluster members. The
message contains the identification number of each node in the cluster. The node which has received
broadcast message and matches it’s ID, replies its collected data to the mobile sink.
The disadvantage of the mobile sink is that data gathering time is longer than a stationary sink because it
should travel entire wireless network area. There is a possibility to lose data if the mobile sink does not visit at
the right time. Time-sensitive applications, which require fast gathering capability, is required to reduce the
number of clusters even if it causes high energy consumption. If the transmission radius is expanded,
forwarding hops are reduced but the energy consumption of nodes will be increased. There exists a trade-off
between traveling time and its network lifetime.

2. Related work

LEACH proposes for a stationary sink with the clustering protocol for applications which gather data
periodically [5]. In LEACH, a node becomes a head node by itself based on the probability. LEACH uses the
balanced energy consumption strategy of nodes. For one-hop transmission to stationary sink, the head node
needs the high transmission power because of the distance. If the head node sends data through multi-hop to
the sink, unbalanced energy consumption of sink-nearby sensors is inevitable.
Recently the mobility of wireless sensor networks has been conducted in many papers. Juang has used
*
* Corresponding author*. Tel.: +8-253-810-2556; fax: +8-253-810-4630.
E-mail address: b.ahn@ynu.ac.kr
290 Djuraev Mamurjon and Byoungchul Ahn / IERI Procedia 4 (2013) 288 – 294

mobile nodes with a radio-tagged to collect sensed data of animals such as zebras and whales [6]. These nodes
attached to animals move around randomly in the field. When animals move close to each other and sensed
data are exchanged. Shah and et. el. present that mobile nodes called MULEs(data Mobile Ubiquitous Local
area Extensions. It collects sensed data from sensor nodes in short distance and save them in its memory. It
uploads collected data of all nodes to the base station [3]. Shah and et. el. implement the movements of
MULEs by modelling as 2 dimensional random walks. However, one drawback of this approach is the latency
of data collection because MULEs has to travel each sensor node to collect sensed data. Kansal et al. proposes
a mobile sink idea. The mobile sink moves along on the designated line and collects sensed data of nodes [7].
This idea is expanded to applications. Several mobile sinks moves on the designated lines in parallel and
collect sensed data from sensor nodes in parallel [8]. In order to reduce data collection latency, the mobile
sinks receive packets relayed from some wireless sensor nodes, which cannot be communicated directly.
Sencar proposes the method for optimizing the trajectory of sink [9]. It minimizes the energy consumption by
building the trajectory where nodes can reach to the mobile sink by load balanced hops.

3. Proposed method

In this section, we introduce novel data gathering method based on clustering method and a mobile sink.
The proposed method uses clustering concept to get energy efficiency. The mobile sink visits every cluster
and gets sensed data of wireless sensor nodes. After it finishes data collection, it returns to a base station and
uploads collected data to the base station.

3.1. Clustering phase

At the beginning of the proposed method a mobile sink divides the sensor network into clusters. To
perform clustering, K-means clustering algorithm is applied. For large WSNs, this algorithm is
computationally faster than hierarchical clustering and also produce tighter cluster, especially if clusters are
globular.
3.1.1. K-means algorithm
The number of clusters is predefined using the K-means clustering algorithm and clustering process is
divided into 4 steps below.
Step 1. If WSNs are divided into ‘k’ clusters, k clusters are defined in advance at random places. These
random places are the center of each cluster.
Step 2. Calculate the Euclidian distance from each node to the center of its cluster. Attach it to nearest
cluster center. Through this calculation ‘k’ initial clusters are built. Assume that, there are n nodes taken such
that each one of them relates to Rd. To find the minimum difference clustering of these nodes into k clusters,
the centers of k clusters are searched in { mj }k j = 1 in Rd such that,
ଵ
ቀ ቁ ‫ ݔ‬σ ቀ݉݅݊ ݀ ଶ ൫ܺ௜ ݉௝ ൯ቁ ǡ for i = 1 to n, (1)
௡
k
where d (Xi , mj) is the Euclidean distance of two points Xi and mj. The points {j} i=1 are known as the
center of clusters.
Step 3. Calculate again each cluster center position and check for the position change.
Step 4. If there is no difference in position of any center of cluster, the clusters are selected and the
clustering process finishes. Otherwise repeat Step 2.
The mobile sink knows locations of all sensor nodes using GPS or position estimating algorithms. The
mobile sink performs clustering algorithm and saves center location of each cluster. The mobile sink will visit
center point of each cluster. Based on these locations, the mobile sink calculates its visiting points to gain
 Djuraev Mamurjon and Byoungchul Ahn / IERI Procedia 4 (2013) 288 – 294 291

optimal total travelling time using heuristic Travel Salesman Problem (TSP). After clustering sensor networks
and calculating TSP, the mobile sink moves along predefined position by TSP.

3.2. Data gathering

The mobile sink travels entire sensor networks in trajectory which was calculated before starting its data
collection travel. As soon as, the mobile sink gets its predefined position, which is cluster center, it broadcasts
a message to cluster members and waits for an acknowledge message. The broadcast message contains the ID
number of cluster members. Sensor nodes receive a broadcast message and check the message. If sensor
nodes find their node ID in the broadcast message, they reply back by sending an acknowledge message to the
mobile sink and becomes member node. Otherwise the sensor node ignores the broadcast message. As soon as
the mobile sink finishes data gathering process from the cluster members and it continues its travel. Next data
collection point will be a following cluster calculated by TSP. After visiting all clusters, the mobile sink stops
to the base station and uploads its collected data to the base station. In Fig. 1, the mobile sink travels the
center of each cluster.

Fig. 1. After clustering WSNs and data collection points of the mobile sink

4. Performance evaluation

To evaluate the proposed method, simulation NS2 is used. Wireless sensor nodes are uniformly distributed
in the field of 100m x100m area. The location information of sensors is assumed to be obtained during the
network deployment phase as Sencar[9]. Table 1 shows the simulation parameters to evaluate the
performance. Each node has the initial energy of 2 Joules.
292 Djuraev Mamurjon and Byoungchul Ahn / IERI Procedia 4 (2013) 288 – 294

Table 1. Simulation parameters

Parameter Value
Network dimension from(0,0) to (100,100)
Number of sensors 50
Start point (0, 0)
The speed of Mobile sink 10 m/s
Einit 2 Joules

Fig. 2. Power dissipation of sensor networks

For comparison, we used LEACH algorithm in the simulation. The simulation results show that the
proposed method performs more energy efficiency than LEACH. In LEACH algorithm, the cluster head of
every cluster spends much more energy to transmit gathered data to the base station. It might decrease
network lifetime. But the proposed method extends the use of power consumption. Also using TSP algorithm
reduces total travelling time. Fig. 2 shows the average network lifetime of WSNs by time using the proposed
method and LEACH. In comparison, the proposed method extends network lifetime of WSNs. In LEACH
network stops at 23th round since some nodes exhaust the battery-power, the proposed method extends the
network lifetime to four more rounds.

5. Conclusion

One of the important factors in WSNs is the network lifetime that is tightly dependent on energy
consumption during data gathering process. Moreover, unbalanced energy consumption impacts on whole
 Djuraev Mamurjon and Byoungchul Ahn / IERI Procedia 4 (2013) 288 – 294 293

network lifetime of WSNs. The novel data gathering method uses a mobile sink to increase fair use of energy
consumption at wireless sensor nodes and to extend the network lifetime of WSNs.
The performance of the proposed method is evaluated and compared with that of LEACH algorithm under
same simulation conditions. The simulation result shows the effectiveness of energy consumption in proposed
method specifically network lifetime increased to about 15% compared with LEACH.

Acknowledgements

This work (00045039) was supported by Business for Cooperative R&D between Industry, Academy, and
Research Institute funded Korea Small and Medium Business Administration in 2011.

References

[1] Manjeshwar A, Agrawal DP, TEEN: a routing protocol for enhanced efficiency in wireless sensor
networks. Parallel and Distributed Processing Symposium. April 2001
[2] Lindsey S, Raghavendra C.S, Data Gathering Algorithms in Sensor Networks Using Energy Metrics.
IEEE Transactions on Parallel and Distributed Systems. Vol. 13, No. 9, Sep 2002.
[3] Shah R. C, Roy S, Jain S, and Brunette W. Data mules: Modeling a three-tier architecture for sparse
sensor networks. IEEE Workshop on Sensor Network Protocols and Applications (SNPA), 2003.
[4] Zhao M, Ma M. and Yang Y. Mobile Data Gathering with Space-Division Multiple Access in Wireless
Sensor Network. 27th IEEE INFOCOM , Phoenix, USA, 2008.
[5] Heinzelman W. B, Chandrakasan A. P, Balakrishnan H. An Application-Specific Protocol Architecture
for Wireless Microsensor Networks. IEEE Transactions on Wireless Communications. Vol. 1, No. 4, October
2002.
[6] Juang P, Oki H, Wang Y, Martonosi M, Peh L. and Rubenstein. Energy-Efficient Computing for Wildlife
Tracking: Design Tradeoffs and Early Experiences with Zebranet. Proc. 10 International Conference
Architectural Support for Programming Languages and Operating Systems (ASPLOS), 2002.
[7] Kansal A, Somasundara A, Jea D, Srivastava M, and Estrin. Intelligent Fluid Infrastructure for Embedded
Networks. Proceding Second ACM International Conference Mobile Systems, Applications and Services
(MobiSys ’04), 2004.
[8] Jea D, Somasundara A.A, and Srivastava M.B. Multiple Controlled Mobile Elements (Data Mules) for
Data Collection in Sensor Networks. Proceding First IEEE/ACM Intenational Conference Distributed
Computing in Sensor Systems (DCOSS ’05), June 2005.
[9] Ma M and Yang Y. Sencar: An energy efficient data gathering mechanism for large scale multihop sensor
networks. In IEEE International Conference on Distributed Computing in Sensor Networks (DCOSS), Vol.
4026, pp. 498-513. Lecture Notes in Computer Science (LNCS), Springer Verlag, 2006.
[10] Intanagonwiwat S. C, Govindan R, and Estrin D. Directed diffusion: A scalable and robust
communication paradigm for sensor networks. 6th ACM/IEEE Annual International Conference on Mobile
Computing (MOBICOM 2000), 2000.
[11] Zhang Z, Ma M and Yang Y. Energy-Efficient Multihop Polling in Clusters of Two-Layered
Heterogeneous Sensor Networks. IEEE Transaction on Computers, Vol. 57, No2, Feb 2008
[12] Saad EM, Awadalla MH, and Darwish RR. Adaptive Energy-Aware Gathering Strategy for Wireless
Sensor Networks. International Journal of Computers, Issue2, Vol. 2, 2008.
[13] Younis O, Fahmy S. HEED: A Hybrid, Energy-Efficient, Distributed Clustering Approach for Ad-Hoc
Sensor Networks. IEEE Trans.Mobile Computing, Vol. 3, No. 4, pp. 366-379, 2004.
294 Djuraev Mamurjon and Byoungchul Ahn / IERI Procedia 4 (2013) 288 – 294

[14] Avin C and Krishnamachari B. The power of choice in random walks: an empirical study. MSWiM '06:
Proceedings of the 9th ACM international symposium on Modeling analysis and simulation of wireless and
mobile systems, pp. 219-228, ACM Press, Oct 2006.
[15] Ettus M. System capacity, latency, and power consumption in multihop-routed SS-CDMA wireless
networks. Proceding Radio and Wireless Conf. (RAWCON), pp. 55–58, Colorado Springs Co, Aug 1998.
[16] Florens C, Franceschetti M, and McEliece RJ. Lower bounds on data collection time in sensory networks.
IEEE Journal on Selected Areas in Communications, Vol. 22, pp.1110– 1120, 2004.
[17] Gupta P and Kumar PR. The capacity of wireless networks. IEEE Transactions on Information Theory,
Vol. 46, No. 2, pp. 388–404, Mar 2000.
[18] Jiawei H, Micheline K, Morgan K. Data Mining: Concepts and Techniques, 2 nd Edition , 2006, pp.
404-407