You are on page 1of 3

2012 Sixth International Conference on Internet Computing for Science and Engineering

A Sensor Network Topology Controlling Algorithm


with Balancing Energy Consumption

Guomin Wu 1, Xiufang Xu1, Sen Xu1, Xiaoxia Wan2


1
School of Information Engineering
Yancheng Institute of Technology, Yancheng, China
2
School of Information Science and Technology
Yancheng Teachers’ University, Yancheng , China

Abstract—Geographical adaptive fidelity (GAF) energy. As to the problem, traditional single hop
algorithm is analyzed theoretically to point out some communication is transformed to others for node energy
reasons for node energy dissipation imbalance. This consumption balance, which is mixed single hop with multi
paper presents some corresponding solutions. Node hop data communication.
residual energy is taken as one of factors to elect cluster In order to balance node energy dissipation, some
head, in case that some nodes with low energy will be improvements are given for GAF algorithm. In the cluster
dead soon for being cluster head. Data communication is head selection epoch, node residual energy is regarded as one
done by means of single-hop or multi-hop. So some factor to select cluster head. During data transmission,
cluster heads get less chance to consume energy soon, network communication is done by mixing single hop and
which is far away with sink node. Simulation results multi hop. So improved IGAF algorithm can balance node
show that improved geographical adaptive fidelity (IGAF) energy dissipation and prolong network lifetime.
algorithm can improve energy consumption balance and
prolong network lifetime, compared with exiting GAF II. GAF ALGORITHM
related algorithms. GAF is topology controlling algorithm as geographic
adaptive fidelity, which network is divided into several
Keywords-sensor network; topology control; cluster virtual cells. Cluster member can sleep selectively
geographical adaptive fidelity; cluster election; network life each cell.

I. INTRODUCTION Running GAF consists of two stages. In the first stage,


network area is separated from many virtual cells by node
As key technology in the sensor network, optimization of location and communication radius on condition that any
topology controlling can prolong network lifetime, improve nodes in neighbored cells can communicate directly each
communication efficiency, and provide higher service. other. In the second stage, cluster heads are selected in the
Researching work is carried out in two aspects: power different cells. Cluster member nodes periodically change
controlling and hierarchical topology structure. Among their working status alternatively: sleeping and working
topology controlling algorithms, GAF is one typical status. After node awake, it can exchange some information
self-organized clustered algorithm, in which cluster head with other node in the same cell, which decides whether it
selection is affected by geographic position [1]. will become cluster head. Stable data transmission is
Nowadays, many scholars did a lot of research work in following.
the researching field. In the literature [2], one hierarchical
topology controlling algorithm is proposed for clustered III. FEATURE ANALYSIS
sensor network, which can balance node energy dissipation As typical cell clustered algorithm, GAF is simple, easy
and prolong network lifetime by rotating cluster head. In the to realize for distributed application, which is suitable for
literature [3-5], node residual energy is regarded as one large scale network. Meanwhile, some shortage exist, the
factor to elect cluster head node for cluster head locating probability of being selected head is same for all nodes.
unbalance. Selected cluster heads are more applicable for Node residual energy is not considered. Cluster member is
data transmission. So network energy is consumed with more selected head at random, which causes some cluster heads
balance. In the literature [6], some cluster heads are far away die out easily for too much data processing. Besides, data
from Sink node, which are easy to consume too much communication is done by single hop way. So those cluster

978-0-7695-4705-3/12 $26.00 © 2012 IEEE 208


DOI 10.1109/ICICSE.2012.32
heads are easy to die for consume too much energy, which wireless communication model. The related equation is
are far away from Sink node. following:
IV. RALATIVE SOLUTION
Etr (k , d )  Eelec (k )  E amp (k , d )
As to above existing shortages, some improvements are

following. kEelec  k  fs d 2 d d 0
 kE  k d 4 d d
A. Considering Node residual energy as dicisive factor for  elec  mp 0

cluster head election Eelec denotes node energy dissipation unit from
transmitting or receiving circuit, ¦-fs is denoted as energy
In the GAF algorithm, main scheme for cluster head
election is following. In the head election epoch, random dissipation parameter in free space mode, ¦-mp is denoted
number is given for each node, which ranges 0 and 1. If energy dissipation parameter in multi path attenuation
random number below threshold T(n), relative node is mode. Its unit is J/ (bit.m4).d denotes communication
selected as cluster head. The equation about T(n) is distance. d0 denotes critical communication distance. In the
following: mode, data is transmitted from cluster head to base station.
When d below d0, single hop way is taken for data
 p transmission. Otherwise, multi path way is taken.
 nG (1)
T ( n)  1  p[r mod(1 / p )]
0 V. IMPROVED ALOGORITHM PROCEDURE
 else
Among these parameters, p denotes node percentage, r is Based on GAF algorithm, two aspects have been
network passed round. G is denoted as unelected node set in improved. Main procedure is following.
the past 1/p rounds. Improved IGAF algorithm takes node a) In the monitoring area, nodes are deployed at
residual energy as election factor. Value of T(n) is changed random.The whole network area is divided into several
as following: subarea.
 p b) In the cluster head selection, node residual energy is
 * T ' ( n) n G
T (n)  1  p[r mod(1 / p)] taken as one factor for dicision. N nodes transmit and receive
0 message in sequence.
 else
and
c) All the nodes run with communcation module in the
different cells. Message (E, ID) is sent to inform node which

En _ residual 1  En _ residual
T (n)'   rs div 1  (2) numbered n.
En _ max
p
En _ max

d) When all the nodes get ID and residual energy from


In the equation (2), En _ residual denotes node residual node, the node becomes cluster head informly.
energy, En _ max denotes maximal residual energy among the e) The cluster head send some message to surrounding
nodes, also the initial energy, r is passed round. nodes ,which announces cluster head itself.
Considering effect from node residual energy and f) All the cluster members join their own cluster.
threshold, some nodes with low energy can avoid early death g) During data transmission,the distance between cluster
as cluster head. head and base station is computed, which is noted as
B. Mixed communication scheme min_dis.
In the GAF algorithm, single hop is taken as h) If min_dis<= d0, single hop way is taken.Otherwise
communication way between cluster head and base station. multi hop way is taken.
In the case, all cluster heads are divided into two parts. The
first one is near to base station. The second one is far away i) If network lifetime ending, exit. Otherwise, jump to
from base station. The second one will consume more energy step b.
than the first with transmitting same amount of data. The VI. SIMULATION VALIDATION
second is easy to die out for consume more energy. With
network running, node residual energy shows to be A. Senario
unbalanced. If nodes with below average energy become to One hundred nodes are deployed in the 100*100
be cluster head, these nodes become blind node instantly. monitroing area randomly. Whole area is divided into four
With blind node increasing, network lifetime will be vitual cells. Each node has its own ID and same initial
affected. energy in the cells. Only one sink node is located in the
To balance energy dissipation among different clusters, center of monitoring area, which position is (50,50).Packet
mixed communication way is taken, which range from single size is 4000bit for data transmission. Controlling packet size
hop and multi hop. Scholar Heinzelman gives one simple is 32bit as cluster head to broad message. Running process is
following:

209
With network running round increasing, all four curves
show to be decreasing, which shows that network residual
energy decreases continuously when network running ahead.
Compared with GAF, GAF-1and GAF-2, IGAF algorithm
improves a lot in the network residual energy. Besides, IGAF
curve decrease steadily, it can provide higher stability for
network running.
Figure 1. Network running process. VII. CONCLUSION
In the Figure 1, o represents ordinary node. Node with Feature is analyzed for GAF algorithm. Two aspects are
star represents cluster head. Node with black o represents proposed. Node residual energy is taken as one factor for
dead node. There is only one cluster head in each cell. In the cluster head selection. And mixed communication way is
network running process, cluster member send some proposed for data transmission, which includes single hop
messages to cluster head. Then cluster head transmit these and multi-hop. Improved IGAF algorithm can better balance
messages to sink node. network energy dissipation, decrease the number of dead
nodes in the same running round, and prolong network
In the experimental scenario, network performance is lifetime. Future work is focused on cluster head election in
evaluated from two aspects: the number of all live nodes, the optimum way.
network residual energy.
ACKNOWLEDGMENT
B. The number of all live nodes
In the same running round, the number for all live nodes This work was partly funded by the National Natural
are compared with four algorithms in Figure 2. Science Foundation of China under Grant No. 60975042 and
61105057, Natural Science Foundation researching project in
Yancheng Institute of Technology No. XKY2011035, and
Talent Introduction Project in Yancheng Institute of
Technology under Grant No. XKR2011019.
REFERENCES
[1] I. F. Akiyldiz, W. Su, Y. Sankarasubramaniam and E. Cayirci,
Ā A survey on sensor networks, ā IEEE Communication
Magazine, 2002, 4(8):102-114.
[2] Wendi.B Heinzelman, P.Chandrakasan, and Hari
Balakrishnan, “An Application-specific Protocol Architecture
for Wireless Microsensor Networks,” IEEE Transactions on
Figure 2. The number of all live nodes. Wireless Communications, 2002, 1(10):660-667.
With network running round increasing, all four curves [3] Ossama, Younis Sonia, and Fahmy, “HEED:Ahybrid,energy-
show to be decent obviously. From other aspect, the numbers efficient, distributed clustering approach for Ad hoc sensor
for all live nodes decrease continuously. Compared to GAF, networks,” IEEE Transactions on Mobile Computing, 2004,
GAF-1and GAF-2, IGAF algorithm owns highest node 3(3):660-669.
living rate among them. With running round increasing,
[4] N. Dimokas, D. Katsaros, and Y. Manolopoulos,
improved result shows to be more obvious. So improved “Energy-efficient distributed clustering in wireless sensor
IGAF algorithm can further reduce the number of network networks,” Journal of Parallel and Distributed of Computing,
dead node, and prolong network lifetime. 2010, 70(4):371-383.
C. Network residual energy [5] Handy M. J, Haase M,and Timmermann D, “Low Energy
Adaptive Clustering Hierarchy with Deterministic
Cluster-head Selection,” in Proceedings of the 4th IEEE Conf.
on Mobile and Wireless CommunicationsNetworks:IEEE
Communications Society, 2002, 10(11):368-372.
[6] Heinzelman,W.B,Chandrakasan, A.P, and Balakrishnan,H,
“Energy- efficient Communication Protocol for Wireless
Microsensor Networks,” in Proceedings of the 33rd Hawaii
International Conference on System Science.Hawaii,USA,
2000, 2(8):1-10.

Figure 3. Network residual energy.

210

You might also like