Lightweight Clustering Schem

e for Disaster Relief Wireless

Sensor Network

Hanshin-Awaji earthquake survival rates versus rescue time

Contents
Introduction
Lightweight Clustering scheme
Performance evaluation
Experimental results
Conclusion

Introduction
Lifetime maximization is very important when designing
WSN especially in disaster relief WSNs.
Several studies on reducing battery power consumption
and extending network lifetime
Sensor node with sleep Mode : SMAC, TMAC, DMAC
Not practical for rescue operations

Clustering: LEACH, LEACH-C, WCA, HEED..

So complicated.

Need a lightweight clustering scheme with a simple algo

rithm for rescue operations.

LiCS did not require exact location information of sensor

nodes
No need localization module result in low cost of sensor node a
nd no energy for localization module

Having no complex localization algorithms

Efficient and effective way for maximizing the lifetime of
a network and the number of communicable sensor nod
es.

Lightweight Clustering scheme

Network and system model
All sensor nodes are stationary an
d uniformly distributed in a field
Only one SINK node in the wireless
sensor network
not have an energy constraint

Cluster head cannot change its st

ate to sleep mode
Cluster heads are elected by a tim
e interval
Initially, each sensor node has the
same energy capacity

Lightweight Clustering scheme

Cluster Head Election Algorithm

Lightweight Clustering scheme

The target of this clustering scheme is a wireless sensor networ

k that elects cluster heads autonomously.
At the first round
Cluster head election depends on the sensor node starting time.
At the start of a round each sensor node calculates its cluster he
ad announcement (CHA) time.
The CHA time is decide by the amount of energy remaining in the se
nsor node.

Low remaining energy will have a later CHA time.

10

Earlier CHA time become cluster heads, and broadcast CHA

Sensor nodes check any cluster heads around before announci
ng they are cluster heads.
Any cluster heads among their neighbor nodes
do not announce they are cluster heads, and join the nearest cluster
as a cluster member.

Sensor node that has become a cluster member checks for new cl
uster heads until an election phase timeout.

11

Changing their cluster head to the new one has stronger RSSI
No cluster head among its neighbor nodes, a sensor node annou
nces that it is a cluster head.

After a round, sensor nodes will start a new election phase and
elect cluster heads of the wireless sensor network.

Performance evaluation
Simulation Outline and Parameters
Energy consumption of the wireless sensor node

Energy consumption of a cluster head

Energy consumption of a cluster member

Energy consumption of sensor nodes without a cluster head

12

13

14

Energy consumption per second of a cluster head in the data collectio

n phase

Energy consumption per second of a cluster head in the cluster head

election phase

Energy consumption per second of a cluster member in the data colle

ction phase

Energy consumption per second of a cluster member in the cluster h

ead election phase

Performance evaluation
Simulation parameters

Simulation topology

15

Experimental results
Number of communicable nodes

16

17

Performance of LiCS was better than that of LEACH

LiCS has the same drop in the numbers of communicabl
e nodes as HEED
Low density network reduces the effectiveness of LEAC
H, LiCS and HEED
High-density point
LEACH: increases the energy consumption of sensor nodes.
LiCS and HEED: need more cluster heads were needed makes t
he node lifetime short.

Experimental results
Number of CHA messages

18

19

LiCS and Heed have the same lifetime.

LiCS cover all sensor nodes with fewer CHs than HEED.
Transmission frequency of the CHA messages in the election ph
ase decreases
A lighter selection algorithm resulted in LiCS having fewer CH
A messages than HEED.

Conclusion

20

A simple clustering scheme to increase the lifetime of wirel

ess sensor networks and maximize the number of commu
nicable sensor nodes.
The performance of LiCS with a dynamic communication r
ange caused by a radio irregularity will be evaluated in fut
ure.