Redeployment for Mobile Wireless Sensor Networks Weihong Fan, Hengyang Zhang and Xuanping Cai Yunhui...

Redeployment for

Mobile Wireless Sensor Networks

Weihong Fan, Hengyang Zhang and Xuanpi

ng CaiYunhui Liu

Yunhui LiuJoint Center of Intelligent Sensing and Systems

National University of Defense Technology

Department of Automation and Computer-Aided Engineering

The Chinese University of Hong Kong

Proceedings of the 2008 IEEE International Conference on Information and Automation

IEEE ICIA 2008

Outline Introduction Problem Description DFS_LF Algorithm

(Depth-First-Search_Leader-Follower) RHR_C Algorithm

(Right-Hand-Rule_Centroid) Simulation Conclusion

Introduction

The wireless sensor networks Application

Environment surveillance Battlefield search/rescue

Characteristic A large amount of sensor node A large range of area

The sensors are deployed randomly

Introduction

Random deployment Fragmentation phenomena Coverage hole problem

Communication disc

Communication or sensing disc

Related Work

Solve the coverage problem Incremental sensor deployment

ICPSAS 06 Movement assisted sensor deployme

nt Grid coverage algorithm Voronoi diagram algorithm VFA and PFA

Network ModleSurveillance area A

The total number of node : NPoisson distribution Poisson density : λ

N/A → λ0<λ<∞

Network Model

All sensor nodes Mobile Homogeneous Location and orientation aware Rc : communication range Rs : sensing range Rc ≥2Rs

Network Model Communication area

πRc2

Sensing area : ||S||

Coveragemax : max coverage ratio

P(C) : coverage probability

2sRS

A

RN

A

SNCoverage s

2

max

SePCP 1)coverednot 0(1)(

DFS_LF Algorithm

DFS algorithm + LF method DFS algorithm

Select a virtual node as the collective object Flocking for Multi-Agent Dynamic Systems : Algorithms and

Theory IEEE Transaction on automatic control, Vol. 51, No. 3, March 2006

DFS_LF Algorithm

2

1

4

5

3

6

8

79

Subnet A

Subnet Bcollective object

p : the position of collective object

pi : the position of node i

L

L

F1F1

F1

F2

F2

F2

F3

DFS_LF Algorithm

8

79

collective object

p : the position of collective object

pi : the position of node i

If d(p,pi )≤1/2Rc

need not moveElse pi aims and moves to the object

L

2

L

F1F1

1

4

5

3

6

F1

F2

F2

F2

F3

d(pi , pj ) ≤Rc

1/2RcF1

F2

F3

F3

F3

F4

RHR_C Algorithm

In the dense mobile sensor networks

A large Rc is likely to cause the congestion of communications

traffic waste their limited energy

Rc≥2Rs

RHR_C Algorithm If there is a communication hole, there

must be a sensing coverage hole If the communication hole does not exit

Discuss the relationship of communication range and sensing range

Sensing coverage with no overlapping Sensing coverage with overlapping

Rc≥2Rs +ε (ε>0)

RHR_C Algorithm

detection coverage hole

Sensing coverage without overlapping

Sensing coverage without overlapping

Sensing coverage without overlapping

Sensing coverage without overlapping

ε→0 X’(dij)>0rc

0222 ijkijk ddd

X(dij) S1

Sensing coverage with overlapping

i

j

l

k

l‘

αB

k

Rs

Sensing coverage with overlapping

2 2

Sensing coverage with overlapping

Bα Y(α)

sinα>0 sinB-sinα>0 Y’(α)> 0

rc djk dki

1

Result

From the two case The shorter communication distance

between the sensor node and the higher sensing coverage ratio

In the dense network rc≥90.69% The coverage hole does not exits

RHR_C Algorithm

Healing the communication hole

Rs

Simulation The interest area A : 80 x 60 m2 Rs = 2m DFS_LF algorithm

Sensor nodes : 385 Rc = 5m Poisson density λ = 0.08 Coveragemax=1.01 P(C) =0.57

Simulation

RHR_C Algorithm Sensor nodes : 1000 , 1200 Rc = 4.1m Poisson density λ = 0.21 Coveragemax=2.62 P(C) =0.93

Conclusion

This paper propose two deployment algorithm for mobile sensor networksFragmentation in sparse networksCoverage hole in dense networks

Simulation result testify that the two algorithm are validity