Yun-Jung Lu 1

Connectivity-Guaranteed and Obstacle-Adaptive Deployment

Schemes for Mobile Sensor Networks

Guang Tan, Stephen A. Jarvis, and Anne-Marie Kermarrec

IEEE Transactions on Mobile Computing, VOL. 8, NO.6, JUNE 2009

Yun-Jung Lu 2

Introduction Preliminaries The Connectivity-Preserved Virtual Force

(CPVF) Scheme The Floor-Based Scheme Performance Evaluation Conclusion

Outline

Yun-Jung Lu 3

In an mobile sensor network, the sensors are able to relocate and self-organize into a network.

The mobility and self-management of sensors are desirable for many application scenarios, including remote harsh fields, disaster areas or toxic urban regions, where manual operations are unsafe or burdensome.

Introduction

Yun-Jung Lu 4

Given a target sensing field with an arbitrary initial sensor distribution, how should these sensors self- organize into a connected ad hoc network that has the maximum coverage, at the cost of a minimum moving distance?

Self-deployment Problem

Yun-Jung Lu 5

Potential Fields or Virtual Force◦ When two electromagnetic particles are too close

in proximity, a repulsive force pushes them apart.

Voronoi Diagrams (VDs)◦ Allow sensors to move to maximize coverage in

its own subarea

A number of proposed scheme to this problem

Yun-Jung Lu 6

The communication range of a sensor may not be large enough to cover all Voronoi neighbors.◦ An incomplete view of the Voronoi neighbors may

result in very inaccurate VDs being constructed.

Several Problems in practice

Yun-Jung Lu 7

Network Connectivity? Network partition can still occur in a dense network. ◦ Generally, connectivity must be considered in

protocol design.

Obstacle-free? ◦ Naturally, the real-world environments have

obstacles or holes render such schemes ineffectual.

The Rest of Problems

Yun-Jung Lu 8

To achieve connectivity for a network with an arbitrary initial distribution, communication/sensing range, or node density

To minimize moving distance, which dominates energy consumption in the deployment process

To be able to work without any knowledge of the field layout, which can be irregular and have obstacles of arbitrary shape

The goals of this paper

Yun-Jung Lu 9

System Assumptions◦ All sensors have the same communication range

rc and sensing range rs.◦ At any given time, a sensor knows its own position

and can recognize the boundary of the obstacles within its sensing range.

◦ Sensors move in steps of variable size. In each step, a sensor moves in a straight line at a

uniform speed for a period and denote by T.◦ There is a reference point O ; all the sensors will

try to connect to O generality.

Preliminaries I

Yun-Jung Lu 10

Obstacle Avoidance◦ BUG2: “Path-Planning Strategies for a Point Mobile

Automaton Moving amidst Unknown Obstacles of Arbitrary Shape,” Algorithmica, 1987

◦ Reference Line: the straight line (Start, Target)◦ H: hitting point◦ Right-hand rule

Preliminaries II

Yun-Jung Lu 11

Lazy Movement (With multiple hop communication, not all disconnected sensors need to move to get connected.)

◦ At the end of each step, a sensor checks its neighbors to see if there are any ahead of it;

◦ If so, then it chooses the nearest neighbor as its candidate path parent.

Preliminaries III

Yun-Jung Lu 12

Achieving Connectivity

Maximizing Sensing Coverage

The Connectivity-Preserved Virtual Force (CPVF) Scheme

Yun-Jung Lu 13

Initially, all sensors are required to decide their states regarding connectivity.◦ Flooding a message to the network

Sensor receives such a message, becomes aware that they are also connected

After a certain period of time, if a sensor still has not received such a message, it can decide that it is disconnected.

◦ It will allow a small random time period to elapse after which it starts to move using the BUG2 Algorithm(with lazy movement) toward the base station.

Achieving Connectivity

Yun-Jung Lu 14

Virtual Force is used to determine the direction to move.◦ The obstacles and neighboring sensors exert

repulsive forces onto a sensor.◦ The sum of all forces determines the subsequent

direction of that sensor.

Maximizing Sensing Coverage

Yun-Jung Lu 15

Connectivity Preserving Conditions◦ The distance between s and s’ at time t’

is no greater than rc

◦ The distance between s’ ’s position at t’ and s’ ’s position at t + T is no grater than rc

A sensor can approximately determine the maximum valid step size by checking a set of possible values, for example, VT, 0.9＊ VT, …, 0.1*VT, 0.

Maximizing Sensing Coverage

A

B

C

fba fca

V : the moving speedT : the moving time of one step

VTAt’

s’

s’

Yun-Jung Lu 16

CPVF

Yun-Jung Lu 17

The Floor-Based Scheme

Yun-Jung Lu 18

Achieving Connectivity

Identifying Movable Sensors

Expanding Coverage

The Floor-Based Scheme

Yun-Jung Lu 19

A High-level View

Yun-Jung Lu 20

Achieving Connectivity

Yun-Jung Lu 21

To identify sensors that can move without partitioning the network and whose move is expected to increase network coverage

The Rules to achieve that:◦ Obtain a list of neighbors within two hops of itself◦ Try to find for each child a new parent◦ Loop check for a particular child◦ If all the children can find parents without crating

loops, then it means that the sensor can safely move away.

Identifying Movable Sensors

Yun-Jung Lu 22

With all movable sensors identified, we can now expand the network’s coverage.

Three types of expansion policy◦ Floor-line-guided expansion

◦ Boundary-line-guided expansion

◦ Interfloor-line-guided expansion

Expanding Sensing Coverage

Yun-Jung Lu 23

Expansion Point Expansion Circle is min(rc, rs)

Floor and Boundary

frontier point

Expansion Circle

Yun-Jung Lu 24

Floor and Boundary

Yun-Jung Lu 25

Interfloor

Frontier Point

26

If a sensor can not find any expansion points in its expansion circle, it will stop the process.

Else, it will flood a Invitation Message to find some sensors to cover these points.

Invitation Message contains an EP to the network and a TTL value.

Inviting Movable Sensors

Yun-Jung Lu

Yun-Jung Lu 27

It collects a certain number of invitations, and picks one with the highest priority.

It sends an AcceptInvitation message to the inviter.

A movable sensor

Yun-Jung Lu 28

The inviting sensor constructs a virtual place-holding fixed node in the tree, and sends a message to the root on behalf of the invited sensor to update the location information maintained by its ancestors.

In the former case,

Yun-Jung Lu 29

Floor-based Scheme

Yun-Jung Lu 30

An event-based simulator using C++ 240 sensors are initially randomly

distributed in a subarea {(x, y):0≦x ≦500m, 0≦y ≦500m} of a target field {(x, y):0≦x ≦1000m, 0≦y ≦1000m}

The base station is located at (0,0). The maximum moving speed is 2 m/s. The period length is 1 second. The simulation runs for 750 seconds.

Performance Evaluation

Yun-Jung Lu 31

Comparison between CPVF, FLOOR, and OPT

Yun-Jung Lu 32

Comparison between FLOOR, VOR, and Minimax

Yun-Jung Lu 33

Moving Distance in Obstacle-Free Fields

Yun-Jung Lu 34

Two sensor deployment schemes are proposed for mobile sensor network in this paper.

The major difference of the proposed schemes with the previous works is their adaptability to arbitrary network densities or communication ranges and to obstacles.

Conclusion