Sensor Deployment and Target Localization in Distributed Sensor Networks

Yi Zou and Krishnendu ChakrabartyACM Transactions on Embedded Computing Systems 2003

Speaker ： Chen-Chi Hsieh

Outline

Introduction Virtual force algorithm Target localization Simulation results Conclusions

Introduction － Motivations

Motivations Distributed sensor networks (DSNs) are

important for strategic applications Target detection, surveillance, and localization

The effectiveness of DSNs is determined to the coverage provided by the sensor deployment

The positioning of sensors affects coverage, communication cost, and resource management

Introduction (cont.) － Motivations

A random placement of sensors in the target area is often desirable Especially if no a prior knowledge of the

terrain is available However, random deployment does not

always lead to effective coverage

Introduction (cont.) － Objectives

Objectives Maximize the coverage for a given number of

sensors within a cluster in cluster-based DSNs Propose an energy-conserving method for

novel target localization

Introduction (cont.) － Key ideas

Key ideas in this paper Coverage

A random deployment can be improved using a force-directed algorithm

Virtual force algorithm (VFA) Target localization

Is based on a two-step communication protocol between the cluster head and the sensors within the cluster

Virtual force algorithm － Environment

Environment ： for a cluster-based sensor network architecture All sensor nodes are able to communicate with the

cluster head

The cluster head is responsible for executing the VFA algo. and managing the one-time movement of sensors to the desired locations

Sensors only send a yes/no notification message to the cluster head when a target is detected

Virtual force algorithm (cont.)－ The virtual force ideas

The virtual force ideas Each sensor behaves as a “source of force” for all

other sensors

S1

S2S3

S1

S2S3

dth

Virtual force algorithm (cont.)－ The virtual force ideas

The virtual force ideas Each sensor behaves as a “source of force” for all

other sensors

X

Y

S2

S3

S4

S1

12FPositive force

(attractive force)

Negative force(repulsive force)

13F

Virtual force algorithm (cont.)－ The virtual force ideas

Virtual Force calculation in the VFA algo. ： the vector exerted on Si by another sensor Sj

Obstacles and areas of preferential coverage also have forces acting on Si

： the total (attractive) force on Si due to preferential coverage areas

： the total (repulsive) force on Si due to obstacles

The total force on Si

ijF

iRF

iAF

iF

k

ijjiAiRiji FFFF

,1

Virtual force algorithm (cont.)－ The virtual force ideas

Express between Si and Sj in polar coordinate notation

dth ： the threshold distance

αij ： the angle of a line segment from Si to Sj

wA (wR) ： the attractive (repulsive) force

ijF

otherwiseifd

w

ddif

ddifddw

F

ijij

R

thij

thijijthijA

i

),1

(

0

)),((

Virtual force algorithm (cont.)－ Assumptions

Assumptions An n by m sensor field grid There are k sensors deployed in the random

deployment stage r ： detection range of a sensor Sensor Si is deployed at point (xi, yi)

d(Si, P) is the distance between Si and P, for any point P at (x, y)

22 )()(),( yyxxPSd iii

Virtual force algorithm (cont.)－ Coverage

The coverage cxy(Si) of a grid point P(x,y) by sensor Si

The binary sensor detection model

otherwise

rPSdifSc i

ixy ,0

),(,1)(

Sir

d(Si, P) < r

P

d(Si, P) ≧ r

P

Virtual force algorithm (cont.)－ Coverage

The probabilistic sensor detection model In reality, sensor detections are imprecise

re is the uncertainty in sensor detection

a = d (Si,P) － d (r － re)

λand β are parameters that measure detection probability

),(,1

),(,

),(,0

)(

PSdrrif

rrPSdrrife

PSdrrif

Sc

ie

eiea

ie

ixy

re

re

rP 1

Si

e － λaβ

Si

0Si

Virtual force algorithm (cont.)－ Coverage

cxy(Si,Sj) ： the probability that a target is detected by two sensors (overlapped)

A region which is overlapped by kov sensors

))(1))((1(1),( ,,, jyxiyxjiyx ScScSSc

)))((1(1)( ,, iyxovyx ScSc

Si SjP

Virtual force algorithm (cont.)－ Energy Constraint on the VFA Algorithm

dmax ： the max. distance that each node can move in repositioning phase

Virtual force algorithm (cont.)－ Procedural description of the VFA algorithm

Virtual force algorithm (cont.)－ Procedural description of the VFA algorithm

Target localization－ Detection Probability Table

The cluster head generates a detection probability table for each grid point Contains all possible detection reports from sensors

that can detect a target at this grid point P

Sxy ： a grid point P(x,y) is covered by a set of kxy sensors

pxy(Sj, i) ： If Sj detects a target ： pxy(Sj, i) ＝ cxy(Sj)

otherwise ： pxy(Sj, i) ＝ 1 － cxy(Sj)

xyj SS

jxyxy iSpitablep ),()(_

Target localization－ Detection Probability Table

Target localization－ Detection Probability Table

Simulation Results

Environment 50×50 sensor field A total of 20 sensors in the sensor field in

random placement stage Each sensor

a detection radius ： r ＝ 5 Range detection error ： re ＝ 3

The simulation is done on a Pentium III 1.0GHz PC using Matlab

Simulation Results－ binary sensor detection model

Simulation Results－ binary sensor detection model

Simulation Results－ probability sensor detection model

Simulation Results－ probability sensor detection model

Simulation Results－ with obstacles and preferred areas

Simulation Results－ with obstacles and preferred areas

Conclusion

The virtual force algorithm (VFA) Uses a force-directed approach to improve the

coverage after initial random deployment Advantages

Negligible computation time One-time repositioning of sensors Flexibility