Tracking Evacuation of Pedestrians during Disastersgsolmaz/publications/GC15-Solmaz... · Gürkan...

Tracking Evacuation of Pedestriansduring Disasters

Gürkan Solmaz and Damla Turgut

Networking and Mobile Computing LabDepartment of Computer Science

University of Central Florida, Orlando, FL

December 9, 2015

Gürkan Solmaz, Damla Turgut (UCF) IEEE GLOBECOM’15 December 9, 2015 1 / 17

Motivation

Increasing use of mobile devices

Application scenario of theme parksI Theme parks in OrlandoI Crowded areas (e.g., city squares, airports)


Tracking pedestrians and emergent events

Internet requires certain infrastructure that can be damaged

Networks resilient to disasters

Communication and safe evacuation in disaster areas

ContributionsI A networked system with mobile sinks

I Sink placement and mobility approaches


Network model

Network consists of sensor nodesand mobile sinks

Sensor nodes:I Store, carry, and transmit the

messages

Mobile sinks:I Personal electric vehicle

(e.g., Segway)I Exist in limited numbersI Able to move to emergent events

Routing protocol: Epidemic routing


Routing protocol

Sink A

P2

P1Initiation

RequestMessages

Sensors act as either initiator or replierMobile sinks always act as repliers


Initial sink placement

Process starts with the creation of a grid layout

The small-sized grids are located only on top of the roads

Base points: The set of quasi-random points with the highestdistance sum

Base points are used for marking grids with particular indices

Each mobile sink is placed on a random point in the subset ofgrids


Sink mobility - Physical force based (PF)

Sink A

P1

P2

F1

Sink B

F2

FB

VA

Inspired by Newton’s law of universal gravitationA pedestrian has a unit mass which attracts the mobile sinksMobile sinks have larger masses causing inverse forces


Sink mobility - Grid allocation based (GA)

Each mobile sink allocates a subset of grids (by indices)Iteratively chooses a random waypoint as the next destinationAims to divide the workload evenly without intercepting


Sink mobility - Road allocation based (RA)

Each sink allocates one or multiple roadsGrid indices are used for allocating neighbor roadsSinks decide random waypoints on top of the roads


Simulation studyOur group’s disaster area mobility model† is used for pedestriansMetrics: Intercontact times, recontact rates, number oftransmissions, rescue success ratioTwo mobility models: Random waypoint distribution (RWD),random target location (RTL)

Table: Simulation parameterssimulation time 2000 ssampling time 2.0 sdisaster area size (≈) 800x800 mnumber of sensor nodes 200sensing range 20 msensor message storage capacity 100transmission probability 0.9grid width/height 50 mnumber of effected people 20rescue failure time 600 ssink relative mass constant(λ) 0.5physical force impact factor α 20.0sink max speed 1 m/spedestrian max speed 1 m/spedestrian visibility 50 m

†G. Solmaz and D. Turgut, “Pedestrian mobility in theme park disasters.” In IEEE Communications Magazine, vol. 53, no. 7,pp.172-177, July 2015.


Processed maps - Magic Kindgom

ru

pS A704-L A704-L

L-4

07A

Main Street

Carolwood Park

Frontierland

Magic Kingdom

The H

u

b

te

ert

S ni

aM

evir

D dlr

oW

yaW seitilicaF

Floridian Way

Flor

idia

n W

ay

Walt Disney World Railroad

Magic Kingdom Drive

evir

D dlr

oW

World D

rive

Caribbean W

ayyaW liaronoM

etuoR edaraP

dnalrietnorF

Tom

orro

wla

nd Indy Speedw

ayMagic Kingdom

TheNew

FantasylandNow

Open! DumboThe

FlyingElephant

First Aid

LibertyBell

Harper'sMill

StoryTimeSpot

JungleCruise

Cinderella'sCastleMoat

Mickey'sPhilharmagic

Monsters,Inc.

LaughFloor

Stitch'sGreat

Escape

It'sa SmallWorld

TheHauntedMansion

Tom Sawyer Island

MadTeaParty

SpaceMountain

Tom Sawyer Island

SplashMountain

Piratesofthe

Caribbean

The map of the Magic Kingdom extracted from (†OSM) and theprocessed map with 1300 waypoints

†M. Haklay and P. Weber, “OpenStreetMap: User-generated street maps.” Pervasive Computing, vol. 7, no. 4, pp. 12-18, Dec.2008.


Results: Intercontact times by sink mobility

PF GA RA RTL RWD0

100

200

300

400

500

Sink Mobility

Inte

rcon

tact

tim

e (s

)

I PF: Physical force basedI GA: Grid allocation basedI RA: Road allocation basedI RTL: Random target locationI RWD: Random waypoint distribution


Results: Intercontact times by transmission range

10 25 50 1000

0.5

1

1.5

2

2.5x 10

4

Transmission range (m)

Inte

rcon

tact

tim

e (s

)

PFGARARTLRWD



Results: Recontact rates by number of sinks

0 2 4 6 8 100

1

2

3

4

5

6

7

8

9

10

Number of mobile sinks

Ave

rage

rec

onta

ct r

ate

PFGARARTLRWD



Results: Number of transmissions by range

10 25 50 1000

50

100

150

200

250

300

350

400

450

500

Transmission range (m)

Num

ber

of tr

ansm

issi

ons

PFGARARTLRWD



Results: Rescue success ratio by number of sinks

0 2 4 6 8 100

10

20

30

40

50

60

70

80

90

100

Number of mobile sinks

Res

cue

succ

ess

ratio

(%

)

PFGARARTLRWD



Thank you

Email: [email protected]

Webpage: http://cs.ucf.edu/∼gsolmaz


[email protected]

http://www.cs.ucf.edu/~gsolmaz

Tracking Evacuation of Pedestrians during Disastersgsolmaz/publications/GC15-Solmaz... · Gürkan...

Documents

Transcript of Tracking Evacuation of Pedestrians during Disastersgsolmaz/publications/GC15-Solmaz... · Gürkan...