Community-based Hazard Warningsin Rural Sri Lanka:

Performance of Alerting and Notificationin a Last-Mile Relay

Nuwan Waidyanatha, LIRNEasiaGordon A. Gow, University of Alberta

Peter S. Anderson, Simon Fraser University

Background

Last-mile Hazard Warning System, Sri Lanka:

IDRC-funded, Sarvodaya and LIRNEasia lead organizations.

32 participating villages, 150-1000 households

Evaluate five wireless ICTs deployed in various conditions for their suitability in the last-mile of a warning system

Not a “public warning” system

Closed user group of first responders, who alert local communities

Multiple components to the study, including:

Common Alerting Protocol

Social and cultural factors (adoption and use, gender)

Paper reports on a subset of indicators for reliability and effectiveness

5 ICTs Tested for Reliability and Effectiveness in the Last-Mile

CDMA Fixed Phone GSM Mobile Phone Remote Alarm Device

Addressable Radios for Emergency Alerts Very Small Aperture Terminals

LM-HWS Components and their functions

ERP-C(n,m)

ERP-C(n,2)

ERP-C(n,1)

National Early Warning Center

Hazard Information

HubDomestic & International

Sources

ICT

-G(n

)

CO

MM

UN

ITY

(n)

Networks

Staff at the HIH monitor hazard events around-the-clock

When an “Event of Interest” is detected, HIH follows a protocol that may result in the issuance of a message to ICT Guardians over multiple modes

ICT-Gs receive messages and acknowledge receipt

If urgent or high-priority alerts, ICT-Gs notify local ERP-Coordinators in the community using locally agreed on methods

In real cases, ICT-Gs and ERP-Cs will await official warnings

In live tests, ERP-Cs activated local ERP and were timed

Multiple Paths, Multiple Technologies and Multiple Gateways

HIH 203.88.69.241

AsiaStar

Singapore58.185.127.202

Touluse82.225.29.106

Server

Colombo202.69.192.51

UDP/IP Recievers

Monitor

GSM Tower

GSM Devices

Group

Group

Hong Kong203.88.69.241

Admin

Ottawa64.26.169.57 TCP/IP PCs

Group

ColomboPSTN

CDMA Tower

CDMA Phones

Group

CDMA Tower

Server

Server

Server

Singapore203.88.69.241 AsiaSat II

WorldSpace

Dialog

Solana

Sri Lanka Telecom

Speedcast

Melbourne203.4.254.115

Colombo172.40.1.249

Server

AsiaSat-II

HIH 202.69.197.113

Server

HazInfo Project Research Design

With ERP Training No ERP Training

Sarvodaya S

tage 1, 2, 3

VSAT Urawatha(Galle)

MoPNidavur(Batticalo)

FxPThirukadalar(Trincomalee)

AREAMoratuwella(Colombo)

MoPMeddhawatha(Matara)

MoP Thambiluvil (Kalmunai)

FxPOluville(Kalmunai)

AREA Maggona(Kalutara)

AREA + RAD Modarapallassa(Hambantota)

AREA + FxPWathegama North(Matara)

AREA + MoPPalmunnai(Batticalo)

Control VillageAbeyasinghepura(Ampara)

AREA + RADThondamanar(Jaffna)

AREA + FxPKarathivu(Kalmunai)

AREA + MoPMunnai(Jaffna)

Control VillageModara(Colombo)

Sarvodaya S

tage 4

VSATModaragama(Hambantota)

MoPDiyalagoda(Kalutara)

FxPPeriyakallar (Batticalo)

AREAPanama North(Ampara)

MoPSatur-kondagnya(Batticallo)

MoPSamodhagama(Hambantota)

FxPIndivinna(Galle)

AREA Brahamana-wattha(Galle)

AREA + RAD Kalmunai II(Kalmunai)

AREA + FxPSamudragama(Trincomalee

AREA + MoPValhengoda(Galle)

Control VillageMirissa South(Matara)

AREA + RADVenamulla(Galle)

AREA + FxPKottegoda(Matara)

AREA + MoPThallala South(Matara)

Control VillageThalpitiya(Kalutara)

AREA: Addressable Radio for Emergency Alerts, Class B configuration of WorldSpace System

MoP: Java enabled Mobile Phone, Dialog-Microimage innovation MiDews application

RAD: Remote Alarm Device, Dialog-University-of-Moratuwa Innovation

FxP: CDMA Wireless Fixed Phones with 1xRTT functions, Sri Lanka Telecom

VSAT: Very Small Aperture Terminals coupled with Internet Public Alerting System Innovative-Tech & Solana Networks

Reliability measure of the Terminal Devices

Basic question: “Did the ICT work on the day of the exercise?”

Two aspects of Reliability measure: Certainty and Efficiency

Certainty is the operational state of the device (variable: Rc)

Efficiency is the time taken to complete the transmission (variable: Re)

Reliability R = Rc x Re

Calculating the CERTAINTY of a device receiving an alert

otherwise0

alerts receivecertainly can device1cR

Examples of mishaps during live-exercises in rural communities

User accidentally deletes the tri-language J2ME applet in mobile phone

Mobile phone is powered down or battery has zero energy

User removes the 2 AA batteries and powers down the AREA

Antenna in AREA is not aligned for maximum signal strength

CDMA phone bill was not paid and service is discontinued

User covers VSAT modem ventilation shafts with news paper and over heats modem

RAD not registered in DEWNS Internet bases alerting application

Calculating the EFFICIENCY of transmitting to a device

HIH-Monitor

ICT GuardianRelay Alert

ERP CoordinatorsDissemminate

Report Status

Acknowledge

Resolve CAP Alert()

Activate ICTG ERP()

Activate Community ERP()

Relay Results

Activate HIH ERP()

Download Alert()

T0

T1

T2

t0

t1

t2

: time taken to complete transmissionT: benchmark time transmission

must be completed)(TE

: minimal allowable time interval to impactT

0

)(1

1

T

TETRe

when )(TET

when

when TT

)(TET

The alert should be received well before the predicted time of hazard impact.

Some devices require more time for inputting and transmitting alert message.

Example of Calculating the Efficiency

HIH-Monitor

ICT GuardianRelay Alert

ERP CoordinatorsDissemminate

Report Status

Acknowledge

Resolve CAP Alert()

Activate ICTG ERP()

Activate Community ERP()

Relay Results

Activate HIH ERP()

Download Alert()

T0

T1

T2

t0

t1

t2

Tsunami Event occurred at 10:15am and will impact at 11:45

External source issued email bulletin at 10:25am

HIH Monitor receives email at 10:35am

HIH Monitor issues CAP alert at 10:46am

The scenario is based on the Panama (Ampara District) simulation data

ICT Guardian receives CAP alert over AREA-B at 11:02am

ERP Coordinator receives alert information at 11:08am

Community completes evacuation at 11:08am

Efficiency of ICT Network and ICT Guardian activities

Assumption: since this is the first set of trials and the LM-HWS has no data to calculate an ‘expected time we set 0)( TE

8222.090

161

cR

(i.e. best case scenario)

Effectiveness of Terminal Devices in conveying a warning message

Evaluated using a set of 11 discrete parameters Language diversity (ethnicity) Full CAP capability (all-media, all-hazards) Audio and text medium availability (communicability) Total cost of ownership (affordability) DC power consumption (longevity) Daily utilization (adoptability) Upstream Downstream communication (bi-directionality) Weight of wireless ICT (portability) Acknowledgment of message receipt (accountability) Geographical Signal coverage (ubiquity) Active alerting functionality (wakeup)

A score between 0 and 1 is given for each parameter

Calculating the overall Effectiveness of the Terminal devices

Each parameter in denoted by a literal: 1121 ,,, qqq

1,0:11,,1 ii q

1121 qqq

A ‘Liken” type rating is used to obtain a real valued score between 0 and 1 is for each literal:

Design requirements of the ICT Terminal device is denoted by a conjunction of the literals:

Overall effectiveness of the of the ICT Terminal device:

1121 qqqGG

1121 qGqGqGG

Calculating the Effectiveness Parameter: Active Alert Function

Enumeration of the Effectiveness parameter: Active Alert Function in

Terminal Devices

0.00

0.25

0.50

0.70

0.85

0.951.00

Active Alert Functions

Eff

ecti

ven

ess

Sco

re

Concerns the ability of the Terminal device to get the attention of the message recipient: ICT-G

Ideally the alert feature should work on all of the sensory domains

Researchers need to develop a better methodology by applying Human Computer Interface (HCI) or Man Machine Interface (MMI) Theory

Wakeup""G

Comparison of Reliability and Effectiveness of ICT as a Warning Technology in a LM-HWS

0.15

0.47

0.59

0.71

0.89

0.07

0.09

0.43

0.71

0.27

0.7

0.69

0.13

0.08

0.39

Control Goup

VSAT

RAD

FXP

MOP

AREA

AREA+RAD

AREA+FXP

AREA+MOP

ICT

Co

nfi

gu

rati

on

Effectiveness

ReliabilityNA

NA

planner should consider the deployment of multiple devices with the aim of achieving complementary redundancy in reliability and effectiveness

NA – Not Applicable

ICT

-G(n

)

Networks

Summary

Last-mile Hazard Warning System, Sri Lanka:

Efficiency is good in most cases (< 1min.)

Effectiveness varied widely, indicating user training and deployment issues

AREA shows high reliability but lower effectiveness than expected

MOP effective but with some reliability issues

When combined, devices can provide complementary redundancy

AREA (reliable) + MOP (effective)

Implications:

deploy technologies that provide complementary redundancy

Adopt Common Alerting Protocol for consistent messaging

Need to further refine reliability and effectiveness measures