2016/7/25

1

Disaster Resilience throughBig Open Data and Smart Things

Jane W. S. LiuInstitute of Information Science andResearch Center for IT Innovation

Big Open Data and Smart Things

Research Center for IT InnovationAcademia Sinica, Taiwan

The 3rd Keio Creativity Initiative Symposium, July 25, 2016

http://openisdm.iis.sinica.edu.tw

OpenISDM(Open Information Systems for Disaster

M t)

DRBoaST (Disaster Resilience through Big open data

Management) and Smart Things)

A framework for building open and

sustainable information systems for disaster

Generation and use of data for effective

disasters preparedness andsystems for disaster

managementpreparedness and

response

Supported by Taiwan Academia Sinica, Sustainability Science Research Program

2012 2016 2018

2016/7/25

2

C. S. ShihCSIE, NTU

J. M. Ho, IIS, AS

D. T. LeeIIS, AS

F. T. LinUP, NCKU

H L Li

J. C. LeeIES, AS

W T Li

W. L. HsuIIS, AS

T. W. KuoCITI, AS

J. K. ZaoCS, NCTU

H. L. LinUP, NCKU

H. C. ChouGeo, NTNU

N. C. HsiaoSC, CWB

W. T. LiangIES, AS

F. S. LinIHP, AS

H. C. LeeSES, NCDR

E. T. H. ChuCS, YunTech

J. W. S. LiuIIS, AS

On Power of Data/Information Information can save lives, livelihoods and resources.

2005 World Disasters Report You cannot have knowledge without data. DKM We cannot stop natural disasters but we can arm ourselves We cannot stop natural disasters but we can arm ourselves

with knowledge; so many lives wouldn't have to be lost if there was enough disaster preparedness P. Nemcova

Small advances in emergency informatics could significantly reduce deaths, accelerate damage assessment, & minimize economic downtime R. R. Murphy

0%0 24h 48h 72h

100%AvailabilityImpact

- 24h

Effective use

2016/7/25

3

fog and mist for IoTs

From http:///www.slideshare.netFrom http://www.ibmbigdatahub.com

g

Water, water everywhere, nor any drop to drink.from The Rime of the Ancientfrom The Rime of the Ancient Mariner, by Samuel Taylor Coleridge

Data, data h d everywhere, nor good

use of them to help us well prepared for calamities!

2016/7/25

4

Alert xmlns: Sender: Central Weather BureauStatus: ActualMsgType: AlertScope: PublicI f

Current Scenario

InfoCategory: GeoEvent: EarthquakeUrgency: ImmediateSeverity: Strong Certainty: Observed

Description: A strong earthquake measuring 7.8 occurred in

Parameters: Magnitude, depth, Areas: Polygons specifying

2016/7/2512

Areas: Polygons specifying affected areas

Resources:

Other EAS and CMAS

Effectiveness is limited due to Non-specificity of alerts Limitation in human reaction

time to take advantage

Cell broadcast

Alert xmlns: Sender: Central Weather BureauStatus: ActualMsgType: AlertScope: PublicI f

A Future Scenario

Location specific

instructionsInfo

Category: GeoEvent: EarthquakeUrgency: ImmediateSeverity: Strong Certainty: Observed

Description: A strong earthquake measuring 7.8 occurred in

Parameters: Magnitude, depth, Areas: Polygons specifying

Earthquake,Stay calm,

Walk in indicated directions

Areas: Polygons specifying affected areas

Resources:

Active use of alerts

2016/7/25

5

A Tornado Emergency Scenario

...THIS IS A TORNADO EMERGENCY FOR THE WARNED AREA... AT 703 PM CDT...NATIONAL WEATHER SERVICE METEOROLOGISTS AND STORM SPOTTERS WERE TRACKING A LARGE AND EXTREMELY DANGEROUS TORNADO. THIS TORNADO WAS LOCATED 6 MILES SOUTHWEST OF SAVOY MOVING NORTHEAST AT 45 MPH. THE TORNADO WILL BE NEAR...

CHAMPAIGN BY 720 PM CDT... RANTOUL BY 729 PM CDTPAXTON BY

A TORNADO WARNING MEANS THAT A TORNADO IS

Sound tornado sirens Unlock shelter doors Turn on spot lights on

shelters Adjust traffic signals on

A TORNADO WARNING MEANS THAT A TORNADO IS OCCURRING OR IMMINENT. YOU SHOULD ACTIVATE YOUR TORNADO ACTION PLAN AND TAKE PROTECTIVE ACTION NOW. SIGNIFICANT DAMAGE HAS OCCURRED WITH THIS SIGNIFICANT TORNADO! THIS IS AN EXTREMELY DANGEROUS AND SERIOUS LIFE THREATENING SITUATION..

j groutes to shelters Enable outside doors in

homes to open when the storm hits.

ADiPLE ADiPLE (Active Disaster Prepared Living Environment)

2016/7/25

6

Technology Enablers

Accurate and timely disaster predictiony p Wide adoption of XML-based international

standard alert message format Platforms for authenticating authorized alert

senders and dissemination of alerts Increasing adoption and use of building

information and facility management digital data exchange standards

DRBoaST

Disaster Prediction and Detection

From Minnesota Public Radio Org.

Tsunami warning, landslide warning, debris flow monitoring systems

2016/7/25

7

Earthquake Detection

$ 29

Affordable detectors based on that P (primary/pressure) wave travels

faster than S (secondary/shear) d t l h k $ 29waves, and generate only shock

Slower S waves are destructive Early earthquake warning systems

consisting of seismic motion sensors networked with computers can Issue warning of quakes of specified g q p

strengths for affected areas and progression of seismic events, and

Allow warning a fraction of a second to a few seconds or more before ground motion starts.

From early earthquake warning system, by P. Y. Lin in AUSMT, 2011

2016/7/25

8

Common Alert Protocol A common format that supports Broadcast of messages of all

types to all public alert systemsyp p y Message exchange between

emergency information systems Aggregation and correlation of

warnings from multiple sources End-to-end authentication and

validation

ITU-T Recommendation X 1303 (07/2007)

validation Automatic reports by sensor

systems to analysis centers Automatic processing by smart

devices and applications

Standard-Based Alert Delivery Authenticated

message senders

Integrated Public Alert and Warning System

http://www.fema.gov/emergency/ipaws/ipaws_cap_mg.pdf

2016/7/25

9

J-Alert, operational since 2009From https://www.city.nasukarasuyama.lg.jp/index.cfm/12,0,61,342,html

Data.gov.tw

CMAS(Google)

g

1212

14

NGOs

Alerts.ncdr.nat.gov.tw

From NCDR 04/2013

2016/7/25

10

iGadiGaDs in

Smart Homes and Buildings

(AlertType==Earthquake)AND(Magnitude>=6)OR

(AlertType==Tornado)AND(Severity>=3)AND(OutsideAirPressure * THRESHOLD RATIO < InsideAirPressure)

CAP-aware elevator controller

CAP-aware

iGaD

and Buildings(OutsideAirPressure THRESHOLD_RATIO=7.5)OR(AlertType==Tornado)AND(Severity>=4)

?xmlns version = 1.0 = 8.0) (AlertType == Earthquake) AND (Magnitude >= 6.0)

2016/7/25

11

BIM

From Daily life of building information modeling by Kristin Dispenza at Buildingpedia.com, June 2010

Old Building (4 floors)

First floor of IIS Building

New Building (8 floors above ground, one underground)

2016/7/25

12

BIM of Institute of Information

B ld Science Building generated from

blueprints

Inventory & floor-plan manager

Local Info

In-Store ServerStore Layouts

LAN access points

Retail data capture & analysis tools

Employee deviceEmployee

Employee device

Customers Mobile devices

Employee device

2016/7/25

13

Architecture and components for CAP Message Processor

Alert message

bufferModem Signaturevalidation

XMLparser

Architecture and components for building diverse iGaDs (intelligent Guards against Disasters)

We now have

pbuilding diverse iGaDs (intelligent Guards against Disasters)

Alert type &information Alert

records

Affectedareas

Location filter Configuration

files

Resources

Rule processor

Activation

Responsive active mechanism for pushing alerts over the Internet

Prototype Active Emergency

Device Controller

Device interfaces

Device location

Local dataRule engine

parametersand rules

yp g yResponse Systems (AERS)

Prototype building/environment data-based IPS (BeDIPS)

Emergency Command &

Control (ECC)Data access control

AERS

Integrated CAP

alert/warningmessaging

FM

Ai lit

Fire alarm system

Indoor Evacuation SystemSensor and device interfaces

messaging& notification

system

BeDIC

Other systems in ICT support infrastructure

Local emergency alert

systems

Security alerts

Air qualityalarm

Local sensors and surveillance system

iGaDs (CAP-aware embedded

devices)

Government emergency

alert/warning authorities &

response agencies

2016/7/25

14

Status of ADiPLE High-level objective: Enabling pervasive

deployment and use of AERS in 5-10 years Case studies on representative buildings: a home Case studies on representative buildings: a home,

an office building and a large public building Tools and support infrastructures, including

AERS framework for building diverse systems BeDIC (Building/environment Information Cloud)

Current efforts, including Location specificity: indoor positioning & delivery

of fine-scale decision support information Safety of active emergency response systems

An Advertisement

Indoor Positioning System

From http://www.extremetech.com/extreme/126843-think-gps-is-cool-ips-will-blow-your-mind

2016/7/25

15

911: What is the address of your emergency?

Caller: I am lost inside the Westfield Garden State Plaza Mall.

911: Where in the mall?

Goal set by US FCC in 2015: To find methods in four years that can pinpoint a mobile caller's location within about 50 meters.

A market arena full of Big players, including Google, Apple, Microsoft,

Qualcomm Broadcom and so on

On IPS: Indoor Positioning Systems work where GPS does not.

Qualcomm, Broadcom and so on 82 startups, 472 investors and 3,500 followers

according to AngleList as of April 18, 2016 A playground for diverse technologies, including

Triangulation of using WiFi, cell, and FM signalsFingerprints of magnetic signals of the building Fingerprints of magnetic signals of the building and WiFi and FM signals in the building

Measurements of light, acoustic, ultra wideband and low frequency signals, and

Proximity beacons

2016/7/25

16

Indoor Triangulation System

Accuracy affected by movements of people and objects High cost of maintaining information on access points

Magnetic-Fingerprint Indoor Positioning

FingerPrints Descriptions 3D coordinates

Eddie Bauer

Walking Co.

1, xxxxx, yyyyy

1, zzzzz, aaaaa

LEGO Co.

North Face

1, LLLL, cccc

1,

Is not scalable with building size and location queries Requires Internet access to fingerprint database Incurs high maintenance cost Can serve only smart devices

2016/7/25

17

IPS for large public buildings Works without Internet Stays responsive in

presence of mass crowd

Is easy to install & maintain Remains accurate despite

movement of people & thingsServes both feature and smart phonesServes both feature and smart phones

You are in front of RM 201 You are in front of

on Level 2

For nearest exit, take escalator

down to level G

RM 201on Level 2

For nearest exit,Take escalator down to level G

As shown in this Google Indoor maps ad On phones with no indoor maps and HereUAre

2016/7/25

18

FM

BeDICLbeacon data

BeDISP monitor &Testing programs

BeDIPS server

Wide Area Network

Lbeacon-URL

mapping

Gateway Gateway Gateway

Networked Lbeacons (Location beacons)

LBeacon APP Setup, Init, & Maintenance Built-in tests

UDP/TCPIP

GATT-based and BR/EDR

profiles and

OPP, IPS, OTP, OTSLNP and ILNP

Objects, synchronization

Lbeacon (Location beacon)

Attribute Protocol

Logical Link Control and Adaptation Protocol

RFCOMM Service Discovery

BNEP

Link Layer Link ManagerLi k M

profiles and protocols

Zigbee Network Protocol

Host-Controller Interface

Object EXchange

Link Layer

Low energy RF

Link Manager

Basic Rate RF

Link Manager

BR/EDR RF

ZigbeePhysical Layer

To Zigbee Gateway

2016/7/25

19

Practical Challenges

In a major transport hub (e.g., Frankfurt Airport): There may be 50 000 smoke detectors There may be 50,000 smoke detectors. 100,000 Lbeacons may be used to achieve the

desired location accuracy (e.g., 5-10 meters)

Where should the beacons be placed ? h h l f h How to pinpoint the chosen location of each

Lbeacon and install it there? How to systematically and reliably maintain

the system during remodeling?DRBoaST

FM

BeDIC

FMLbeacon data

BeDISP monitor &Testing programs

Lbeacon-URL

mapping

BeDIPS server

2016/7/25

20

FM

BeDICLbeacon data

BeDISP monitor &Testing programs

BeDIPS server

Wid A N t k

Lbeacon-URL

mapping

Wide Area Network

When design completes, the BeDIPS server contains For each area, the coordinates of

the reference point of the area

Reference point of the area

the reference point of the area For each Lbeacon in the area,

Type and QR code 3D coordinates Location description

DRBoaST

Installing Lbeacons Using a Servo-Controlled Laser Pointer When design completes, BeDIPS server contains

- QR code, coordinates, and type of each Lbeacon, and

- Coordinates of the reference point of each area The installer prepares each Lbeacon as follows

Reference point of the area

p p- Take a new beacon of the selected type,- Fix the QR code of the beacon on it, and- Load from the server its coordinates

To install Lbeacons in an area, the installer places the laser pointer at the reference point of the area and retrieves from the BeDIPS server the coordinates of the pointer. p

To find the location of a new Lbeacon, - The installer: have the laser pointer scan the

QR code of the beacon. - The pointer: reads the coordinates of the

beacon and points to the location on the ceiling with the coordinates.

laser pointer placed at the reference point.

Servo-controlled

2016/7/25

21

Response Time Measurements

Hardware : Raspberry Pi 2 Model B Raspberry Pi 2 Model B 1 and 2 Bluetooth classic dongles 10-16 feature phones

Software: Linux operating system

Bluetooth programming in C with BlueZ Bluetooth programming in C with BlueZ #include #include #include

DRBoaST

One dongle doing both discovery and object push

Inquiry for 10.24 sec. Object Push for ~ 40 sec. Resume InquiryInquiry

h tim

e in

sec

s

10

50

8.57

17.87

39.99

Variation in push time

0 2 4 6 8 10 121

Ave.

pus

h

No. of devices

1.45

4.884.25

0.73 1.5 2.48

DRBoaST

2016/7/25

22

Multiple dongles: One does discovery andthe others do object push

Inquiry for 10.24 sec. Resume Inquiry Timeout

Buffers for FHS packets of discovered devices

CO

NN

EC

T

PU

T

DIS

CO

NN

EC

T

Push time under 5 seconds with one dongle per 5 devices

DRBoaST

Plans for BeDIPS Demo BeDIPS in representative public buildings Release beta ersion of BeDIPS de elopment Release beta version of BeDIPS development

environment Extend the system for delivery of fine-scale

location-specific decision support data and emergency preparedness and response i t tiinstructions

Exploit the beacons for data and information delivery when Internet and cell coverage area disrupted

DRBoaST

2016/7/25

23

Squat under a

FM

BeDIC

How to provide hundreds of phones with fine-scale location specific decision support data?

Stay. Squat at a

columnGo out of front door

under a seat

End-Points of a Building Information Mist

URLsURL s of

nearby stores

Fine-scale, location-specific building/facility decision support data

Location-specific emergency response

instructions

2016/7/25

24

Message fileEmergency

Response inst.Setup, init

andmaintenance

File

Local navigation instructions

Location spec emergency response instructions

L ti ifi N i i i

BIM-FM data.

DZigbeed Bluetoothd

Obexd

3-D Coordinates

Obex PusherLbeaconself-test

File transfer

Location.-specific BIM-FM datasets

Emergency CMD RCVBeDI mist

Location descript.Navigation inst.

BLE driverBLE driverBLE driverBluetooth EDR driverZigbee driver

Bluetooth subsystem (GAP & GATT)Zigbee subsystem

Kernel

Daemonsg

OUTLINE Sample applications:

Opportunities and challenges Wh t ill it t k f ti di t d What will it take for active disaster prepared homes and buildings to be everywhere?

Answer: Three things - dependability, dependability, and dependability!

2016/7/25

25

Dependability Attributes

Maintainability Security

Interface standardsIntegration and testing frameworkCustomization &

Reliability

Availability Delivery of fine-

scale locationlocalization

Maintenance SOPs

Safety(of active devices & people)

scale location specific data

Data/information access during emergencies

Safety is an important requirement of active emergency response systems (AERS).

Observations and Definitions

A common definition of safety is the absence of dangerous conditions that can cause death, injury, and property damage.

An AERS is safe if its actions never create new dangerous conditions and never increase gthe probability of dangerous conditions.

2016/7/25

26

iGaDs may need to process concurrently multiple alerts that call conflicting responses, e.g.,

Challenges in Ensuring Safety

Strong earthquake: open all doors Local fire: open some doors and close some doors

Alerts may be cancelled and reissued Some advanced disaster prediction system may

issue false alarms Human actions may create unsafe conditions

USER(S)

Input Output XML Controller

Core Controller CONTRO

AERS Simulation Framework

User Interface

GUI Main Window

VIEW

Updates

Dynamic Loader

Thread Pool

OLER

User

Input

Manipulates

UserM Abstract agent

Crowds

Women OfficesWomen

Men

Mobile devicesElevators

Power switchesTest scenarios

Protocols and SOPsCAP alert sequencesDoors

Other AERS components

MODELS

Global & local environment

Stairwells Halls

Other building models

g

2016/7/25

27

GUI of AERS-Simulation Framework

Building objects

Furnishings People Crowd CAP-aware

objects

Case 1: Send nearly simultaneous multiple alerts that require conflicting responses

Evaluation of AERS Safety

Case 2: Send arbitrary sequences of alerts Case 3: Cancel alarms / reissue alarms Case 4: Issue false alarms Agent-based simulation: simulate

Actions and interactions of iGaDs and human actions Actions and interactions of iGaDs and human actions Normal and faulty behaviors

2016/7/25

28

Tools for converting building information models and interior layout models into inputs of matured,

Missing Pieces

commonly used simulators, e.g., A commercial tool for converting BIM 3D models into input for FDS/SMV

An open source 3D graphics tools for creating input to fluid and smoke simulatorssimulators

Integration of agents modeling active devices, people and crowd into simulated disaster scenarios.

DATA

DATA EVERYWHEREoffer us opportunities to become

much better prepared for disasters

2016/7/25

29

Thank You!You!