Indian Institute of Technology Kharagpur
Indian Institute of Management, Kolkata
Bengal Engineering and Science University, Shibpur
National institute of technology, Durgapur
Heritage Institute of Technology, Kolkata
Kalyani Government Engineeting College, Kalyani
Personal Background Indian Scenario Disaster management….
(personal experience)
Scenario User Meet
Advanced Country and Backward Case study to bring out contrast and hopes
Motivation
Summary of criticisms
No comparison or mention of the recent Japan disaster (A.2)
Lack of a detailed sketch of the architecture to be deployed (A.1, A.3)
Absence of a precise problem definition (A.4-A.6), (B.1, B.2)
Our connections with the disaster relief personnel are not clearly stated (C.1, A.9).
Communication infrastructure (A.1 – A.3)
No cellular infrastructure
We propose latency-aware ad-hoc network architecture
SP – Shelter points
MCS – Master control Station
Four tier architecture
Communication infrastructure (A.1 – A.3) Tier - 1: Rescue personnel carrying smart phones that can form a DTN; exchange information among themselves through the DTN
Tier -2: Message packets shall be unloaded into Throwboxes belonging to SPs
Tier – 3: Communication among far apart Throwboxes shall be facilitated through Data Mules
Tier – 4: One of the Throwboxes within a small group shall be elected as the center to communicate through Wi-Fi (Line-of-Sight) devices with other groups as well as with the outside world
MCS
Latency o
f 3hr 2
0 mins
matures…
.Laten
cy = 3
hr. 20 m
ins
Group 1Group 2
Group Center
Group Center
TBTB
Communication infrastructure (A.1 – A.3)
[Saha 12] S. Saha, V. K. Shah, R. Verma, R. Mandal, S. Nandi,” Is It Worth Taking A Planned Approach To Design Ad Hoc Infrastructure For Post Disaster Communication?”, ACM MobiCom Workshop on Challenged Networks (CHANTS) 2012.
Four Tier hybrid Architecture using DTN Nodes, ThrowBoxes(TBs), DataMules(DMs) and WiFi Towers(WTs)
Research problems: 4 tier architecture (A.5, B.1)
Given a (i) initial disaster hit area map which can be realized as a graph G(V,E), where V= {set of ThrowBox (TB) at each shelter point} and E={set of pathways among those TB}. (ii) finite pool of resources and (iii) load (function of no of victims and size of an affected area which is under coverage of one Throwbox) -- what is the minimum achievable latency (L) such that almost 100 % packet delivery is ensured?
Conversely for aforesaid graph with given (i) latency, and (ii) load -- what is the optimal number of network resources that are required to ensure almost 100 % of packet delivery?
As disaster hit area may change with time it is also important to answer with existing network setup and given resource what is the minimum achievable latency with least/ tolerable movement of resources to ensure almost 100% of packet delivery?
Designing the mixed-mode routing protocol lie in dealing with tradeoff between fairness and prioritized access, protocol inter-operability, authentication, universal user/device identity, group multicast, etc.
24 K
M
20 KM
Input Parameters
Area 480 Sq KM
Location Sundarban
Number of TBs 19
Location of TBs Given
Location of MCS Given
No of DTN nodes 10 DTNs/TB
Traffic Rate 4 Packets/hr/DTN11 Packets/hr/DTN
Mobility Model Post office cluster mobility model
Simulator Enhanced ONE Simulator
Affected Area
MCS
Our System Model
List of Components
Quantity
DataMules(DM) 11
WiFi Tower 7
List of Components
Quantity
DataMules(DM) 11
WiFi Tower 8
Required Resources for Satisfying 3hr 20 minutes & 2 hrs 40 Minutes Latency 3
hrs
20
min
sla
tenc
y
2 hr
s 40
m
ins
late
ncy
Comparison between Planned & Random Placement with 3 hrs 20 minutes Latency
Planned Approach Random Approach
Time vs. Delivery Probability for Planned and Unplanned Approach
Degradation of delivery probability due to Deployment Overhead for topological
changes
Time vs. Delivery Probability with Low Load
Topological change towards betterment with Time Keeping Same Latency Constraint
Initial topology Improved Topology
Google Map Based User Interface for Network
Resource Planning
Feeding InputHighlighting Activity Area and Variable Constraints are Fed
ThrowBox & Shelter Point CreationSP and TB are Drawn on Google Map using Google Map
API
Determining Distance Finding out the Geographical and Geodesic Distance between TB’s
for each Combination
Group FormationTB’s form various groups as per the Heuristic
Approach Adopted & Wi-Fi Tower is placed on the Group Center
Finding Data Mule Trajectory
Finding DM trajectory and the number of DM required for each group
Location awareness
Problem Definition
Build an annotated people/resource map Approximate position of the victims and the
resource situation shall be highlighted on a time-varying basis
Location awareness
Solution
Physical co-ordinates provided by smart-phones carried by rescue workers shall serve as anchor points
Clever Crowd Sourcing to identify other users.
Location awareness CaveatsGPS Constraints Many mobile phones (Give a snapshot of your phone)
cannot do GPS localization without wireless connection Atmospheric Condition hinders GPS localization
Mobility and Delay Delay in aggregating the data brings in accuracy as
the user has already moved from the reported position
Mobility induced error correctionMovement is predominantly deterministic
Landmark based localization
Annotate the disaster prone area graph with landmarks Identify them as anchor points
Location awareness
Proposed methodology
In-built sensors of smart phones produce/identify anchor points
Gyroscope, accelerometer – can sense turns/bumps on roads
Relative humidity sensor – can sense the presence of a place like pondBarometric sensor / gravity sensor – can sense the different floors of a building
Location awareness
Initial study: landmarks of the 2nd floor of the CSE Department, IIT Kharagpur Phone : Samsung Galaxy S2
i9100GPlatform : Android 2.3.6 App Used : SensoSaur
CSE IIT Kgp 2nd Floor Plan
04/22/23Swadhin
28
Toilet
Starting/Ending Pt.
Landmarks using GSM signal strength (Manual)29
Strongest (15-22)
Weakest (7-12)
Landmarks using Wi-Fi (Manual) 30
Linksys,HWLAB,Research,Abhishe
k_PC
Linksys,HWLAB,Research,Abhishek_PC,58.x
AP
Linksys,HWLAB,Research
Linksys,Research,HWLAB,58.x AP
Landmarks using light (Day) 31
Highest Light
Highest Light
Lowest Light
Landmarks using light (Night) 32
Bright (Tube Light)
Dark (No Tube Light)
Landmarks using Gyro/Lacc/Magnetometer/Rotation vector
33
Using online social media to gather authentic situational information (A.4) Good indicators of the situation of victims of
man-made calamities – victims themselves can tweet about the extent of damage caused and the specific help required
Recent studies show that 30% of tweets posted immediately after calamities contain situational information …
But only 17% are credible
Using online social media to gather authentic situational information (A.4) Research questions
Judge credibility of the tweets posted immediately after a calamity
Retweets not a good metric since rumors might also get retweeted millions of times
Proposed: identify authoritative experts on calamities Challenge – to discover such experts in Twitter Utilize knowledge of location of the person posting a tweet Analyze local flow patterns Sensor information can be coupled with tweets
Example of rumors after Blasts in Mumbai
http://blogs.wsj.com/indiarealtime/2011/07/15/mumbai-blasts-did-twitter-really-help/
http://www.in.com/news/current-affairs/mumbai-blast-13th-and-on-kasabs-birthday-19724405-in-1.html
Example of rumors after terrorist attack
Examples of rumors after UK Riots
http://www.guardian.co.uk/uk/interactive/2011/dec/07/london-riots-twitter
Examples of rumors after UK Riots
Comparison with relief measure Japanese earthquake India and other developing countries are still not
equipped with an organized post-disaster relief programme
Japan enjoys State and private machineries, Huge economic power, A strong socio-cultural backbone
to combat with post-disaster situation that is non-replicable in the context of India
Japan use UAVs (unnamed aerial vehicles ) and under water robots to analyze post-disaster situation and fix damaged cables
The economic strength of India does not allow of such an infrastructure and an alternative could be to use smart-phones (augmented with various sensors) to design low-cost (possibly not the best) solutions
Online social media: Source of authentic situational information (A.4)
Works as good indicators of the situation of victims of man-made calamities – victims themselves can tweet about the extent of damage caused and the specific help required
Recent studies show that 30% of tweets posted immediately after calamities contain important situational information
Research questions – credibility of the tweets posted Retweets -- not a good metric since rumors might
also get retweeted millions of times
Rumors about disaster on the social media (Mumbai Blast)
http://blogs.wsj.com/indiarealtime/2011/07/15/mumbai-blasts-did-twitter-really-help/
http://www.in.com/news/current-affairs/mumbai-blast-13th-and-on-kasabs-birthday-19724405-in-1.html
Rumors about disaster on the social media (Mumbai Blast)
Rumors about disaster on the social media (UK Riots)
http://www.guardian.co.uk/uk/interactive/2011/dec/07/london-riots-twitter
Rumors about disaster on the social media (UK Riots)
Online social media: Source of authentic situational information (A.4) Identify local authoritative experts – difficult
to track since Twitter-like social media are full of celebrities that completely shadow the presence of these experts
Challenge – to discover such experts Analyze local flow patterns
Sensor information can be coupled with tweets Relevance of a tweet reporting a damage can be
better judged if the location information of the person tweeting is available.
Comparison with relief measure Japanese earthquake India and other developing countries are still not
equipped with an organized post-disaster relief programme
Japan enjoys State and private machineries, Huge economic power, A strong socio-cultural backbone
to combat with post-disaster situation that is non-replicable in the context of India
Japan use UAVs (unnamed aerial vehicles ) and under water robots to analyze post-disaster situation and fix damaged cables
The economic strength of India does not allow of such an infrastructure and an alternative could be to use smart-phones (augmented with various sensors) to design low-cost (possibly not the best) solutions
Collaboration and management plan: Past Experience (Research) A Secure Decentralized Disaster Management
Information Network using Rapidly Deployable Wireless and Mobile Computing Technologies -- A DIT Funded project successfully accomplished by Prof. Somprakash Bandyopadhyay (IIM Calcutta), Dr. Siuli Roy (HIT, Kolkata) and Mr. Sujoy Saha, NIT Durgapur
Work on directional antenna, DIT funded projects on peer-to-peer networks, Vodafone funded projects on mobile communication networks (Prof. Niloy Ganguly)
The communication was setup among three island near gosaba using 802.11 enabled with optilink devices configured in point to multipoint bridge with 15dbi antenna which was found to be able to cover near about 7 to 8 km range in line of sight. Using this link voice communication was established with NGOs as shown in figure.
Collaboration and management plan: Past Experience (Field work)
Collaboration and management plan: On-going work IIT Kharagpur
Comparison of infrastructure for disaster management after calamities in Japan, Pakistan, Haiti and India
Information dissemination in DTN using bipartite networks
Online social media for gathering situational updates (jointly with BESU Shibpur )
SensoSaur - Landmark based location tracking Collaborative download framework
IIM Kolkata
Collaboration and management plan: On-going work NIT Durgapur
Layered architecture for post-disaster communication Disaster management services - Android application Google map based user interface for disaster
management BESU Shibpur
Using online social media for gathering situational updates (jointly with IIT Kharagpur)
Installing smart-phone based DTN in disaster-hit region HIT Kolkata
Wireless sensor networks – tracking miners and the mining environment, agro-parameter monitoring system, pollution monitoring system, traffic congestion detection system
Impact on curriculum
Study groups Planet-lab installation Tweet collection on social media PG level electives:
Wireless Networks for Crisis Management: IIM Kolkata + KGEC
ICTs for Disaster Management: IIM Kolkata + Heritage Institute of Technology (HIT)
Distributed Systems and Unreliability: IIT KGP + BESU
Impact on curriculum
Disaster related database development Building computer solutions Refresher course (through AICTE) Recruitment of new faculty members Summer/winter training programmes Measure to attract good MS/PhD students
Societal sensitivity development Through the established banners of
Initiative for Community Action (INCA) – IIM Kolkata National Service Scheme -- IIT Kharagpur
A structured list of members (User-Groups) NGOs Government agencies Community based organizations Victims of disaster
First User-Group meeting already held at IIM Kolkata on March 30, 2012
Four workshops for sensitization planned to be held at IIT KGP and IIM Kolkata
References http://cse.iitkgp.ac.in/resgrp/cnerg/disaster_dtn
/