Lynne Grewe California State University East Bay lynnegrewecsueastbayedu Presentation by Funda Erdin Disaster Recovery Issues Disaster Incident DefinitionProtocol Categories of Data representing Incident ID: 540199
Download Presentation The PPT/PDF document "Metrics of a System for Disaster Relief" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Metrics of a System for Disaster Relief
Lynne GreweCalifornia State University East Baylynne.grewe@csueastbay.eduPresentation by Funda ErdinSlide2
Disaster RecoverySlide3
Issues
Disaster Incident Definition/ProtocolCategories of Data representing IncidentMethods of Data CaptureData and Uncertainty Computer RepresentationFusion Processes to Reduce Data Size.Storage and Data Distribution NeedsPersonnel Roles and related Security Needs
Data Presentation
Communication
Needs and Performance
Administration and Incident Control ToolsSlide4
Previous Work
Currently, mostly disaster relief is performed without any computer management tools as shown in imagesProtocols (NIMS/ICS)Describe human interactions … not as quantifiable as neededSpecialized ComponentsSome previous work concentrates on very specific issues (e.g. ambulance routing)
C3 related
work
Command, Control, Communications – C3, US military related work, goals are not disaster reliefSlide5Slide6
Our System: DiRecT(Disaster Recovery Tool)
VisualizationUncertainty VisualizationDistributed InfrastructureCommunications Application of Mobile AgentsProtocol StructureClient ControlSlide7
DiRecT
OverviewSlide8
DiRecT Server
EJB components for persistence and business logic JMS and Mobile agents for instant-memoing JMS for incident updates Oracle database for persistenceSlide9
DiRecT Field & Incident Command
Clientscreate a new incidentmanage multiple incidentsrequest resources, personnel and equipment for a given incident
assignment of personnelSlide10
DiRecT Admin Client
Activation / Deactivation of incidentsCreation of new personnel, equipment and resources.Assigning personnel to incidentsFulfilling resource requestsPurging incidents from the databaseSlide11
Incident
Tracking of victims, personnelHazardous materialsNatural hazards responseSearch-and-rescue missionsFire controlAir, rail, ground, and water transportation accidentsIncidents with multiple casualties…and others.Planned human events, e.g., large crowd gatherings, concerts, etc.Slide12
Incident CreationSlide13
Beginning of IncidentSlide14
Incident Data
Many kinds of data possible, different operating units (firefighters, police, etc) may want different kinds of data.DiRecT’s Data CategoriesBioTargets (victims)Search AreasEquipmentPersonnel
Hazards
ImagerySlide15
Data and Uncertainty Metric Goals
Quickly enter in dataCapture the essence of the underlying dataAllow for longer more narrative (open) forms of data capture optionally.Have quantitative metrics when possible to make visualization easier and storage efficient.Have metrics easily understood by all types of possible users.Have metrics be intuitive when possible.Slide16
BioTarget Data Metrics
Capture common dataQuantizeCapture uncertainties.Data measured: IdentityLocationHealth
SafetySlide17
Biotarget Data Capture
IdentitySlide18
Biotarget Data Capture
LocationSlide19
Biotarget Data Capture
HealthSlide20
Biotarget Data Capture
Safety / StatusSlide21
Search Area Metrics
Represent areas of search (or past search)Typical geometries – rectangular, circularCapture uncertainties.Data measured: Geometry size, shape, location, searching or searchedSlide22
Search Area Data Capture
Give identity through nameLabel as searching or searchedGive geometry and associated certainty of thisSlide23
Equipment Data Metrics
Represents various kinds of equipment and resources in the fieldCould be lots of different kinds of equipmentData measured: Identity, type, locationSlide24
Equipment Data Capture
Give identity through nameLabel Category or type in your own if not listedLocation same as biotarget data captureRight screen shows listing of current equipment in Incident.Slide25
Personnel Data Metrics
Represents various personnel in the field (firefighters, police, etc).Data measured: Identity, personnel type, locationSlide26
Personnel Data Capture
Give identity through nameType indicates unit personnel belongs to.Location same as BioTarget data captureSlide27
Hazard Data Metrics
Represents hazards in the scene.Many kinds (fire, water, wind, etc).Data measured: Identity, locationSlide28
Hazard Data Capture
Give identity through name, type, optional description, condition and certaintyCondition – scale of 0 to 100 indicating level of containment & severity. 0 is contained/low level problem. 100 is not contained/criticalSlide29
Imagery Data Metrics
Represents raw imagery type data collected about incident.Many kinds possible – photographic imagery, infra-red, maps, etc.Data is ALREADY quantifiedProblem here is to register the data against our Incident Visualization Grid so it can be fused.Data measured: File upload, User entered Registration pointsSlide30
Imagery Data Capture
Name of image layerActivate layer or notData fileRegistration informationSpecified by user by clicking with mouse on upper-left and lower-left location ofimage boundaries.Opacity control
Optional description
Rotation controlSlide31
DiRecT Visualization
DiRecT take these data and certainty metrics and performs visualizationSlide32
Some Visualization Cues
Opaqueness-TransparencyIcons/GlyphsColor (pseudo-coloring or color representation)Brightness/IntensityTextureAtmospheric EffectsAdding/Altering GeometryLayers
Focus
Pop-up textual information
Animation
Morphing
Time Fading
Sounds (volume, key, duration, fade)Slide33
Biotarget Visualization
IconicColorTransparencyLocationBloom
indicates
BioTarget
idenity
type
Indicates combined Health
and Safety w/Certainty
indicates identity and presence certainty
indicates location information
indicates location certaintySlide34
Biotarget Color
C
H
C
S
ALTERED COLOR
ICON
100
100
Unchanged – RED
50
100
Unchanged – RED
75
75
Some Uncertainty – ORANGE
50
50
Significant Uncertainty – ORANGE YELLOW
0
0
Very Much Uncertain - YELLOW
Color of Icon = F(
Health,Saftey
, Certainty)
Red = Max(
ColorSafetyR
,
ColorHealthR
)
Green =Min(
ColorSafetyG,ColorHealthG
)
This visualization is intuitive.
Red –alarm/high concern. Green – safe Yellow –caution/uncertain.Slide35
Example
Visualiztion
Image, map,
BioTargets
, Search Areas, Equipment, Hazard and Personnel
.Slide36
Search Area Visualization
Colorsearching searchedShapeTransparencyLocationSlide37
Equipment Visualization
Iconic &Color Medical Hazard WaterTransparencyLocation&BloomSlide38
Personnel Visualization
IconicTransparencyLocation&BloomSlide39
Hazard Visualization
Iconic &Color Water Explosion ChemicalTransparencyLocation&BloomSlide40
Visualization Control
Control clutterBetter DecisionsView only desired dataSlide41
ContrastSlide42
Biotarget
Before
Highlight
<= 60% healthSlide43
AreaSlide44
Filter
BeforeCertainty > 60%Slide45
Infometrics – basic statistics about Incident.
CountSearch
Search and highlight for biotargets Health <=60%Slide46
Count
Count in Area BioTargetsSlide47
DiRecT Infrastructure Issues
Remote method invocations Transparent fail-over Back-end integrationTransactions Clustering Dynamic redeployment Clean shutdown
Logging and auditing
Threading
Object life cycle
Resource pooling
Security
Caching
CommunicationsSlide48
DiRecT ServerSlide49
DiRecT Communications
We can’t in the time allowed look at all the components of InfrastructureLook at communications modulePerforms a kind of “Instant- Memoing” for asynchronous communications in the field.2 implementations
Will look at Performance metricsSlide50
Instant MemoingSlide51
Implementation 1 : Messaging via JMSSlide52
Implementation 2: Mobile Agents – using AgletsSlide53
Aglets over JMS
Aglets can very easily and efficiently send private messages, while with JMS it is not so simple.Aglets is explicitly asynchronous while JMS can be made asynchronous through durable subscriptionsEach mobile agent can carry different a encoding/decoding algorithm.Agents can be controlled and can react dynamically to unfavorable situations on a host JMS reliable, mature technology.Slide54
Size of Message
Avg Time in Milliseconds for 10 messages
Aglets
(Milliseconds)
15KB class file + 16 chars data
51.3
JMS
(Milliseconds)
16 chars data
49.7
Aglets
(Milliseconds)
15KB class file + 32 chars data
54.1
JMS
(Milliseconds)
32 chars data
49.9
Aglets
(Milliseconds)
15KB class file + 64 chars data
56.7
JMS
(Milliseconds)
64 chars data
50.3
Aglets
(Milliseconds)
15KB class file + 564 chars data
91.2
JMS
(Milliseconds)
564 chars data
52.8Slide55
Performance Aglets over JMS
JMS Superior, especially as message size grows.Aglets have some other possible advantages.Slide56
Future Work
TestingSystem IntegrationPDA, other devicesPredictive & Planning (AI)Other forms of MediaUser-specified/assigned visualization queues