Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
Dynamic Data-DrivenApplication Systems
(DDDAS)What is DDDAS?
Dynamic Data Driven Application Systems
DDDAS is a new paradigm.
Old ways: Application simulations use staticdata input and start up.
DDDAS (continued)
Application simulations are able to
RECEIVE AND RESPOND to ONLINE
physical data and measurements and/or
control the measurements.
Why DDDAS?
Computer models were not designed to dealwith dynamic conditions as the simulationsalready started.
Example: simulating wild fire. We need toknow which community needs to beevacuated. The wild fire simulation needs toadjust based on the wind direction and otherfactors.
Why DDDAS (continued)
Also fueled by advances in: Applications and algorithms for parallel and
distributed platforms. Computational steering and visualization. Computing. Networking. Sensors and data collection. System software technologies.
What DDDAS NeedsThree research areas: Application: data driven technology.
Algorithm: dynamic data injection and dataperturbation tolerance.
System software: support for dynamic environments.
Static applications can be changed into a more usefuldynamic data driven applications.
DDDAS General PropertiesWhat’s new in DDDAS?Feedback and control interactions betweencomputations and the physical measurement systems.
There are three ways to interact with DDDAScomputation: Human interaction. Physical system interaction. Computational infrastructure interaction (machines
and their connections).
DDDAS Interactions
DDDAS Key Characteristics
Key characteristics that need to be addressed:
Time dependency and/or real time aspect.
Data streams in addition to data sets.
Interactive visualization and steering.
DDDAS ExampleTsunami simulation.(http://www.pgc.nrcan.gc.ca/geoscapevictoria)
Where we get the data from?Tsunami can be caused by tectonic platemovement, or sea floor quakes.
Data sources: GPS stations, sea floor sensors.
DDDAS Example 1 (continued)
DDDAS Example 1 (continued)
DDDAS Example 1 (continued)
DDDAS Example 2
Air traffic simulation.(http://www.simlabs.arc.nasa.gov/cvsrf/atcs.html)
Where we get the data from?We need the world location for all the aircrafts, as wellas their schedules. Outside data: weather simulations.
Data source: Aircraft transponders.
DDDAS Example 2 (continued)
DDDAS Example 3Medical imaging and simulation.(http://www.bitc.gatech.edu/bitcprojects/eye_sim/eye_surg_sim.html)The Georgia Institute of Technology and the Medical College ofGeorgia.
We need the real life model of the object ofinterest. We can do this by using IntensityModulated Radiation Therapy (IMRT).
Simulating eye surgery. To blind or not blind.
DDDAS Example 3 (continued)
DDDAS Example 3 (continued)
DDDAS Researches: Past,Present, and Future.
Reference taken from DDDAS.org HomePage,
NSF Official DDDAS Page, NSF 2000Workshop on Dynamic Data-Driven
ApplicationSystems, and Performance EngineeringTechnology Notes.
DDDAS: Past and Present
Past ExperienceFebruary 17, 2000: Meteorologists missed
predicting the track and magnitude of amajor storm in January 24-25, 2000, thatblanketed major cities from South Carolinato New England.
May 7, 2000: The National Park Servicestarted a controlled burn near Los AlamosNational Laboratory. Within a day, the firewas labeled a wildfire.
What is being sought fromDDDAS?
DDDAS Capability: simulation applications thatcan dynamically accept and respond to fielddata and measurements, and can control suchmeasurements in a dynamic manner throughthe symbiotic measurement and simulationsystems.
How DDDAS Affects OurPresent and FutureExamples of Applications benefiting fromDDDAS:
Engineering Design and Control: aircraft design, oilexploration, semiconductor manufacturing,structural engineering, computing systemshardware and software design and runtime.
Medical: customized surgery, radiation treatment,BioMechanics/BioEngineering
How DDDAS Affects Our Presentand Future (continued)
Economy: Production planning and control,financial trading (stock market, portfolioanalysis).
Crisis Management and EnvironmentalSystems: Weather, hurricane/tornadoprediction system, floods, fire propagation,transportation systems (emergencyplanning, accident response)
Challenges in EnablingDDDAS Capabilities
Application Simulations Developments
Algorithms
Computing Systems Support
Challenges in EnablingDDDAS Capabilities(continued) Applications:
Ability of the application to interface with measurementsystems.
The stream data might introduce new modalities todescribe the system, like a different level of the physicsinvolved, in cases where the analysis is about a physicalsystem.
Ability to dynamically select the application componentsdepending on the dynamically streamed data.
Challenges in EnablingDDDAS Capabilities(continued)
Algorithms:
Stable to dynamically injected data/ tolerant toperturbations of dynamic input data.
Visualization with a human in the loop: feedbackto the simulations.
Handling data uncertainties: continuoussensitivity analysis
Challenges in EnablingDDDAS Capabilities(continued)
Computing Systems Support:
Support environments where the applicationrequirements change during the execution dependingon the streamed data/dynamic execution support onheterogeneous environments.
Extended spectrum of platforms: Grid Computing andbeyond.
What is Grid Computing?
What is Grid Computing?(continued)
Grid computing enables the virtualization of distributedcomputing and data resources such as processing, networkbandwidth and storage capacity to create a single system image,granting users and applications seamless access to vast ITcapabilities.
Just as an Internet user views a unified instance of content via theWeb, a grid user essentially sees a single, large virtual computer.
At its core, grid computing is based on an open set of standardsand protocols that enable communication across heterogeneous,geographically dispersed environments. With grid computing,organizations can optimize computing and data resources, poolthem for large capacity workloads, share them across networksand enable collaboration.
What is Grid Computing?(continued) Like the Web, grid computing keeps complexity hidden: multiple users enjoy a
single, unified experience.
Unlike the Web, which mainly enables communication, grid computing enables fullcollaboration toward common business goals.
Like peer-to-peer, grid computing allows users to share files.
Unlike peer-to-peer, grid computing allows many-to-many sharing — not only filesbut other resources as well.
Like clusters and distributed computing, grids bring computing resourcestogether.
Unlike clusters and distributed computing, which need physical proximity andoperating homogeneity, grids can be geographically distributed and heterogeneous.
Like virtualization technologies, grid computing enables the virtualization of ITresources.
Unlike virtualization technologies, which virtualizes a single system, gridcomputing enables the virtualization of vast and disparate IT resources.
Why DDDAS Now?DDDAS has the potential to revolutionizescience, engineering, medicine, economy,management systems, and etc.
We have technological progresses that has advancesthe level of overcoming the challenges:
- computing speed (terascale, uni- and multi-processor systems, Grid Computing, SensorNetworks).
Why DDDAS Now? (continued)- System software
- Applications (parallel and grid computing)
- Algorithms (parallel and grid computing,numeric
and non-numeric techniques, data assimilation,and chaotic Monte-Carlo method).
Summary and What the FutureHolds
NGS: Next Generation Software (1998 - …) Develops systems software supporting dynamic
resource execution
Summary and What the FutureHolds (continued)
SES: Scalable Enterprise Systems Program(1999, 2000-2003) Geared toward commercial applications.
ITR: Information Technology Research (NSF-wide project) Has been used as an opportunity to
support DDDAS-related efforts.
Summary and What the FutureHolds (continued) Simultaneous advances on the models,Simultaneous advances on the models,
methods, and algorithms that underpin themethods, and algorithms that underpin thecomponents components –– and on their and on their systematicsystematicintegrationintegration to target strategic applications to target strategic applications ––are crucial for realizing the potential ofare crucial for realizing the potential ofDDDAS.DDDAS.
But hardware, software, and CyberBut hardware, software, and CyberInfrastructure alone are insufficient toInfrastructure alone are insufficient toachieve this goal.achieve this goal.
Case Study: Architecture of the DDDASWildfire Model
Major Components of the Model
Coupled atmosphere/fire model Legacy NCAR code Combined with the latest techniques, such as OpenMP and Multigrid
Data acquisition From the Internet: GIS maps, past fire information, weather Field information: photos taken from aircraft, field sensors
Visualization and user interface runs on PDAs or cell phones in the field
Dynamic Data Assimilation control module Incorporates data from the field Bayesian data assimilation Data assimilation steers the data acquisition
Guaranteed secure communication infrastructure
The NCAR Coupled Atmosphere/FireModel
The interaction of fire and atmosphere is important Heat flux from the fire to the atmosphere produces fire wind Wind facilitates the fire spread
Traditional fire models cannot represent this interaction Wildfires are difficult to model Limited computational resources
An meteorological model is coupled with an fire spread model The atmosphere model is based on the Clark-Hall Atmospheric Model Fire model can be an empirical model, or a more realistic Stochastic
Reaction-Diffusion Equation Fire Model Represents the important interaction between fire and atmosphere,
more accurate and closer to reality
Data Acquisition
Geographical, weather, and fire information available from theInternet
Weather information: NOAAPORT broadcast, MesoWest weather stations,and the Rapid Update Cycle (RUC) weather system by NOAA
Past fire information: the GeoMAC project at USGS Fuel type: national database
Advancement of the fire front infrared pictures taken from aircraft GPS for aircraft position 3-axis inertial measurement to get the pointing direction of the camera
Field fire and weather information data Sensors arbitrarily placed in the vicinity of a fire for point measurements of
fire and weather information
Visualization and User Interface
Simulation results (prediction) will be transferred to thefirefighters in the field
Need an easy and portable way to visualize the simulation result
Firefighters don’t want to carry a notebook when fighting fires
PDA (or cell phone) and Java is the natural choice
Limited computing power and memory
Needs Java-based graphic software
Why Dynamic Data Assimilation?
Wildfire is a complex process with lots of uncertainties, such aswind, humidity, temperature
Neither empirical model nor physical model cannot represent thewildfire very well
Lots of parameters cannot be measured accurately
All in all, the system is heavily non-linear and ill-posed
Sequential Bayesian data assimilation can be used to guide thesimulation
How Dynamic Data Assimilation works?
The state of the system at any time - physical variables andparameters of interest at mesh points
Time-state vector x - snapshots of system states at differentpoints in time
The knowledge of the time-state of the system - probabilitydensity function p(x)
p(x) is represented by a ensemble of time-state vectors x1, x2, …,xn
Number of system states maintained = size of the ensemble *number of snapshots
Thousands of simulation run simultaneously
Sequential Bayesian Filtering
Current state of the model prior probability density –
Incorporate data from the field Measurements – vector y How the data is derived from
x – p(y|x)
Posterior probability density –
The system advances in timefrom the posterior probabilitydensity until new data arrives;this process is called ananalysis cycle
)(xp f
!=
""" dpyp
xpxypxp
f
fa
)()|(
)()|()(
Standard Approach of DataAssimilation by Ensemble Filter
Initialization: generate initial ensemble by a randomperturbation of initial conditions
Repeat the analysis cycle: Advance ensemble states
to a target time by solvingthe model PDEs in time
Inject data with time-stampsequal to the target time:modify ensemble states bya Bayesian update
Overall Pictures
References Geoscape Victoria http://www.pgc.nrcan.gc.ca/geoscapevictoria/ Air Traffic Controls Simulator
http://www.simlabs.arc.nasa.gov/cvsrf/atcs.html Eye Surgery Simulation
http://www.bitc.gatech.edu/bitcprojects/eye_sim/eye_surg_sim.html
DDDAS by Dr. Craig Douglas http://www.dddas.org/ DDDAS by NSF
http://www.nsf.gov/cise/cns/darema/dd_das/index.jsp Jan Mandel’s DDDAS Website http://www-
math.cudenver.edu/~jmandel/dddas03/