Dynamic Data Driven Application System

"'Dynamic Data Driven Applications Systems" ("'DDDAS'") is a paradigm whereby the computation and instrumentation aspects of an application system are dynamically integrated with a feedback control loop, in the sense that instrumentation data can be dynamically incorporated into the executing model of the application, and in reverse the executing model can control the instrumentation. Several closely related concepts predate DDDAS, including:
* N-armed bandit problems (Thompson 1933, Gittins 1979) and, subsequently, Reinforcement Learning:
* Chernoff'
Sequential Design of Experiments
* Fedorov's Design of Experiments (70s), where the experiments determine the parameters and the parameters guide the experiments, including for data and model selection.
* MacKay's Information-based Active Data Selection (1991) is where the expected informativeness of candidate measurements is used to select salient ones for learning, improving the expected informativeness.
* Active Learning (the 80s-90s), where the actively sampled environment improves learning, improving the sampling.
* Data Assimilation an
Adaptive Observation
(90s) -- In contrast to traditional data assimilation, new ideas integrated supplementary targeted observations into the assimilation system using forecasts and their uncertainties.
* Information Retrieval (90s), where queries generate searches, and the results refine the queries with relevance feedback.

Each of these well-developed concepts predates and embodies iterative two-way coupling that DDDAS' feedback loops for dynamic integration restate following the optimal design of experiment principles. DDDAS's primary contribution was expressing these concepts for multiple computational science application (simulation) scales. Indeed, much of the early effort (see Figure) connected data assimilation and adaptive observation, which is already well established. Descriptions of DDDAS as maximizing information gain naturally follow Mackay's arguments. Like them, DDDAS also sought to enable more accurate and faster modeling and analysis, calling this "systems analytics" and adding many more layers -- humans, software, architectures, and services, etc. as complex systems typically involve. The novelty of its central claim remains unestablished because no detailed comparisons by the community have been made concerning many of these prior concepts.

The DDDAS concept - and the term - is claimed by Frederica Darema. Its general contours emerged from the workshop co-chaired by Profs. Craig Douglas and Abhi Deshmukh for th
National Science Foundation (NSF) workshop in March 2000
(now defunct link). Many PIs were influenced into using the DDDAS term and concept by Darema's programs at NSF and AFOSR, which continued after Blasch took over as program manager at AFOSR. Thus, a community formed and advanced several new concepts.

For several years, DDDAS was organized as a workshop with ICCS. The first full-fledged but environmentally-focussed DDDAS conference was DyDESS, held at MIT. As time progressed, it was suggested by Dr. Ravela that DDDAS grow into its conference, adding workshops to special subjects. Through its Earth, Atmospheric, and Planetary Sciences department, MIT sponsored and organized the DyDESS and DDDAS 2016 and 2017 conferences and hosted the DDDAS 2020 and DDDAS 2022 conferences.

The affiliated meetings and conferences are:

* DDDAS workshops at ICCS (since 2003) organized by C. C. Douglas et al.
DyDESS
conference and workshop at MIT organized by Sai Ravela and Adrian Sandu
* DDDAS special session at the ACC organized by Puneet Singla, Dennis Bernstein, and Sai Ravela
* DDDAS Special Sessio
Information Fusion
* DDDAS 2016 at Hartford is the first full-fledged conference hosted and sponsored by MIT (Sai Ravela, EAPS).
* DDDAS 2017 at MIT is the second conference hosted and managed by MIT (Sai Ravela, EAPS).
* DDDAS 2020 at MIT is the third conference hosted online (Sai Ravela, EAPS).
* DDDAS 2022 at MIT is the fourth conference hosted by MIT (Erik Blasch and Sai Ravela, Co-Chairs) and organized wit
CLEPS22
(Sai Ravela).
* DDDAS 2024 will be held at Rutgers University (Erik Blasch and Dimitris Metaxas, Co-Chairs).

References

#F. Darema, "Dynamic Data Driven Applications Systems: A New Paradigm for Application Simulations and Measurements. Computational Science." Int’l Conf. on Computational Science (ICCS), LNCS, 3038, 662–669, 2004.
#F. Darema, "Grid Computing and Beyond: The Context of Dynamic Data Driven Applications Systems," Proceedings IEEE, 93(3), p. 692-697, 2005.
#G. Allen, "Building a Dynamic Data Driven Application System for Hurricane Forecasting," Int’l Conf. on Computational Science (ICCS), LNCS, vol. 4487, p. 1034–1041. Springer, Heidelberg, 2007.
#M. Denham, A. Cortes, T. Margalef, E. Luque, "Applying a Dynamic Data Driven Genetic Algorithm to Improve Forest Fire Spread Prediction," M. Bubak et al. (Eds.): ICCS 2008, LNCS 5103, pp. 36–45, 2008.
#E. Blasch, Y. Al-Nashif, and S. Hariri, “Static versus Dynamic Data Information Fusion analysis using DDDAS for Cyber Trust,” Procedia Computer Science, Vol. 29, pp. 1299-1313, 2014.
#X. Shi, H. Damgacioglu, N. Celik, "A Dynamic Data Driven Approach for Operation Planning of Microgrids," Procedia Computer Science, 2015.
#E. Blasch, S. Ravela, A. Aved, Handbook on Dynamic Data Driven Applications Systems, Springer, 2018.

External links

1DDDAS.org
Has a list of active projects and slides from the current DDDAS program and past contributions from NSF.
CAOS
Cooperative Autonomous Observing Systems for Mapping and Monitoring the Atmosphere @ MIT, jointly between EAPS and Aero-Astro
FireGrid
{{Webarchive, url=https://web.archive.org/web/20140714152013/http://www.firegrid.org/ , date=2014-07-14 FireGrid is a previous example of Emergency Systems.

See also

*Data assimilation
*Control systems
*Uncertainty Quantification
Theoretical computer science