May 24, 2016
Professors Biao Chen, Yingbin Liang, Jian Tang, and Pramod Varshney, along with researchers from the University of Illinois at Urbana-Champaign, have been awarded a multi-year grant from the Air Force Office of Scientific Research (AFOSR) to pursue a dynamic data driven approach to information fusion for intelligence, surveillance, and reconnaissance missions. The multi-disciplinary project brings together signal processing experts with computer scientists, and was funded under the Dynamic Data Driven Application Systems (DDDAS) program, spearheaded by Dr. Frederica Darema at AFOSR.
Information fusion is the extraction of meaningful info from a variety of sources including physical data, cyber data, and human input. It has become even more vital in this era of big data, where we are forced to obtain reliable, meaningful information from data that is often noisy, corrupted, heterogeneous, and overabundant.
The research team will re-examine all levels of traditional fusion system design through a dynamic data driven lens. Their study has three main objectives:
explore a data driven approach to dynamically drive the sensing system toward the desired state while imposing a minimum RF footprint for low probability detection/intercept;
facilitate accurate and timely inference and control actions using data-driven learning of signal models, interference environment, and dynamically varying system objectives; and
tackle the computational challenges encountered in the proposed dynamic data driven information fusion paradigm.
The team’s research will create a new science for information fusion that is centered on the notion of a creating a dynamic and data driven information system that has tangible value for Department of Defense and Air Force mission.
Chen, the project’s principal investigator, says, “The project takes a data-driven approach to situational awareness. This departs from the traditional information fusion paradigm where predefined inference objectives drive the sensing hierarchy in data collection, processing, communications, and fusion. Such a data-driven approach inherently leads to a dynamic system where inference objectives are defined on the fly from the collected and processed data which are then used to drive subsequent data collection and processing.”