Data-Driven Performance Assessment and Process Management for Space Situational Awareness

Data Fusion & Neural Networks (DF&NN), LLC, Broomfield, CO 80020 Data-driven processes are becoming increasingly useful for decision support as large data sets become available, the computational resources become more powerful, and the data analysis approaches become more mature. Because the development of data-driven approaches only minimally relies on domain experts, they can provide decision support aids rapidly and with low development cost. Several ongoing projects in our development group are applying data-driven applications to help satellite operators quickly and confidently build real-time situational awareness during potentially dangerous events. Among the most significant challenges in building, testing and deploying data-driven applications are performance assessment (PA) issues. PA allows developers to determine whether a given solution meets expectations and requirements. In addition, because data-driven applications are often based on highly configurable algorithms with many non-obvious options (e.g. parameter settings, data sets, algorithm variants, etc.), PA can provide crucial feedback to the developer in tuning an approach and comparing alternatives. This paper discusses PA technologies and presents a semi-automated approach to multiobjective parameter optimization – all development was in the context of datadriven application tuning for Space Situational Awareness.

[1]  John Salerno.  Measuring situation assessment performance through the activities of interest score , 2008, 2008 11th International Conference on Information Fusion.

[2]  Christopher Bowman,et al.  Toward an Automated Situation Assessment for Abnormal Behavior of Satellites in Low Earth Orbit (LEO) , 2010 .

[3]  Andrew B. Whinston,et al.  A model of decision-making with sequential information-acquisition (Part 1) , 1986, Decis. Support Syst..

[4]  Christopher Bowman,et al.  Possibilistic versus probabilistic trade-off for data association , 1993, Defense, Security, and Sensing.

[5]  W. Wonham.  On the Separation Theorem of Stochastic Control , 1968 .

[6]  Y. Bar-Shalom.  Tracking and data association , 1988 .

[7]  A. N. Steinberg.  Data fusion system engineering , 2000, Proceedings of the Third International Conference on Information Fusion.

[8]  Nour-Eddin El Faouzi,et al.  Sensor and Data Fusion , 2012, Handbook of Soft Computing for Video Surveillance.

[9]  Yaakov Bar-Shalom,et al.  Estimation and Tracking: Principles, Techniques, and Software , 1993 .

[10]  Belur V. Dasarathy,et al.  Sensor fusion potential exploitation-innovative architectures and illustrative applications , 1997, Proc. IEEE.

[11]  James Llinas.  On the Scientific Foundations of Level 2 Fusion , 2004 .

[12]  Christopher Bowman.  The Dual Node Network (DNN) Data Fusion & Resource Management (DF&RM) Architecture , 2004 .

[13]  K. Kastella.  Joint multitarget probabilities for detection and tracking , 1997 .

[14]  D. L. Hall,et al.  Mathematical Techniques in Multisensor Data Fusion , 1992 .

[15]  James Llinas,et al.  Multisensor Data Fusion , 1990 .

[16]  Christopher Bowman,et al.  Statistical Methods for Correlating Space Environment Measurements and Effects amongst Multiple GEO Satellites , 2010 .

[17]  Jay B. Jordan,et al.  Comparative analysis of statistical, fuzzy, and artificial neural pattern recognition techniques , 1992, Defense, Security, and Sensing.

[18]  Christopher Bowman.  Engineering Resource Management Solutions by Leveraging Dual Data Fusion Solutions , 2010 .

[19]  Lawrence D. Stone,et al.  Bayesian Multiple Target Tracking , 1999 .