Joseph Coughlin, Stinger Ghaffarian Technologies, Inc., Rohit Mital, Stinger Ghaffarian Technologies, Inc., Shashi Nittur, Stinger Ghaffarian Technologies, Inc., Benjamin SanNicolas, Stinger Ghaffarian Technologies, Inc., Christian Wolf, Stinger Ghaffarian Technologies, Inc., Rinor Jusufi, Stinger Ghaffarian Technologies, Inc.
Keywords: Big Data, Analytics, Sensor analysis, SSN
Abstract:
Operational analytics when combined with Big Data technologies and predictive techniques have been shown to be valuable in detecting mission critical sensor anomalies that might be missed by conventional analytical techniques. Our approach helps analysts and leaders make informed and rapid decisions by analyzing large volumes of complex data in near real-time and presenting it in a manner that facilitates decision making. It provides cost savings by being able to alert and predict when sensor degradations pass a critical threshold and impact mission operations. Operational analytics, which uses Big Data tools and technologies, can process very large data sets containing a variety of data types to uncover hidden patterns, unknown correlations, and other relevant information. When combined with predictive techniques, it provides a mechanism to monitor and visualize these data sets and provide insight into degradations encountered in large sensor systems such as the space surveillance network. In this study, data from a notional sensor is simulated and we use big data technologies, predictive algorithms and operational analytics to process the data and predict sensor degradations. This study uses data products that would commonly be analyzed at a site. This study builds on a big data architecture that has previously been proven valuable in detecting anomalies. This paper outlines our methodology of implementing an operational analytic solution through data discovery, learning and training of data modeling and predictive techniques, and deployment. Through this methodology, we implement a functional architecture focused on exploring available big data sets and determine practical analytic, visualization, and predictive technologies.
Date of Conference: September 20-23, 2016
Track: Poster