Summaries - Office of Research & Innovation
Back Planning Big-Data Distributed Processing for Combat ID
|Division||Research & Sponsored Programs|
|Department||Naval Research Program|
Schwamm, Noboru E.
Rowe, Neil C.
|Sponsor||NPS Naval Research Program (Navy)|
Many architectures proposed for doing careful or deep analysis of combat ID information require centralized aggregation of information. This is important for analysis of long-term phenomena such as patrol patterns, but may provide a bottleneck for getting timely information back to the "edge" or the sensor-equipped platforms such as aircraft that need it for tactical purposes. Furthermore, a centralized site can be a single point of failure, bandwidth to it may be insufficient for full use, and overdependence on it may reduce reliability of Navy operations.
An alternative is to develop methods for either distributing some of the processing of a centralized node to the edge or by giving the sensor platforms additional data and software for analytic capabilities. For instance, if we push data to platforms about what adversary units have been operating in their area recently, software can compare sensor readings to history to recognize adversary platforms, and provide operators with immediate feedback about potential threats, faster than trying to contact a centralized site.
This requires a plan for how to push processing onto distributed nodes using modern ideas about big-data processing. What processing can be pushed requires careful analysis, since some processing requires too much data to transfer given bandwidth limits, some processing is too speculative to provide useful guidance, and some processing is too computationally demanding to be doable on the limited resources of an edge node such as an aircraft. It is also important for nodes on the edge to be able to request specific data pulls. We will examine the factors and come up with a set of numerically based criteria for making decisions about what processing can be distributed and to where. We will then build a simulation to test the main conclusions in simple scenarios.
|Publications||Publications, theses (not shown) and data repositories will be added to the portal record when information is available in FAIRS and brought back to the portal|
|Data||Publications, theses (not shown) and data repositories will be added to the portal record when information is available in FAIRS and brought back to the portal|