Fusion of Asynchronous, Parallel, Unreliable Data Streams

Report No. ARL-TR-5344
Authors: Ann Bornstein; John Brand; Michelle McVey; Andrew Neiderer
Date/Pages: September 2010; 70 pages
Abstract: This report documents progress in an investigation of the fusion of parallel data streams composed of dissimilar data types of varying reliability, arriving at different times. The technique will serve to determine resemblance between objects, in this case, people, which can be described by a stream of data expressed as state vectors. Objects described by similar state vectors will group together. The data fusion methodology is based on the use of multidimensional scaling to evaluate data composed of attribute vectors of high-spatial dimensionality. A focus problem has been used for development of the technique, identification of high-value individuals (HVIs). The technique applies to other problems as well; the identification of HVIs is only one topical application. Data available to field operators and analysts includes traditional identification information, biometric identity files, situational and observational data, archival data, and intelligence information. These data may be used to classify individuals in terms of resemblance to key groups and cue an analyst to whether they are HVIs. A proof of concept of the methodology has been conducted with promising results.
Distribution: Approved for public release
  Download Report ( 1.112 MBytes )
If you are visually impaired or need a physical copy of this report, please visit and contact DTIC.

Last Update / Reviewed: September 1, 2010