On the Scientific Foundations of Level 2 Fusion
Abstract
The JDL Data Fusion process model has served the Data Fusion (DF) community well as a means for describing and understanding the nature of the DF information process. However, it has not yet been examined as to its compatibility with a distributed information and fusion environment. Distributed Data Fusion (DDF) will no doubt form an important underpinning for the realization of Network-Centric Warfare, and it is in this motivational context that the preliminary ideas herein are offered. Other ideas involving suggested modifications to the model also are offered. This Keynote speech addresses two focal topics: (1) suggestions for extensions and modifications to the well-known JDL Data Fusion process model, and (2) perspectives on an overarching approach toward developing first Theories and then Algorithms for Level 2 Data Fusion. Four major extensions to the JDL model in terms of DDF are suggested. These extensions are related to the Pedigree function, an Adjudication function, the need for both local fusion and network-based fusion algorithms, and to the sharing of information across the nodes of the network. The authors also see a deficiency in the current JDL model regarding the role of quality control and inter-level processing. They suggest repairing this by making obvious the need for internodal bi-level processing. They also envision a two-step quality control process before a node issues an estimate: one that checks whether a just-formed estimate is of adequate quality, and a Belief Revision process that checks for both internodal consistency and consistency with recent global estimates. Lastly, the authors suggest that the basic DF model be modified to make obvious the co-processing of both inductively based and deductively based inferencing processes. The remainder of the paper focuses on theories and algorithms for level 2 data fusion. Fifty-three briefing charts summarize the presentation.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 2004
- Accession Number
- ADA428555
Entities
People
- James Llinas
Organizations
- University at Buffalo