Data Fusion Cube: A Multi-Dimensional Perspective

In the classical sense, data fusion can be viewed as a one-dimensional entity having five distinct levels. However, this view does not convey the multi-dimensional aspect of data fusion. This paper argues that data fusion is not one-dimensional, but rather a three-dimensional entity. These three attributes are sensor fusion, system fusion and information fusion. Sensor fusion can be thought of as taking the raw sensor data and fusing it together so it seems to have come from a single sensor. System fusion can be thought of as combining the output of various heterogeneous systems together into a single fused output. Information fusion can be thought of as taking information gathered from various sources and fusing it into a single output. This paper provides an overview and discussion of this three-dimensional perspective of data fusion in order to illustrate its multi-faceted capabilities and applications. For each of the three dimensions, a definition and possible application, along with a discussion and comparison to the classical Level-0 to Level-4 levels of data fusion, is presented. Finally, this new data fusion cube is offered for consideration in which each axis (Sensor, System, Information) has a corresponding relationship to the classical aspects of data fusion. Presented at the Command and Control Research and Technology Symposia held at Monterey, CA on 11-13 Jun 2002. The original document contains color images.