Efficient Track-to-Task Assignment Using Cluster Analysis

Abstract

This paper reaches beyond the data fusion community to find a novel way to approach the assignment problem in track-to-track (T2T) data fusion. Cluster analysis techniques are explored as a source of assignment algorithms that are practical and efficient. The proposal to apply cluster analysis to the assignment problem is based on a re-examination of the unique requirements of track-to-track fusion. This paper will present the computational argument for a more sophisticated approach to T2T assignment. It will show how the assignment problem may be recast as a clustering problem. Clustering techniques that present promising approaches to T2T assignment are briefly surveyed. A prototype T2T fusion system is described which uses an agglomerative hierarchical clustering assignment algorithm. Experimental results from this system are presented that demonstrate the feasibility, usefulness and efficiency of the proposed approach