A Novel Clustering Algorithm for Faster Passive Sensor Measurement Association

Abstract

This paper proposes a novel clustering algorithm for static passive sensor measurement association. The algorithm clusters line of sight (LOS) measurements based on the valid solutions of pairwise triangulation between measurements of different sensors. Each valid triangulation solution will give a range value along the LOS as possible target location. Clusters are formed based on the pairwise distance between these range values. The criteria for the validity of a triangulation solution and cluster formation depend on the measurement noise of sensors and geometry of possible target and sensor locations. Intensity of space intersection and association cost of each cluster are used to select association solutions among available clusters. Several algorithms for passive sensor measurement association exist in literature like S-D assignment, Seq(2-D), $\mathrm{S}_{o}-\mathrm{D}+\text{Seq}(2-\mathrm{D})$ and algorithms based on intensity of space intersection, however, they are still cumbersome to be implemented in real time. An effort has been made through this novel algorithm to reduce computation time for a feasible real time implementation.