Preliminary Design Review: Kinematic Tracking for the PCA Integrated Radar-Tracker Application

Abstract

A kinematic tracker is a classical tracking system. The idea behind kinematic tracking is to store positions and velocities and then, using this information, associate incoming targets with the stored profiles. Using these new targets, the tracker updates its information about positions and velocities to prepare to do the entire process again when it is given the next set of targets.

Polymorphous Computing Architecture (PCA) Kernel-Level Benchmarks

“…The pattern matching kernel is extracted from the feature-aided tracking portion of the integrated radar-tracker application [2]. Fundamentally, this step entails overlaying two length-N vectors a and t and computing a metric that quantifies the degree to which these two vectors match.…”

“…The major motivation for this is to avoid generating the large amount of data necessary for the database. A typical record from a feature-aided track application is on the order of 650 bytes per record (see document PCA-IRT-4, [2]), and test cases of interest may require up to 100,000 such records. Thus, in the benchmark, we do not actually generate and maintain the contents of the database itself, only the indexing structures.…”

Morphing Scenarios for the GMTI Portion of the PCA Integrated Radar Tracker

“…Either piece may be run separately or combined by feeding GMTI output to KT as input. Complete descriptions of GMTI and KT may be found in [3] and [2], respectively. A very high level overview of GMTI will be offered in this section.…”