Intelligent agents for radar systems

Abstract

This article describes a system using multilevel agents which offer a way of tracking radar objects that fly close to the ground or are otherwise difficult to detect. Our research uses a pre-track system that exploits spatio-temporal Doppler correlations to help reduce ghost targets, as well as reducing false alarms due to noise. Further, it makes use of intelligent agents. The two main processing methods we have used are dynamic programming and Hough transforms. In summary, the system is a simplified multiple hypothesis tracker, tightly coupled to a self-adaptive, context sensitive, spatio-temporal CFAR system. In environments with diverse clutter characteristics, the self-adaptive nature of the agent system self-organises using simple processing and by assuming that there will be too few data measurements to establish the clutter statistics accurately, a robust sub-optimal solution is formed.