Neural Network Aided False and Ghost Detection Suppression

This document examines the efficacy of intentionally lowering the detection and binary integration thresholds in an effort to improve the radar's detection performance while mitigating the number of false alarms generated. The increased number of false alarms introduced by lowering the detection thresholds are suppressed by a neural network added to the signal processing chain after target disambiguation to label targets as real or false. The performance of standard processing techniques are compared to the proposed architecture and it is shown through simulation that probability of detection and number of false alarms of the proposed architecture can outperform the standard processing method.