Hybrid clutter canceler with feedback guided predicative filtering on a heterogeneous parallel processing system

In this paper we describe a novel clutter cancellation platform based on a two stage approach that combines a feedback guided predictive front-end hybrid clutter canceller with high performance back-end filtering and target detection. The front-end architecture is based on an FPGA implementation of a Kalman filter that predicts target locations in real time and removes the target signals from the incoming data prior to hybrid cancellation. The back-end is user configurable and exploits high performance GPU and multi-core parallel hardware to simultaneously compute multiple clutter suppression and target detection algorithms coupled to an intelligent selection strategy for selecting the most accurate result. These target locations are fed back to the FPGA Kalman filter periodically to update the target predictions.