Stable filters: A robust signal processing framework for heavy-tailed noise

Abstract

The radar systems of the twenty-first century are being pushed to improve signal detection beyond current capabilities. Current systems perform poorly when the clutter departs from the Gaussian model and becomes heavy-tailed. We introduce stable filters as a general framework for signal processing in heavy-tailed scenarios. The performance gains attainable from these filters can provide significant improvements to digital systems in impulsive noise. Stable filters make use of advanced numerical optimizations to overcome the computational complexity that is common in non-linear approaches. Current performance results show superior performance of stable filters when compared with conventional linear approaches.