CFAR Target Detection Based on Gumbel Distribution for HF Radar

Abstract

High frequency (HF) radars are capable to detect and track targets at extremely long ranges. But the signal environment that includes external noise, different kinds of clutter and interference will significantly limit the detection and system capability. This paper considers a new approach to solve the target detection problem in a complex HF radar signal environment. It uses conventional constant false-alarm-rate (CFAR) detection procedure but based in this case on an untypical assumption of Gumbel distributed power spectrum values along range cells for thresholding. The CFAR detection test rule was combined with a local peak determination procedure. The proposed detection scheme was tested using real HF radar data and gave very promising results.