CFARnet: deep learning for target detection with constant false alarm rate

TL;DR

This paper introduces CFARnet, a deep learning framework that approximates classical CFAR detectors, enabling flexible tradeoffs between false alarm rate and detection accuracy in target detection tasks.

Contribution

It develops a neural network-based approach for CFAR detection, bridging the gap between classical methods and data-driven deep learning techniques.

Findings

CFARnet achieves a flexible balance between CFAR and detection accuracy.

Theoretically, CFARnet's detector is asymptotically equivalent to the classical GLRT.

Experimental results demonstrate CFARnet's effectiveness across various target detection scenarios.

Abstract

We consider the problem of target detection with a constant false alarm rate (CFAR). This constraint is crucial in many practical applications and is a standard requirement in classical composite hypothesis testing. In settings where classical approaches are computationally expensive or where only data samples are given, machine learning methodologies are advantageous. CFAR is less understood in these settings. To close this gap, we introduce a framework of CFAR constrained detectors. Theoretically, we prove that a CFAR constrained Bayes optimal detector is asymptotically equivalent to the classical generalized likelihood ratio test (GLRT). Practically, we develop a deep learning framework for fitting neural networks that approximate it. Experiments of target detection in different setting demonstrate that the proposed CFARnet allows a flexible tradeoff between CFAR and accuracy.