When is Cognitive Radar Beneficial?

TL;DR

This paper compares reinforcement learning-based cognitive radar with rule-based strategies in dynamic spectrum access, analyzing when learning methods outperform fixed rules based on channel conditions and problem horizon.

Contribution

It provides analytical and empirical insights into the conditions under which online reinforcement learning benefits cognitive radar performance.

Findings

Learning-based approaches generalize better in realistic scenarios.

Short time-horizon problems may hinder learning performance.

Analytical methods via stochastic dominance offer insights for simple cases.

Abstract

When should an online reinforcement learning-based frequency agile cognitive radar be expected to outperform a rule-based adaptive waveform selection strategy? We seek insight regarding this question by examining a dynamic spectrum access scenario, in which the radar wishes to transmit in the widest unoccupied bandwidth during each pulse repetition interval. Online learning is compared to a fixed rule-based sense-and-avoid strategy. We show that given a simple Markov channel model, the problem can be examined analytically for simple cases via stochastic dominance. Additionally, we show that for more realistic channel assumptions, learning-based approaches demonstrate greater ability to generalize. However, for short time-horizon problems that are well-specified, we find that machine learning approaches may perform poorly due to the inherent limitation of convergence time. We draw conclusions as to when learning-based approaches are expected to be beneficial and provide guidelines for future study.