Position-Aware Self-supervised Representation Learning for Cross-mode Radar Signal Recognition

TL;DR

This paper introduces RadarPos, a position-aware self-supervised learning framework that improves radar signal recognition by modeling pulse-level temporal dynamics, enhancing robustness and generalization in diverse electromagnetic environments.

Contribution

The paper presents a novel position-aware self-supervised approach for radar signal recognition that captures semantic dependencies without complex data augmentations.

Findings

Enhanced discriminability and robustness in radar recognition tasks.

Improved generalization to unseen radar modes in long-tailed settings.

Effective modeling of pulse-level temporal dynamics without complex augmentations.

Abstract

Radar signal recognition in open electromagnetic environments is challenging due to diverse operating modes and unseen radar types. Existing methods often overlook position relations in pulse sequences, limiting their ability to capture semantic dependencies over time. We propose RadarPos, a position-aware self-supervised framework that leverages pulse-level temporal dynamics without complex augmentations or masking, providing improved position relation modeling over contrastive learning or masked reconstruction. Using this framework, we evaluate cross-mode radar signal recognition under the long-tailed setting to assess adaptability and generalization. Experimental results demonstrate enhanced discriminability and robustness, highlighting practical applicability in real-world electromagnetic environments.