Dualformer: Efficient Feature Extractor for Complex-valued Blind Communication Signal Analysis

2026-06-30

AI Summary

This work addresses the challenge of efficient feature extraction in blind communication signal analysis by proposing DualNN, a dual-branch neural network that, for the first time, shares parameters between the real and imaginary components of complex-valued signals. Theoretical analysis demonstrates that this parameter-sharing strategy reduces generalization error while preserving representational capacity. Building upon this foundation, the authors further introduce Dualformer, a Transformer-based architecture that treats segmented signal blocks as tokens and integrates parameter sharing with modular design to enable multi-granularity feature extraction. The proposed framework provides a unified solution for diverse tasks including automatic modulation classification, signal scheme identification, and structural parsing, significantly outperforming existing Transformer and conventional deep learning models across multiple benchmarks. Moreover, it exhibits strong generalization to challenging scenarios such as blind source separation and low-SNR spectrum sensing.

Abstract

Designing effective feature extractors is critical for blind signal analysis tasks such as automatic modulation recognition (AMR), signal scheme recognition (SSR), and \color{black} signal structure parsing (SSP). In this work, we propose dual-channel neural network (DualNN) that efficiently exploits complex-valued signals through parameter sharing across IQ channels. Unlike traditional real-valued or complex-valued models, DualNN is a groundbreaking framework which shares the network parameters for processing the real and imaginary parts of the complex-valued signals, and is theoretically shown to reduce generalization error while preserving expressive capacity. Specifically, we propose a novel Transformer-based architecture to implement DualNN, called Dualformer. The Dualformer segments input signals into patch-level tokens and captures multi-granularity features, enabling robust performance across diverse signal analysis tasks. Furthermore, we conduct extensive experiments comparing Dualformer with three Transformer-based baselines and four conventional DL-based approaches. Results demonstrate consistent performance improvements on AMR, SSR, and SSP tasks. Besides, the modular design of DualNN allows it to generalize well to blind signal processing tasks such as blind source separation and low-SNR spectrum sensing. This work paves the way for a broader application of DualNN architectures in unsupervised and weakly supervised complex-valued signal analysis scenarios.

Problem

• blind signal analysis
• complex-valued signals
• feature extraction
• automatic modulation recognition
• signal structure parsing

Innovation

• DualNN
• complex-valued signal processing
• parameter sharing
• Transformer architecture
• blind signal analysis