Exploiting Temporal Structures of Cyclostationary Signals for Data-Driven Single-Channel Source Separation

TL;DR

This paper addresses single-channel source separation of cyclostationary signals using a data-driven deep learning approach, demonstrating near-optimal performance with reduced computational complexity.

Contribution

It introduces a U-Net based deep learning method for cyclostationary source separation, outperforming classical approaches and approaching theoretical optimal bounds.

Findings

U-Net approach achieves near-optimal separation performance.

The method reduces computational burden compared to traditional techniques.

Theoretical bounds guide the design of effective deep learning architectures.

Abstract

We study the problem of single-channel source separation (SCSS), and focus on cyclostationary signals, which are particularly suitable in a variety of application domains. Unlike classical SCSS approaches, we consider a setting where only examples of the sources are available rather than their models, inspiring a data-driven approach. For source models with underlying cyclostationary Gaussian constituents, we establish a lower bound on the attainable mean squared error (MSE) for any separation method, model-based or data-driven. Our analysis further reveals the operation for optimal separation and the associated implementation challenges. As a computationally attractive alternative, we propose a deep learning approach using a U-Net architecture, which is competitive with the minimum MSE estimator. We demonstrate in simulation that, with suitable domain-informed architectural choices, our U-Net method can approach the optimal performance with substantially reduced computational burden.