Automatic Modulation Classification: A Deep Learning Enabled Approach

TLDR

Conventional AMC methods are either maximum‑likelihood or feature‑based, but ML‑AMC suffers from high computational complexity and feature‑based approaches require expert knowledge. The paper investigates AMC using deep learning and proposes an end‑to‑end CNN‑AMC that automatically extracts features from long symbol‑rate sequences and estimated SNR. CNN‑AMC uses a unit classifier to handle varying input sizes and employs a two‑step training scheme with transfer learning to ease training complexity. Simulation results show CNN‑AMC outperforms feature‑based methods, approaches optimal ML‑AMC performance, is robust to carrier phase offset and SNR estimation errors, and is 40–1700× faster than ML‑AMC during inference.

Abstract

Automatic modulation classification (AMC), which plays critical roles in both civilian and military applications, is investigated in this paper through a deep learning approach. Conventional AMCs can be categorized into maximum likelihood (ML) based (ML-AMC) and feature-based AMC. However, the practical deployment of ML-AMCs is difficult due to its high computational complexity, and the manually extracted features require expert knowledge. Therefore, an end-to-end convolution neural network (CNN) based AMC (CNN-AMC) is proposed, which automatically extracts features from the long symbol-rate observation sequence along with the estimated signal-to-noise ratio (SNR). With CNNAMC, a unit classifier is adopted to accommodate the varying input dimensions. The direct training of CNN-AMC is challenging with the complicated model and complex tasks, so a novel two-step training is proposed, and the transfer learning is also introduced to improve the efficiency of retraining. Different digital modulation schemes have been considered in distinct scenarios, and the simulation results show that the CNN-AMC can outperform the feature-based method, and obtain a closer approximation to the optimal ML-AMC. Besides, CNN-AMCs have the certain robustness to estimation error on carrier phase offset and SNR. With parallel computation, the deep-learning-based approach is about 40 to 1700 times faster than the ML-AMC regarding inference speed.