Towards Scalable Automatic Modulation Classification via Meta-Learning
- Authors
- Jang, Jungik; Pyo, Jisung; Yoon, Young-Il; Seo, Sang Yong; Lee, Eun Jae; Jung, Gyeong Hun; Choi, Jaehyuk
- Issue Date
- Oct-2023
- Publisher
- IEEE
- Keywords
- Automatic modulation classification; few-shot learning; meta-learning; Transformer; unseen dataset
- Citation
- MILCOM 2023 - 2023 IEEE MILITARY COMMUNICATIONS CONFERENCE
- Journal Title
- MILCOM 2023 - 2023 IEEE MILITARY COMMUNICATIONS CONFERENCE
- URI
- https://scholarworks.bwise.kr/gachon/handle/2020.sw.gachon/91885
- DOI
- 10.1109/MILCOM58377.2023.10356268
- ISSN
- 2155-7578
2155-7586
- Abstract
- Driven by recent technological breakthroughs in deep learning (DL), many recent automatic modulation classification (AMC) methods utilize deep networks to classify the type of modulation in the incoming signal at the receiver. However, existing DL-based approaches suffer from limited scalability, especially for unseen modulations or input signals from new environments not used in training the DL model, thus not ready for real-world systems such as software defined radio devices. In this paper, we introduce a scalable AMC scheme that provides flexibility for new modulations and adaptability to input signals with diverse configurations. We propose a meta-learning framework based on few-shot learning (FSL) to acquire general knowledge and a learning method for AMC tasks. This approach allows the model to recognize new unseen modulations by learning with only a very small number of samples, without requiring the entire model to be retrained. Additionally, we enhance the scalability of the classifier by leveraging a transformer-based encoder, enabling efficient processing of input signals with varying configurations. Extensive evaluations demonstrate that the proposed AMC method outperforms existing techniques across all signal-to-noise ratios (SNRs) on RadioML2018.01A dataset.
- Files in This Item
- There are no files associated with this item.
- Appears in
Collections - ETC > 1. Journal Articles
![qrcode](https://api.qrserver.com/v1/create-qr-code/?size=55x55&data=https://scholarworks.bwise.kr/gachon/handle/2020.sw.gachon/91885)
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.