Signal Recognition Based On Federated Learning

Signal modulation recognition is widely used in spectrum detection, channel estimation, and interference recognition, and is a prerequisite step for signal decoding and demodulation. With the development of artificial intelligence, great progress has been made in signal recognition using deep learni...

Full description

Saved in:
Bibliographic Details
Published inIEEE INFOCOM 2020 - IEEE Conference on Computer Communications Workshops (INFOCOM WKSHPS) pp. 1105 - 1110
Main Authors Shi, Jibo, Zhao, Haojun, Wang, Meiyu, Tian, Qiao
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.07.2020
Subjects
Convolution
Data models
federated learning
Machine learning
Modulation
modulation recognition
Privacy security
Servers
Training
Training data
Online AccessGet full text

Cover

More Information
Summary:Signal modulation recognition is widely used in spectrum detection, channel estimation, and interference recognition, and is a prerequisite step for signal decoding and demodulation. With the development of artificial intelligence, great progress has been made in signal recognition using deep learning methods. However, the method of deep learning requires a large amount of data. Today, with more and more emphasis on data privacy and security protection, there are usually hard-to-break barriers between data sources. This makes the data limited and of poor quality, which is not enough to support deep learning training. Federated learning may be a feasible direction to solve this problem. In this article, we will discuss signal modulation recognition based on federated learning, and the results show that an acceptable recognition rate is achieved while satisfying privacy protection and data security.
DOI:10.1109/INFOCOMWKSHPS50562.2020.9162958