Communication signal modulation classification system and method based on deep learning

Abstract

The invention discloses a communication signal modulation classification system and method based on deep learning. The method includes a unified representation of a communication signal, including thefollowing steps: step one, applying big data and deep mining in the monitoring of a complex electromagnetic environment to make the monitoring process more intelligent; and step two, converting the unified representation of communication signals to a database to enable the environmental monitoring modulation identification framework to change, thereby solving the problem that the previous increase in the monitoring signal leads to changes in the identification structure and change the identification structure. The system and the method have the ability of deep learning; the signal classification representation does not depend on limited learning samples; the increase in the signal type does not need to change the modulation identification structure, and the scalability and the versatilityare good; and the system and the method can be used in radio signal monitoring and can also be used in other application areas requiring signal modulation mode identification or parameter extraction.

Description

technical field [0001] The present invention relates to a communication signal modulation classification system and method, in particular to a communication signal modulation classification system and method based on deep learning. Background technique [0002] Traditional spectrum monitoring is faced with a sparse signal environment, so it can be analyzed one by one according to the prior knowledge of the signal, and can achieve good results. With the extensive use of wireless devices and the increase of signal types, current spectrum monitoring faces a more complex and unknown electromagnetic environment. By studying the latest theory of big data and deep learning, the present invention proposes a new method for unified expression of wireless signal features, uses this expression for modulation recognition, breaks through the key technology of using limited learning samples to recognize unknown radio signals, and improves the understanding of radio signals. The knowledge ...

AI Technical Summary

Problems solved by technology

With the extensive use of wireless devices and the increase of signal types, current spectrum monitoring faces a more complex and unknown electromagnetic environment

By studying the latest theory of big data and deep learning, the present invention proposes a new method for unified expression of wireless signal features, uses this expression for modulation recognition, breaks through the key technology of using limited learning samples to recognize unknown radio signals, and improves the understanding of radio signals. Knowledge acquisition and value discovery capabilities of big data to solve the problem that the current spectrum monitoring cannot effectively understand the complex electromagnetic environment

Images