An Approach Based on Multi-Entropy and Artificial Intelligence for Communication Modulation Signal Identification

With the increasingly complex electromagnetic influence in the modern communication environment and the explosive growth of communication business data, the modulation recognition system needs to constantly learn new modulation categories and signal characteristics in the process of data increment, and gradually acquire stronger modulation recognition capability. This paper proposes a communication modulation signal recognition method based on multi-entropy and artificial intelligence technology. Firstly, the power entropy, singular spectral entropy, sample entropy, fuzzy entropy, and permutation entropy of the signal are calculated, and a composite entropy feature calculation method based on entropy weight method is proposed to further study the feature extraction of the signal. Finally, it is found that when using neural network for learning, without feature extraction, the recognition accuracy rate is only 70.7^%, while using the feature extraction method proposed in this paper, the accuracy rate can reach 89.5^%, and the recognition speed is significantly improved. In addition, the feature extraction method proposed in this paper can improve all kinds of modulation recognition based on artificial intelligence.