A communication signal modulation recognition method based on LMD decomposition intelligent analysis

By combining local mean decomposition and improved residual network, the problem of signal distortion in signal modulation recognition is solved, achieving efficient and noise-resistant recognition of communication signals and improving the accuracy of signal modulation method recognition.

CN116232820BActive Publication Date: 2025-11-07XIAN INSTITUE OF SPACE RADIO TECH
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202210899383.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-07-28
Publication Date
2025-11-07
Estimated Expiration
2042-07-28

AI Technical Summary

Technical Problem

Existing empirical mode decomposition methods suffer from problems such as over-envelope, under-envelope, and mode confusion in signal modulation identification, leading to signal distortion and making it difficult to effectively identify the modulation mode of communication signals.

Method used

Local mean decomposition (LMD) combined with an improved residual network is used to decompose the communication signal into multiple PF components through local mean decomposition, construct a feature matrix and convert it into a two-dimensional image, and use the improved residual network to identify the modulation mode.

Benefits of technology

It effectively identifies the modulation mode of communication signals, has noise resistance, improves the accuracy and robustness of signal recognition, overcomes the shortcomings of traditional methods, and achieves efficient recognition under low signal-to-noise ratio.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116232820B_ABST
    Figure CN116232820B_ABST
Patent Text Reader

Abstract

The application discloses a communication signal modulation recognition method based on LMD intelligent analysis, belongs to the technical category of signal intelligent processing, is widely applied to the field of electronic reconnaissance information acquisition, and is used for intelligent recognition of the modulation mode of intercepted target signals in electronic reconnaissance. The application introduces the thought of deep learning in the traditional modulation recognition, first analyzes the LMD characteristics of common communication frequency modulation signals, and constructs a new type of feature matrix composed of partial components after decomposition. The matrix is an effective feature domain capable of representing the individual differences of the frequency modulation signals in multiple levels, and the feature domain is not sensitive to noise. On the basis, the feature matrix is combined with an improved residual neural network (ResNet), intelligent feature extraction and nonlinear mapping classification methods in deep learning are used, and effective recognition of the communication frequency modulation signals is realized. The method overcomes defects such as poor adaptability caused by artificial feature extraction and threshold setting, has low sample acquisition complexity, and has high intelligence.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Method for identifying digital modulation signals in presence of complicated noise

    CN104052702A

  • Identification method of modulated signals based on LMD approximate entropy, high-order cumulants, and SVM

    CN107395540A