Modulation type identification method based on sparse stack self-coding

Abstract

The invention discloses a modulation type identification method based on sparse stack self-coding. The method comprises the following steps: step 1, processing down-converted signals to obtain baseband signals of modulation signals; step 2, performing one-dimensional vectorization and normalization on the baseband signals; step 3, reforming a self-coder to obtain a sparse self-coder, and then constructing a stack sparse self-coder network using the reformed sparse self-coder to extract signal features; introducing a soft max regression model, and taking the signal features extracted by the stack sparse self-coder network as an input; step 4, training the stack sparse self-coder network; and step 5, judging whether the adjustment of network parameters meets the requirement, if so, ending the training, otherwise, repeating step 4 to continuously train the network. The method does not need to artificially extract signal feature parameters, thereby solving the problem of large parameter error in artificial extraction; and the method has good modulation type identification performance in a network having low signal-to-noise ratio and excellent anti-noise performance.

Description

technical field [0001] The invention belongs to the field of blind signal processing, and in particular relates to a modulation pattern recognition method based on sparse stack self-encoding. Background technique [0002] Modulation pattern identification of communication signals is an intermediate process of signal detection and demodulation. There are very important applications in both civilian and military fields, such as spectrum management, intelligent modem, signal confirmation, radio interception, electronic countermeasures, and threat analysis. Therefore, a large number of scholars at home and abroad have done in-depth research on modulation pattern recognition. At present, there are two main categories of communication signal modulation pattern recognition algorithms: the maximum likelihood method based on decision theory and the pattern recognition method based on feature extraction. [0003] The maximum likelihood method based on decision theory regards the mod...

AI Technical Summary

Problems solved by technology

[0007] The purpose of the present invention is to provide a modulation pattern recognition method based on sparse stack self-encoding to process the original signal received by down-conversion in order to solve the problem of complex feature extraction and low recognition rate under low signal-to-noise ratio in the identification of signal modulation methods , to complete feature extraction and modulation pattern recognition

Images