Communication Interference Recognition Based on Multiscale Convolutional Feature Extraction

In the present complex electromagnetic environment, useful signals are subjected to various interference. How to suppress interference is an urgent problem that needs to be solved. For the premise of the suppression of different interference, we need to clarify the type of interference first. In order to solve the problem of incomplete extraction of existing interference signal features and complex identification of interference types by machine learning. A multiscale neural network model is constructed to explore various depth features of interference signals. Attention mechanism is introduced to enhance the expression of important features, so as to improve the performance of our network to recognize interference types. The effects of noise, frequency offset and delay on recognition accuracy are verified in experiments. The experimental results show that our proposed method is able to achieve 96.3% accuracy when ISR=20dB, which is a significant improvement compared to existing methods through machine learning.