A feature fusion-based communication jamming recognition method

The electromagnetic environment is becoming increasingly complex, with communication jamming intensifying. Accurately identifying the type of jamming is essential for maintaining the integrity of communication and ensuring the safety of individuals and organizations. In this article, a jamming identification method based on feature fusion is proposed for the rapid and accurate identification of communication jamming. Five typical communication jamming signals are simulated and a set of features is extracted from the time domain, frequency domain, and time–frequency domain that are resistant to noise and distinguishable from one another. These features are then input into various classifiers, including support vector machine, K-nearest neighbor, decision tree model, and naive bayesian model, and their identification results are compared. It is shown that the combination of features selected has a strong classification performance for the five types of communication jamming signals. When the jamming-to-noise ratio is − 4 dB, the overall recognition accuracy of the method exceeds 90% and reaches 100% at 3 dB.