Modeling of recoverable radioelectronic systems with redundancy

Abstract

This paper explores the features of applying a neural network approach to solve the problem of recognizing intra-pulse modulation of radar signals. The relevance of the problem is highlighted in the context of radio monitoring and analysis of the electronic environment. The methodology for building the features used in classification is described, including the stages of forming a training dataset, adding noise with different signal-to-noise ratio (SNR) levels, and implementing an algorithm for assessing model accuracy. It is demonstrated that the developed algorithm effectively classifies intra-pulse modulation signals, such as simple radio pulses, linear frequency-modulated signals (up-chirp and down-chirp), and that the proposed algorithm effectively solves the classification problem even under significant noise levels. To test the relevant approaches, software was implemented in the Python programming language. Its operational features are noted. The obtained results confirm the relevance and applicability of the method for radio monitoring tasks in real-world electronic environment conditions.