Estudo do uso de técnicas simples de inteligência artificial na classificação de sinais radar modulados

Abstract

In the modern military scenario, Electronic Warfare Capability becomes increasingly vital for maintaining national security and effectively achieving a nation's military and geopolitical objectives by providing a decisive strategic advantage and dissuasion power. In light of this, this paper is guided by the idea of making a contribution to the development of Electronic Warfare Support Measures for the Brazilian Navy. To achieve this goal, this research analyzes the efficiency of intrapulse modulation classification algorithms in radar signals and aims to serve as an initial step for a more in-depth study on the subject. Thus, this study encompasses a comparative analysis of three classifiers employing different Artificial Intelligence techniques (Elman Neural Networks, Convolutional Neural Networks, and Support Vector Machines). Furthermore, all three algorithms aim to perform accurate classifications of the investigated signal types, whether they are linearly frequency modulated or phase modulated using Barker codes, polyphase codes, or polytime codes. In this context, the correct classification of radar signal modulation allows for early identification of hostile emitters, thereby enhancing efficiency and expediting decision- making by authorities and reaction of combat systems. Therefore, the classification of modulated radar signals assumes paramount importance in threat detection and provides a significant advantage in the realm of Electronic Warfare for the nation.