Classification of communication and navigation systems and jammer signals using deep learning algorithms

With the developing technology, the number of signal-emitting devices used for various purposes in many fields is increasing day by day. While these devices can be beneficial, they can also be used for intentional signal jamming and deception, causing harm to high-cost sectors like aviation. Considering the critical applications where time is of the essence, it is extremely important to classify and identify a large number of signals in a short time and to distinguish interfering signals from them. This study within a certain region aims to develop a model for the classification and identification of signal emitters by processing a large number of signal data collected from ground stations in a short period. The goal of the developed classification model is to propose an analytical approach that minimizes human workload and cost. In the proposed approach, the classification process is carried out using three different deep learning networks (AlexNet, ResNet, UNet) along with K-Means and K-Nearest Neighbors algorithms, taking into account the bandwidth, wavelength, power, and amplitude values of GSM, radar, radio, and jammer signals. The purpose of using Deep Learning and Machine Learning simultaneously in the study is to determine the architecture or algorithm that provides the best performance. After the classification process, the processing times and accuracy rates of the networks used were compared. When the developed classification model was run with sampled datasets created from real data, a high success rate was achieved in with the sampled dataset.