A Deep Neural Network Approach for Detection and Classification of GNSS Interference and Jammer

Abstract

<p>Global Navigation Satellite Systems (GNSS) are one of the most important infrastructures in the modern world, also enabling many critical applications that require the reliability of the received signals. However, it is well known that the power of the GNSS signals at the receiver's antenna is extremely weak, and radio-frequency interference affecting the GNSS bandwidths might lead to reduced positioning and timing accuracy or even a complete lack of the navigation solution. Therefore, in order to mitigate interference in the GNSS receivers and guarantee reliable solutions, interference detection and classification becomes of paramount importance. This paper proposes an approach for the automatic and accurate detection and classification of the most common interference and jammers based on the use of Convolutional Neural Networks (CNN). The input for the network is the visual time-frequency representation of the received signal, together with statistical features in the time and frequency domains. The time-frequency representation is obtained using both the Wigner-Ville and the short-time Fourier transforms. Moreover, the performance of the proposed method is compared using two different CNN architectures, AlexNet and ResNet. The effectiveness of the method is shown in two case studies: Monitoring and classification by a terrestrial monitoring station and from a Low Earth Orbit satellite (LEO). The results show that the proposed method has quite a high accuracy in detecting and classifying interference, even with low power, and can be implemented as a real-time tool for monitoring jammers.</p>

Optimal Machine Learning and Signal Processing Synergies for Low-Resource GNSS Interference Classification

“…The most popular approach is discrete Fourier transform (DFT)-based transforms [7], [16], [19], [20], [50]. The popularity is attributed chiefly to the processing efficient fast Fourier transform (FFT) implementation of the DFT.…”

“…ML is a popular modern choice for signal classification and shows good performance in various applications [2], [18], [19]. It especially shows improved resilience to classification in scenarios where the interference signals are affected by multipath [20], [21].…”

“…Fourth, ML models may become big, limiting their low size, weight, power, and cost (SWAP-C) use for edge devices. For example, convolutional neural networks (CNNs) are a popular choice for signal classification [19], [20], but they require multiple multi-dimensional convolutions to process the data, which is not particularly useful when optimizing for low-power systems like a single-board computer (SBC). A counter around it sends the data to computer clusters for processing, but it introduces communication overhead.…”

Deep temporal semi-supervised one-class classification for GNSS radio frequency interference detection

“…Radio frequency interference (RFI) is a major challenge in aviation when using the Global Navigation Satellite System (GNSS) for navigation and surveillance purposes (Scaramuzza et al, 2014(Scaramuzza et al, , 2015(Scaramuzza et al, , 2016(Scaramuzza et al, , 2017(Scaramuzza et al, , 2019Truffer et al, 2017;Ala'Darabseh and Tedongmo, 2019;Jonáš and Vitan, 2019;Morales Ferre et al, 2019;Liu et al, 2020aLiu et al, , 2020bLiu et al, , 2021Liu et al, , 2022Lukeš et al, 2020;Eurocontrol, 2021;Swinney and Woods, 2021;Mehr and Dovis, 2022;Ebrahimi Mehr and Dovis, 2023). Aircraft flying under instrument flight rules (IFR) increasingly depend on GNSS as a main navigational aid.…”

Deep Learning CNN-GRU Method for GNSS Deformation Monitoring Prediction

“…Meanwhile, the GRU model represents an improved model of the RNN, adept at capturing long-term dependencies in the dataset [35]. Previous studies have applied CNN for GNSS residual processing [36] and classification [37], GRU for landslide prediction [38] and multipath modeling [39], as well as CNN-GRU for forecasting wind power [40], particulate matter concentrations [41], and the pressure of a concrete dam [42]. The CNN-GRU model integrates the characteristics of CNN and GRU, which has better a performance in predictions of GNSS deformation monitoring.…”