Signal Detection and Modulation Classification for Satellite Communications
A4 Artikkeli konferenssijulkaisussa
Proceedings of the 2020 3rd International Conference on Signal Processing and Machine Learning
AbstractAmateur ground stations are gaining increasing importance as both academic and hobby activities. However, due to the limited energy resources available in amateur satellites, ground stations need to be located in isolated places in order to establish a reliable communication. This usually implies limited Internet access. Hence, ground stations need to be able to recognize incoming signal without completely relying on an Internet connection. For this reason, we propose an algorithm to estimate parameters such as amplitude, center frequency, bandwidth and modulation type for amateur radio applications. For signal detection, we use an absolute-valued sinc approximation which estimates the center frequency and bandwidth of signals with signal-to-noise ratios over -6 dB with a precision of 5% and 2% respectively. In addition, Support Vector Machines (SVM) binary classifiers are used in series to classify the four most common modulation types used in amateur satellites. With accuracies over 90%, SVM outperforms solutions based on Artificial Neural Networks.
Modulation classification, signal detection, Spectrum sensing, Support vector machines, Artificial neural networks, Amateur Satellites
Toro Betancur , V , Carmona Valencia , A & Marulanda Bernal , J I 2020 , Signal Detection and Modulation Classification for Satellite Communications . in Proceedings of the 2020 3rd International Conference on Signal Processing and Machine Learning . , 3432297 , ACM , New York, NY, USA , pp. 114–118 , International Conference on Signal Processing and Machine Learning , Beijing , China , 22/10/2020 . https://doi.org/10.1145/3432291.3432297