Bazı modülasyon türlerinin sınıflandırılmasında yüksek mertebeden istatistiksel özelliklerin performans analizi
Loading...
Date
2020
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Open Access Color
OpenAIRE Downloads
OpenAIRE Views
Abstract
Modülasyon Sınıflandırma algoritmaları, alıcıda elde edilen sinyalin modülasyon tipini belirlemek ve uygun demodulator seçimi için kullanılır. Özellik tabanlı ve Olabilirlik tabanlı olmak üzere 2 tür vardır. Bu tezde yapı olarak daha az karmaşık olan FB yöntemi kullanılmıştır. 12 Analog ve Dijital Modülasyon tipli sinyalleri sınıflandırmak için algoritma geliştirilmiştir. İstatistiksel özellikler, Yüksek Dereceli Momentler ve Yüksek Dereceli Kümülantlar kullanılmıştır. Havadan kaydedilen ve sentetik simüle kanal etkileri eklenen sinyaller Lineer, Kuadratik ve Kübik Destek Vektör Makinesi (DVM) ile sınıflandırıldı. SNR'de 0 dB ile 20 dB arasında incelenen sinyallerin sınıflandırma performansı sunulmuştur. Performansın 10 dB ve 20 dB arasında kararlı olduğu ve yaklaşık %73, en yüksek performansın ise Karesel SVM'de 12dB'de % 75.5 olduğu gözlenmiştir. Bu tezde, geliştirilen algoritmasının sınırları, 12 modülasyon tipinin özellikleri ve SVM yapısı ile başarılı bir şekilde sunulmuştur.
Modulation Classification algorithms are used to determine the modulation type of signal obtained at the receiver and to use the appropriate demodulator. There are 2 types as Feature-based(FB) and Likelihood-based(LB). In this thesis, FB method is used, which is less complex in structure. Algorithm has been developed to classify the signals that were modulated by 12 Analog and Digital Modulation types. Statistical features, Higher-order Moments(HOMs) and Higher-order Cumulants(HOCs) were used as features. Signals, which are recorded as over-the-air adding synthetic simulated channel effects, were classified with Linear, Quadratic, and Cubic Support Vector Machine(SVM). The classification performance of the signals examined at SNR from 0 dB to 20 dB were presented. As a result, the classification performance was found to be stable between 10 dB and 20 dB and is approximately 73%. The highest value of performance was observed in Quadratic SVM as 75.5% at 12dB. In this thesis, the limits of the developed modulation classification algorithm successfully presented with the features and SVM structure of 12 modulation types.
Modulation Classification algorithms are used to determine the modulation type of signal obtained at the receiver and to use the appropriate demodulator. There are 2 types as Feature-based(FB) and Likelihood-based(LB). In this thesis, FB method is used, which is less complex in structure. Algorithm has been developed to classify the signals that were modulated by 12 Analog and Digital Modulation types. Statistical features, Higher-order Moments(HOMs) and Higher-order Cumulants(HOCs) were used as features. Signals, which are recorded as over-the-air adding synthetic simulated channel effects, were classified with Linear, Quadratic, and Cubic Support Vector Machine(SVM). The classification performance of the signals examined at SNR from 0 dB to 20 dB were presented. As a result, the classification performance was found to be stable between 10 dB and 20 dB and is approximately 73%. The highest value of performance was observed in Quadratic SVM as 75.5% at 12dB. In this thesis, the limits of the developed modulation classification algorithm successfully presented with the features and SVM structure of 12 modulation types.
Description
Keywords
Elektrik ve Elektronik Mühendisliği, Analog modülasyonlar, Electrical and Electronics Engineering, Analog modulations, Sayısal modülasyon sistemleri, Digital modulation systems
Turkish CoHE Thesis Center URL
Fields of Science
Citation
WoS Q
Scopus Q
Source
Volume
Issue
Start Page
0
End Page
67