Graduation Year

2022

Document Type

Dissertation

Degree

Ph.D.

Degree Name

Doctor of Philosophy (Ph.D.)

Degree Granting Department

Electrical Engineering

Major Professor

Ravi Sankar, Ph.D.

Committee Member

Nasir Ghani, Ph.D.

Committee Member

Ismail Uysal, Ph.D.

Committee Member

Srinivas Katkoori, Ph.D.

Committee Member

Leslaw Skrzypek, Ph.D.

Keywords

Automatic Modulation Recognition, Feature Extraction Techniques, Analog Modulations, Digital Modulations, Principal Component Analysis

Abstract

A newly developed, near real-time, well-performing and potentially universally applicable Automatic Modulation Recognition (AMR) technique for discrimination of numerous modern modulated waveforms found in commercial as well as military communication systems applicable to the new Air Force’s Advanced Battle Management & Surveillance (ABMS) framework as well as Versatile Depot Automatic Test Station (VDATS) Test Program Set (TPS) development is presented. It involves generating complex feature vectors composed of High-Order Direct Cumulant, Cyclostationary and Fourier of Wavelet Transform features created with the help of Principal Component Analysis and Variance Data Compression.

Twelve modulated waveforms are used to evaluate the performance of the expanded feature vectors: eight commercial modulated waveforms [Quaternary Amplitude Shift Keying (QASK), Quaternary Frequency Shift Keying (QFSK), Quaternary Phase Shift Keying (QPSK), 16-Point Quadrature Amplitude Modulation (QAM-4,4), Gaussian Minimum Shift Keying (GMSK), Frequency Quadrature Amplitude Modulation (FQAM), Filter Bank Multi Carrier (FBMC) and Universal Filtered Multi Carrier (UFMC)], (Cosine) Binary Offset Carrier - BOC(1,1) - waveforms used in the European Galileo Navigation System and three waveforms utilized in defense military systems [Quaternary Linear Frequency Modulation (QLFM), Quaternary Pulse Width and Pulse Position Modulations (QPWM and QPPM)]. Generated complex feature vectors are categorized with the help of a neural network and compared with corresponding library feature patterns.

The presented experimental results are rather unprecedented in the literature since to the best of the authors knowledge, no research team has considered such a varied and comprehensive modulation format test set of waveforms in a single study. There have been some hierarchical modulation classifications schemes proposed in the earlier research dating to the early 2000s; however, their modulation format test sets were very limited and the classification performance for those that could be compared to the current research was dramatically lower. Also, there have been some attempts to generate the required probabilities for the other class of maximum-likelihood techniques generating very complex mathematical formulas in numerous cases for a handful of modulation formats. The latest research also addresses some of the feature-based non-hierarchical techniques but again the techniques presented are detached from one another and modulation format test sets are usually limited too. The research presented in this work gives a convenient and effective non-hierarchical method for modulation format classification that is potentially universally applicable to any conceivable modulated waveform that can be represented in the time domain while keeping the computational complexity reasonable so as to be applicable for real-time implementation in the future communications systems. The study described which feature-based techniques are broadly applicable and of these which are actually performing best and are readily suitable for easy implementation.

Download

Share

COinS