Modulation classification for orthogonal frequency-division multiplexing (OFDM) signals

This report proposed a comprehensive modulation classification system for recognizing the Orthogonal Frequency Division Mult iplexing (OFDM) signal and extracting its parameters. Since, OFDM is asymptotically Gaussian, a Gaussianity test is introduced first to distinguish multi-carrier signal (OFDM) from single carrier signal and Additive White Gaussian Noise (AWGN) by analysing the signal distribut ion. Modulation classifiers, including Maximum-Likelihood approach and the Sixth Order Cumulants method, are then presented to ident ify the modulat ion type for each sub-carrier of OFDM signal. The parameter extraction methods are also addressed: cyclic-correlation based sampling frequency estimation, correlation test to estimate the Cyclic Prefix duration, iterative approach to detect the frequency offset, as well as the Gaussianity test for detecting subcarriers number. The main contributions of report are to illustrate and to implement the three methods for estimating the symbol rate which are the Classical method, the Weighted approach, and the Filter-based cyclic correlation method. Simulations are provided to compare and evaluate these three methods. It is shown in simulation results that the Filter-based method is the most effective and accurate one in sampling frequency estimation for signals of different modulation types, such as QAM, 4QAM, 16QAM and 64QAM. All the simulations are carried out using Matlab.Bachelor of Engineerin