Blind symbol rate and modulation type estimations of linear modulated signals received from an unknown signal

In this study, blind symbol rate and modulation type estimations have been proposed to identify the linear modulated signals received from an unknown source. The symbol rate and modulation types are required for further operations, such as equalization, demodulation and finally identifying the information signal. Here, a method which is the cyclic correlation has been found quite effective on accurately estimating the symbol rates. So, the thesis initially is concentrated on the symbol rate estimation using cyclic correlation, the performances of known techniques in the literatures have been made. A new method proposed in this thesis has been developed to improve the symbol rate estimation. The proposed technique uses the same features of the second order statistics and cyclic stationary feature of the incoming linear modulated signal as it is in the cyclic correlation based methods however a better accuracy is provided by its frequency domain analyses after the required low complexity time domain operations. The system performances have been analyzed for the AWGN and frequency selective fading channels. The second important task is the modulation type recognition. The digital modulation classifiers using the features of the cumulants obtained in various conditions are studied. A novel feature is discovered in order to classify the MQAM modulation types. For the digital modulation classification a decision tree with two branches are developed by using features and the performances of the developed method are compared with best methods found in the literature.