Signal identification algorithms for MIMO systems

Signal identification methods developed for recognizing and identifying unknown communication signals in a blind and non-cooperative manner are currently used in military and civilian applications frequently. The main task of signal identification systems is to blindly and non-cooperatively determine the transmission parameters specific to an unknown or partially known communication signal, such as the modulation type, bandwidth, carrier frequency, employed multiple access and the frequency spreading methods etc. The widespread use of digital wireless communication systems and the user demands for increasing data rates and service quality have lead to a surge in the development of new wireless transmission techniques, and paved the way for an increasing diversity in the transmission methods used in wireless communications. The existing signal identification systems need to be constantly extended and updated to include these newly emerging communication techniques, in order to be able to handle this diversity. Multiple-Input-Multiple Output (MIMO) systems, which offer an increased data rate and robustness by using multiple antennas at the transmitter and the receiver is one of the most promising of the above mentioned new communication methodologies, which have emerged in the last decade. The MIMO systems present new parameters to the signal identification systems, which need to be identified, such as the number of transmit antennas and the employed space time block code, which do not exist in conventional single antenna systems. Furthermore, existing modulation type identification algorithms desgined for single antenna systems cannot be aplied to the MIMO systems, due to the presence of self interference at each receive antenna. The aim of this thesis is to develop novel signal identification techniques for MIMO signals. In this context, two novel signal identification algorithms have been proposed. The first one is a joint antenna number and space time block code classification algorithm, which exploits the joint wide-sense cyclostationary characteristics of the space-time block coded transmit signals for discriminating between different codes. The second one is a joint modulation type and antenna number classification algorithm based on the minimum description length (MDL) criterion. In contrast to the existing literature, in which the unknown signal parameters are extracted separately, both of the proposed algorithms handle the classification problem in a joint manner. The proposed joint space-time block code and antenna number classification algorithm is capable of discriminating amongst a much larger number of space-time block codes than the methods existing in the literature, whereas the proposed MDL based joint modulation type and antenna number classification algorithm offers a high classification performance, while requiring very little a-priori information.