9321813 Tong The objective of this research is to explore theoretical and practical issues of fast converging "blind" channel identification, equalization, and sequence estimation schemes to combat multipath fading and co-channel interference. It aims to eliminate or significantly reduce training signals for the equalization of time- varying wireless communication channels. The research exploits the second -order (cyclostationary) statistical and algebraic properties of communication signals and the communication channel; it leads to a faster convergence rate than those of existing methods using higher-order statistics. By combining equalization and coding techniques, the research identifies a frequency selective fading channel and suppresses co-channel interference. Using the so-called signal subspace concepts, an optimal blind sequence estimation method via the Viterbi algorithm is investigated; its structure is particularly attractive to time- varying channels. The research consists of theoretical investigation, algorithm development, and experimental verification. The significance of the work is that it addresses the two important limitations of TDMA and GDMA based systems: the multipath interference and the multiuser interference. It has the potential to increase transmission capabilities by eliminating training sequences. The techniques developed in this research have a wide range of applications including identification of periodic systems and medical image processing. ***
