Fast Converging Blind Equalization for Wireless Communications

本研究的目的是探索快速收敛的“盲”信道识别、均衡和序列估计方案的理论和实践问题，以对抗多径衰落和同信道干扰。它旨在消除或显著减少时变无线通信信道的训练信号。研究利用了通信信号和通信信道的二阶（循环平稳）统计和代数性质；与现有的使用高阶统计量的方法相比，该方法的收敛速度更快。通过将均衡技术和编码技术相结合，确定了频率选择性衰落信道，抑制了同信道干扰。利用信号子空间概念，研究了一种基于Viterbi算法的最优盲序列估计方法；它的结构对时变信道特别有吸引力。该研究包括理论研究、算法开发和实验验证三个部分。这项工作的意义在于它解决了基于TDMA和GDMA系统的两个重要限制：多径干扰和多用户干扰。它有可能通过消除训练序列来提高传输能力。本研究开发的技术具有广泛的应用，包括周期系统的识别和医学图像处理。＊＊＊

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. ***