Practical communication systems suffer from inter-symbol interference (ISI) due to multipath propagation so that channel equalization is important to restore the correct transmitted signals. Blind equalization is an effective equalization approach for reducing inter-symbol interference and improving system throughput. This dissertation introduces several new methods of blind equalization as well as joint blind equalization and multiuser detection in CDMA systems. A common feature of these new methods is that they are all based on second-order statistics.; Blind equalization can be performed using full channel information obtained by channel identification, using partial channel information, or using direct equalization without channel information. This dissertation addresses new algorithms in all these three categories. First, a new subspace tracking method is introduced for blind channel identification by rank-one ULV decomposition, whose advantages include lower computations and no requirement of rank estimation. Then linear prediction based methods are introduced. They estimate equalizers using partial channel information or using only the structure of the channel matrix. They are computationally efficient and robust to channel length over-estimation. Third, a direct equalization method is given which relies on channel output whitening. It is quadratic so that closed form solutions can be obtained. Finally, algorithms of joint blind equalization and multiuser detection in CDMA systems are introduced, which are based on linear prediction. Simulations are shown to demonstrate their performances.
展开▼