利用多階線性估測作分數取樣盲式等化

Abstract

ZF 和 MMSE等化器能以多階估測錯誤濾波器形式的演算法來實現。在本論文□，我們將介紹兩種利用多階估測的方法(FF-PEF和FB-PEF)而做成的盲式ZF等化器。我們調整Giannakis所提出的演算法，並將FF-PEF或FB-PEF代入，此調整過後的演算法將提供FF-PEF和FB-PEF額外的延遲。我們將建立一個模型來探討此額外的延遲對等化器的效應。我們發現此額外的延遲將會壓抑等化器輸出端的高斯雜訊。我們也建立我們的演算法，使它完全以線性估測的方法來估算等化器。再者，使用MMSE和ZF等化器間的線性關係，藉由FF-PEF或FB-PEF我們能實現MMSE等化器。從電腦模擬中，可看出額外的延遲將改善FF-PEF和FB-PEF的表現，另外也看出batch和RLS形式的演算法有比CMA好的表現。

Zero forcing (ZF) and mean minimum square error (MMSE) equalizers can be implemented with the multi-step PEF-based algorithms. In this thesis, we will introduce two methods (FF-PEF and FB-PEF) for blind ZF equalization based on multi-step linear prediction. We modify the algorithm proposed by Gianniakis and apply FF-PEF or FB-PEF to the modified algorithm, which provides the extra delays to FF-PEF and FB-PEF. We build a model to investigate the effects of the extra delays in the equalizers. We discover that the extra delays will suppress the white noise at the output of the equalizers. We also developed our algorithms for estimating equalizer completely based on linear prediction. Moreover, we use the linear relation between MMSE and ZF equalizers to accomplish MMSE equalizer applied by FF-PEF or FB-PEF. From the computer simulation, we can see that the extra delays make an improvement for FF-PEF and FB-PEF. And the batch algorithms and RLS based algorithms have better performance than the high-order CMA method.