用於高階QAM之盲蔽等化技術

Abstract

在現代的無線電訊系統中,人們常利用適應等化器來解決多重路徑的干擾 問題。傳統上的作法是在接收端利用一連串週期性的已知序列,來求得等 化器的參數。此類處理過程常是昂貴的, 不切實際的,而且在多點網路系 統中是無法做到。為了避免傳統等化技術的缺失,而發展出一套只利用通 道的輸出及傳送訊號的統計特性來估測等化器的參數。這類等化器我們稱 為盲蔽等化器。在論文中,我們所建議的盲蔽等化器,其操作是對從QPSK訊 號的幾何和統計性質所推出的非凸成本函數做最小化。直接對此成本函數 做一小幅調整 ,我們可將此方法擴展到QAM訊號上。模擬的結果在適當的 傳輸情況下,我們所建議的方法和Godard所提的方法有一個優點是Godard 方法所沒有的,那就是免除了傳輸延遲所導致的隨機相位差。我們也嚐試 結合適應陣列訊號處理來進一步改善等化器的效率。最後,我們亦分析了 此等化器的收斂性質,並且由電腦模擬的結果得到證明 。 In terrestial digital radio systems,the use of adaptive equalizers to combat with multipath distortion is desired. Conventional equalization techniques require that a training sequence known to the receiver is sent periodically such that the equalizer can determine the inverse channel transfer function via a deconvolution procedure. This kind of processing can be costly and may not be achievable in many applications such as multi-point networks. The desirability of avoiding the difficulty of the training sequence driven process prompts the development of a class of blind equalizers which achieve parameter estimation based on the channel output and prior knowledge about the input sequence. In this thesis,we propose a blind equalizer whose operation is based on the minimization of a noncovex cost function derived from the geometric and statistical properties of the QPSK constellation. Extension of the proposed method to QAM signals is straightforward with a minor modification of the cost function. Simulation results demonstrate that the proposed method achieves the performance of Godard's method for modreate channel conditions, and outperforms the latter for high order QAM signals. A merit of the proposed method not shared by Godard's is that it is immune to the random phase error incurred with the channel distortion. As a preliminary trial, We also incorporate spatial domain processing via an adaptive array to further improve the performance of the equalizer. The convergence property of the proposed algorithm is analyzed mathematically and confirmed by computer simulations.