用於無線通訊之等化器設計

Abstract

針對高速的無線傳輸，本篇論文研究使用正規化最小均方根演算法(NLMS)的適應性等化器之效能評估。關於通道環境的建立，以多重路徑Rayleigh通道模型作分析，每一條路徑的時間延遲和平均功率是參考ITU的室內通道模型，並使用Jakes模型產生每一個路徑所需要的Rayleigh分佈。本文主要研究使用複數架構的等化器，包括線性橫向等化器（LTE）及決策迴授等化器（DFE），同時以相同環境下不使用等化器的例子作對照。本文同時評估演算法中的步階大小和等化器中暫存器數目多寡等許多重要參數的影響，並分析使用不同步階對演算法的收斂速度與穩態時的錯誤變異量之影響。基於上述完整的分析，我們可調整上述重要參數來設計滿足系統規格的等化器。最後，針對原本的正規化最小均方根演算法做一些修改以降低實踐硬體上計算複雜度。

The performance of adaptive equalizers using the normalized least mean square algorithm (NLMS) with QPSK modulation technique for high data rate wireless communications is investigated. Multi-ray tapped delay line Rayleigh channel model is used for propagation analysis. The tapped delay and average power of the each path refer to ITU tapped delay line indoor channel model. The Jakes’ model is used to generate the Rayleigh distribution of the each path. The effect of using linear transversal equalizer (LTE) and decision feedback equalizer (DFE) in complex structure for such channel model is evaluated and compared with that without equalizer. The influence of some important parameters, such as tap number of the equalizers, and step size setting of the adaptive algorithm, is examined. The convergence speed and final convergence variance of the NLMS with different step size for LTE and DFE is presented. Design specifications can then be adjusted according to system requirement. A modified NLMS is proposed and had been implemented to reduce hardware computational complexity.