離散多頻帶傳收器之最佳時域等化

Abstract

在離散多頻帶（discrete multitone，DMT ）接收器中，使用了一個時域等化器（time-domain equalizer，TEQ）來縮短通道響應。然而，現有的TEQ設計方法通常會把訊雜比（signal to noise ratio ，SNR）降低而導致傳輸率下降。近年來，有些學者已注意到這個問題也提出了減輕這個問題的方法。然而這些方法有的是用嘗試觀察得來，有的是假設不夠實際。通道縮減使SNR下降的原因和解決這個問題最好的方法都還是未知的。在本論文中，我們將要研究這個問題。我們首先會推導出TEQ係數和SNR之間的數學關係式，然後我們會提出能夠同時縮短通道響應和達到最大傳輸率的新演算法。經模擬之後可以看出，我們的演算法確實優於現存的所有TEQ設計方法。

In a discrete multitone（DMT） receiver, a time-domain equalizer（TEQ） is usually used to shorten the channel response. However, the existing TEQ design methods often decrease the signal to noise ratio (SNR) resulting in a lower transmission rate. In recent years, some researchers have noticed this problem and proposed methods to alleviate this problem. However, these methods are either heuristic or their assumptions are not realistic. The reason for the SNR reduction (due to channel shortening) and the optimal way to solve the problem still remain unknown. In this thesis, we intend to investigate these problems. We first derive the mathematical relationship between the TEQ coefficients and SNR, and show how the TEQ reduces the SNR. Then, we propose a novel algorithm than can simultaneously shorten the channel response and achieve the maximum transmission rate. Simulations show that our algorithm outperforms all existing TEQ design methods.