一個新的數位資料接收機同步演算法

Abstract

在時序回復演算法中有一種為將取樣造成之頻譜空乏效應降至最低之方法, 然而現有的演算法都無法得到一無偏差之結果, 故本論文提出一新的實現方法來解決此問題。利用向量的表示法, 我們分析了傳統的時序調整機制。然後由幾何的關係我們得到一新的演算法。我們證明了本篇所提出之演算法能得到一零偏差的結果, 而傳統的演算法則有一不為零之偏差。在模擬中顯示本篇的方法不僅有較小的時序均方誤差, 同時也有較快的收斂速度。

One type of timing recovery algorithm is to minimize the "spectral null" effect caused by sampling. However, most existing algorithm are either difficult to implement or unable to obtain bias-free results. This thesis proposes a new approach to solve the problems. Using a vector representation, we analyze the timing adjustment mechanism in the conventional algorithm. Then, from the geometric relations, we derive a new class of timing recovery algorithms. For the trackable analysis, we only focus on the algorithm minimizing the conventional cost function. We prove that the proposed algorithm can produce a unbiased result while the conventional on cannot. Simulations show that not only the mean square timing error of the proposed algorithm is much lower, but also the convergence rate is faster.