• 作者： 陳巽璋
• 作者服務機構： 國立中山大學電機系
• 中文摘要： 本篇論文以適應性線增益器為分析對象。利用理論分析，我們證明最小均方值濾波器的收斂特性，與輸入信號中正弦波間的能量分佈有著密切的關係。除此之外，我們亦發現如果濾波器的階數與演算法中之步距的乘積保持一定，則收斂速度將完全一致。
• 英文摘要： A comprehensive study of the convergence property of the time domain complex least-mean square(LMS) adaptation algorithm is presented. In the case of the adaptive line enhancer (ALE), a theoreticalanalysis is carried out under the assumption that the signals consist of multiple sinusoids with un-equalpower burried in broadband noise. Each sinusoidal signal is narrowband，statistically independent, ort-hogonal, and Rayleigh fading. Under such circumstances, a closed form expression for the transient meansquare error (MSE) can be obtained. Based on this expression of the transient MSE, we are able to studythe effect on the convergence rate of the complex LMS ALE. Here, we have showed that the convergencerate of the complex LMS adaptation algorithm，in the case of the ALE, depends not only on the disparityin eigenvalues but also depends highly on the power ratio of the sinusoids in the received signal. This isdifferent from the conventional study for the convergence rate of the LMS algorithms. Furthermore, wehave also observed that a similar convergence property can be obtained when the length of the adaptivefilter is increased and the step-size is reduced accordingly.
• 中文關鍵字： adaptive filtering; line enhancer; eigenvalue spread; power effect; convergence rate
• 英文關鍵字： --