一種改進(jìn)的變步長(zhǎng)ELMS算法
A Refrained Variable Step-Size ELMS-Like Algorithm
-
摘要: 在簡(jiǎn)單討論基本最小均方(LMS)算法的基礎(chǔ)上，引入了擴(kuò)展的最小均方(ELMS)算法，并分析說(shuō)明了該算法能達(dá)到更小的穩(wěn)態(tài)MSE。改進(jìn)的變步長(zhǎng)ELMS算法是在對(duì)有用信號(hào)的預(yù)測(cè)中采用了自適應(yīng)為歸一化的的最小均方(NLMS)預(yù)測(cè)估計(jì)器，步長(zhǎng)的迭代中引入遺忘因子i，利用其與誤差信號(hào)的加權(quán)和來(lái)產(chǎn)生新的步長(zhǎng)參與迭代。理論分析與計(jì)算機(jī)仿真結(jié)果表明，該算法有較好的收斂性能和較小的穩(wěn)態(tài)失調(diào)。
-
關(guān)鍵詞:
- LMS算法;變步長(zhǎng);ELMS算法;信號(hào)估計(jì)器;遺忘因子
Abstract: Following a brief discussion on basic LMS algorithms, the ELMS algorithms with steady small MSE is introduced. The new algorithm using the adaptive NLMS signal-estimator to predict the signal s(k). Forgetting factor i and error signal are used to control the step size update for iteration. Theoretics analysi and computer simulations demonstrate that the presented algorithm has good performance both in convergence properties and steady small misadjustment. -
吳光弼，祝琳瑜. 一種變步長(zhǎng)LMS自適應(yīng)濾波算法[J]. 電子學(xué)報(bào)，1994，22(1): 55-60.[2]葉華, 吳伯修. 變步長(zhǎng)自適應(yīng)濾波算法的研究[J].電子學(xué)報(bào), 1990, 18(4): 63-69.[3]Apolnario J A Jr , Werner S, Diniz P S R. Constrained normaliz- ed adaptive filters for CDMA mobile communications. http :/ / www. lps. ufrj . br/ profs/ diniz/ conf/ confi61. pdf.[4]Kwong R H, Johnston E W. A variable step size LMS algorithm. IEEE Trans.on Signal Processing, 1992, 40(7) : 1636-1642.[5]Widrow B. Adaptive Signal Processing. New York: Prentice-Hall Inc, 1985.[6]YuXiao.[J].Wang Qicai. An extended LMS algorithm in ANC. ICNNSP95, Nanjing, China.1995,:-[7]Dirk T M Slock. On the convergence behavior of LMS and the NLMS algorithms. IEEE. Trans.on TASSP, 1993, 141(9): 2811-2825. -
計(jì)量
- 文章訪問(wèn)數(shù): 2193
- HTML全文瀏覽量: 102
- PDF下載量: 798
- 被引次數(shù): 0
下載:
