A novel Variable Step Size LMS algorithm employing cross correlation between channel output and error signal

Abstract

ÖZETÇEBu bildiride, en küçük ortalama kareler (Least Mean Squares, LMS) algoritmasının yavaş yakınsama dezavantajına bir çözüm olarak kanal çıkışı ve hata işareti arasındaki çapraz ilintiyi kullanan yeni bir değişken adım büyüklüklü LMS (Variable Step Size LMS, VSS-LMS) algoritması önerilmektedir. Yeni algoritma sabit adım büyüklüklü klasik LMS algoritmasının yakınsama hızı ve doğruluğu arasındaki uyumsuzluğa çözüm sunmaktadır. Önerilen yöntemin başarımını doğrulamak için frekans seçici Rayleigh sönümlemeli kanallarda bilgisayar benzetimleri yapılmaktadır. HIPERLAN/1 standardını kullanarak elde edilen benzetim sonuçları önerilen VSS-LMS algoritmasının klasik LMS, RLS, normalize edilmiş LMS (Normalized LMS, N-LMS) ve diğer VSS-LMS algoritmalarından daha iyi başarıma sahip olduğunu göstermektedir. ABSTRACT A novel variable step size least mean squares (VSS-LMS) algorithm employing cross correlation between channel output and error signal has been proposed as a solution to disadvantage of slow convergence of LMS algorithm. The new algorithm resolves the conflict between the convergence rate and precise of the fixed step-size conventional LMS algorithm. Computer simulations have been performed to verify the performance of the proposed method in frequency selective Rayleigh fading channels. The obtained simulation results using HIPERLAN/1 standard have demonstrated that the proposed VSS-LMS algorithm has considerably better performance than conventional LMS, RLS, normalized LMS (N-LMS) and the other VSS-LMS algorithms.

A Distributed Speech Algorithm for Large Scale Data Communication Systems

“…LMS is the most popular adaptive algorithm due to its simplicity and robustness [22,23]. Adaptive filters can automatically estimate the statistical characteristic at such situation that the input signal and noise statistical properties were unknown [24][25][26], and on the basis of the automatic adjust filter parameters, the best filtering effect can be achieved.…”

Investigation of 20Gb/s WDM-PON transmission over 40-km employing 16QAM signal with blind equalization

“…Therefore, the equalization to suppress ISI should be performed to achieve reliable performance [4]. Traditional equalization techniques cause a waste of bandwidth and increase the complexity of both the transmitter and receiver due to using a training sequence [5], while blind channel equalization can suppress the ISI caused by the non-ideal channel for multilevel modulation signal without using any training sequence [6]. So we use blind equalization at the ONU receiver to equalize the damaged 16QAM signal for the purpose of improving the system performance and saving bandwidth.…”