Student Member and Masayuki KAWAMATA Member SUMMARY In
Abstract:
thispi er, we consider the steady state mean square error (MSE) analysis for 2- LMSadap3R e filtering algorithm in which the filter's weights areup dated along both vertical and horizontal directions as a doubly-indexed dynamical system. The MSE analysis is conducted using the well-known indep endenceassump3WR6 First we show that comp06C0D# of the weight-error covariance matrix for doubly-indexed 2- LMS algorithm requires anap60 ximation for the weight-error correlation coe#cients at largespgeD6 lags. Then wep06 ose a method to solve thispisD393 Further discussion is carried out for the sp ecial case when theinp6 signal is white Gaussian. It is shown that the convergence in the MSE sense occurs forstep size range that is significantly smaller than the one necessary for the convergence of the mean. Simulationexp eriments arepeD01 ted to sup ort the obtained analytical results. key words: 2-D LMS, steady state analysis, doubly-indexed system
Citations
| 136 | Adaptive Signal Processing – Widrow, Stearns - 1985 |
| 49 | Stationary and nonstationary learning characteristics of the LMS adaptive filter – Widrow, McCool, et al. - 1976 |
| 19 | Convergence analysis of LMS filters with uncorrelated Gaussian data – Feuer, Weinstein - 1985 |
| 3 | Stability analysis of 2-D systems – Fornasini, Marchesini - 1980 |
| 3 | e Filter Theory," Third edition – Haykin - 1996 |
| 1 | The two-dimensional adapme e LMS (TDLMS) algorithm – Hadhoud, Thomas - 1988 |
| 1 | Two-dimensional LMSadap tive FIR filters – Ohki, Hashiguchi - 1991 |
| 1 | Performance advantage ofcomp0C LMS for controlling narrow-band adapdD e array – Horowitz, Senne - 1981 |
| 1 | Performance analysis of the LMS algorithm with atap ed delay line (two dimensional case – Florian, Feuer - 1986 |
| 1 | Exactexp ectation analysis of the LMSadap=j e filter – Douglas, Pan - 1995 |
| 1 | oulis, "Probability, Random Variables, and Stochastic Processes – Pap - 1984 |
