Performance analysis of sign-sign algorithm for transversal adaptive filters

Abstract

The sign-sign algorithm (SSA), which is obtained by clipping both the reference input and the estimation error of the least mean square (LMS) algorithm, is analysed for transversal adaptive filters with correlated Gaussian data. The analysis focuses on the expected behaviours of the filter coefficients and the mean square error of the filter. The previous analysis of this type for the SSA is based on the assumption that the input data of adaptive filters are independent, identically distributed Gaussian, but this restriction is removed in our analysis. The analytical results are verified numerically through computer simulations for two examples - an adaptive linear predictor and an adaptive system identification. (C) 1997 Elsevier Science B.V.