Analysis of Adaptive Learning Rate Strategies for Sign-Sign LMS: Stability and Speed

Abstract

This paper addresses coefficient adaptation techniques applicable to DSP-based equalization. The conventional least mean square (LMS) algorithm utilizes both the data and the error signal to update coefficients, leading to robust performance but often at the cost of increased hardware complexity. To address these limitations, this paper proposes an adaptive learning rate strategy based on autocorrelation analysis. The proposed method aims to automatically adjust the learning rate μ in response to the system state, thereby enhancing both convergence stability and speed in SS-LMS adaptation. Simulation results demonstrate superior convergence speed and reduced bit error rate (BER) performance compared to the conventional LMS adaptation scheme.