A Family of Constrained Adaptive filtering Algorithms Based on Logarithmic Cost

TL;DR

This paper proposes a new constrained adaptive filtering algorithm called CLMLS based on a logarithmic cost function, which outperforms traditional methods and is extended to sparse scenarios with improved results.

Contribution

Introduction of the CLMLS algorithm using a logarithmic cost function and its extension to sparse cases with $ ext{l}_1$-norm penalty, demonstrating superior performance.

Findings

CLMLS outperforms conventional CLMS in stability and accuracy.

Analytical expressions for MSD are derived and validated.

Sparse CLMLS shows significant improvement over existing sparse adaptive filters.

Abstract

This paper introduces a novel constraint adaptive filtering algorithm based on a relative logarithmic cost function which is termed as Constrained Least Mean Logarithmic Square (CLMLS). The proposed CLMLS algorithm elegantly adjusts the cost function based on the amount of error thereby achieves better performance compared to the conventional Constrained LMS (CLMS) algorithm. With no assumption on input, the mean square stability analysis of the proposed CLMLS algorithm is presented using the energy conservation approach. The analytical expressions for the transient and steady state MSD are derived and these analytical results are validated through extensive simulations. The proposed CLMLS algorithm is extended to the sparse case by incorporating the $\ell_1$-norm penalty into the CLMLS cost function. detailed Simulations confirms the superiority of the sparse CLMLS over the state-of-the-art.