Transient performance degradation of the LMS, RLS, sign, signedregressor, and sign-sign algorithms with data correlation

摘要

The transient performance degradation, for correlated input data,nis studied for various adaptive algorithms. The algorithms are leastnmean square (LMS), recursive least squares (RLS), sign (SA), signednregressor (SRA), and sign-sign (SSA). Analysis is performed for adaptivenplant identification with stationary Gaussian inputs. A closed-formnexpression for the mean convergence time is derived for each algorithm.nThe degradation measure used is the ratio of convergence time forncorrelated data to the convergence time for white data. The LMS and SRAndegradations are the same. The SA and SSA degradations are also thensame. The smallest degradation occurs for RLS, the largest for LMS andnSRA. The SRA, RLS, and LMS degradations are independent of plant-noisenvariance. The SA and SSA degradations increase with increased noisenvariance. The LMS and SRA degradations do not depend upon the weightninitialization, RLS (SA and SSA) depends weakly (significantly) upon thenweight initialization. The analytical results are supported bynsimulations