Degradation of the tracking performance of adaptive filteringalgorithms with data correlation

摘要

Degradation of the tracking performance of adaptive filteringnalgorithms with data correlation is analyzed. The analysis is done innthe context of the identification of a randomly time varying plant. Fivenalgorithms are considered. They are least mean square, recursive leastnsquares, sign, signed regressor, and sign-sign algorithms. The analysisnis done in terms of the steady-state excess mean-square error and thensteady-state mean-square weight misalignment. The degradation measurenadopted is the ratio of the performance index for correlated data tonthat for white data