Fundamentals of matrix computations
Fundamentals of matrix computations
Hi-index | 0.00 |
It is shown that for both the LMS and the RLS algorithms the normalized bias introduced due to perturbation may not remain bounded, implying possible divergence. A new algorithm is developed for which the normalized bias due to such perturbation in the data remains bounded. Robustness analysis and mean squared error analysis are also presented for this algorithm.