Implements the order recursive adaptive laguerre filter from Fejzo and Lev-Ari (1997). Advantages:
*Fewer parameter to model than FIR filters modeling long impulse responses
*guaranteed stability, unique global minimum, small number of parameter for prescribed modelling error
*numerically robust & and low computational complexity ( O(M) ).
*Fast convergence (RLS equivalent) with no apriori learning rate required
*Order recursive, so filter coefficients are uncorrelated with each other, allowing for modular implementation.
Setting the pole parameter to 0 yields the standard classical gradient adaptive FIR algorithm.
Includes simple example on cross-talk resistant echo cancellation to show how to use it. |
