Adaptive system modeling with critically decimated multirate filters

Multirate adaptive filtering in subbands at critically reduced sample rates is motivated by two objectives. Such filters facilitate a highly parallel adaptive filter structure operating at a reduced data rate. It is possible to achieve a tighter bound on the eigenvalue disparity of the input correlation matrix of the adaptive filter in each branch of the parallel structure. A tighter bound on the eigenvalue disparity is sought because it increases the convergence rate. The adaptive model that is developed by multirate adaptive filtering in subbands is examined and compared and to the model that is developed by traditional adaptive filtering