M. Ghavami - Electerical Engineering, University of Science and Defense Technology

 A step-by-step algorithm for enhancement of periodic signals that are highly corrupted by additive uncorrelated white gausian noise is proposed. In each adaptation step a new parallel second-order section is added to the previous filters. Every section has only one adjustable parameter, i.e., the center frequency of the self-tuning filter. The bandwidth and the convergence factor of each section is adjusted nonadaptively by a deterministic simple method which results in a stable and accurate regulation of the adaptive parameters. The step-by-step detection of sinusoidal signals prevents the convergence difficulties encountered in adaptive parallel IIR filters. Computer simulation results are presented to show the noise canceling performance of the proposed algorithm. Some comparisons with a new adaptive lattice notch filter for detection of multiple sinusoids are also provided.

adaptive filters, IIR Filters, Line Enhancement, Parallel Structures