Frequency Guided Generalized Adaptive Notch Filtering—Tracking Analysis and Optimization

Generalized adaptive notch filters (GANFs) are estimators of coefficients of quasi-periodically time-varying systems, encountered e.g., in RF applications when Doppler effect takes place. Current state of the art GANFs can deliver highly accurate estimates of system variations´ frequency, but underperform in terms of accuracy of system coefficient estimates. The paper proposes a novel multistage GANF with improved coefficient tracking properties and smaller sensitivity to choice of estimation gains. The processing pipeline consists of three stages. First, a pilot filter computes preliminary frequency estimates. Second, a special linear filter reshapes the pilot frequency estimates. Third, a frequency guided GANF works out final estimates of system coefficients. A nontrivial design of the second stage filter assures that the proposed solution has a considerably better performance than current stage of the art solutions or a simpler two-stage approach consisting of the pilot and the frequency guided filter only.