Pipelined signal adaptive architecture of a time-frequency Wiener filter for highly nonstationary FM signals estimation

Abstract

Pipelined signal adaptive architecture of an optimal (Wiener) timefrequency (TF) filter has been designed. It is developed based on the real-time results of TF analysis, on the correspondence of the filter’s region of support to the instantaneous frequency (IF) of the filtered signal, and on the TF analysis-based IF estimation. The implemented pipelining technique allows the proposed design to overlap in execution unconditional steps performing in neighboring TF instants. In this way, the considered filter significantly enhances critical design performance related to the time required for execution. The achieved improvement corresponds to the one clock cycle by a TF point, which means that the improvement by a single TF point can reach up to 50%. The design is tested on multicomponent highly nonstationary FM noisy signals.