Loading...

Spectral estimation by computationally reduced Kalman Filter

Kazemi, R ; Sharif University of Technology | 2012

646 Viewed
  1. Type of Document: Article
  2. DOI: 10.1007/s00034-012-9429-7
  3. Publisher: 2012
  4. Abstract:
  5. This paper introduces an innovative frequency estimation approach that relies on a new and innovative structure for Kalman Filter (KF) in signal processing. Kalman Filtering, because of its state estimating nature, excels other methods in realtime processing. However, the main drawback of Kalman Filtering is that it involves time-consuming matrix operations which makes it inferior to faster methods like fast Fourier transform (FFT). The most evident privilege of the new structure is that it "synthesizes" the components rather than "extracts" them as standard Kalman Filter does. This parallel synthesis of signal components reduces the computational order of the Kalman Filter dramatically. In addition, in the proposed parallel structure, each component can be controlled separately by adjusting its own parameters. Hence, the convergence time and signal-to-noise ratio (SNR) can be improved. This capability surpasses the proposed structure to both KF and FFT. Equations needed for these adjustments are derived analytically and by simulation
  6. Keywords:
  7. Computational order ; Convergence time ; Innovative structures ; Kalman-filtering ; Matrix operations ; Parallel structures ; Parallel synthesis ; Realtime processing ; Signal components ; Signaltonoise ratio (SNR) ; Spectral Estimation ; Standard Kalman filters ; Adaptive filtering ; Fast Fourier transforms ; Signal processing ; Spectrum analysis ; Kalman filters
  8. Source: Circuits, Systems, and Signal Processing ; Volume 31, Issue 6 , 2012 , Pages 2205-2220 ; 0278081X (ISSN)
  9. URL: http://link.springer.com/article/10.1007%2Fs00034-012-9429-7