| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 852.78 KB | Adobe PDF |
Advisor(s)
Abstract(s)
Compressed sensing (CS) is a theory that allows us
to recover sparse or compressible signals from a much smaller
number of samples or measurements than with traditional
methods. The problem of detection and estimation of the
frequency of a signal is more difficult when the frequencies of
the signal are not present on the DFT basis. The Fourier
coefficients are not exactly sparse due to the leakage effect if the
frequency is not a multiple of the fundamental frequency. In
this work we present a high frequency resolution spectrum
estimation algorithm that explores the CS, for this type of
nonperiodic signal from finite number of samples. It takes
advantage of the sparsity of the signal in the frequency domain.
The algorithm transforms the DFT basis into a frame with a
large number of vectors by inserting columns between some of
the existing ones. The proposed algorithm can estimate the
amplitudes and frequencies even when the frequencies are too
close together, a particularly difficult situation which are not
covered by most of the known algorithms. Simulation results
show good convergence and a high resolution when compared
with other algorithms
Description
Keywords
Compressed sensing redundant frames signal sparse representations spectral estimation
Citation
Publisher
Prof. Chandratilak De Silva Liyanage