Signal processing method for improving the signal-to-noise ratio of a noise-dominated channel and a matched-phase noise filter for implementing the same
First Claim
1. A matched-phase noise filter comprising a processor implementing the filter function:
-
wherein;
φ
(ω
) is a set of phases;
{tilde over (P)}(ω
) is the frequency spectrum of a received signal;
Ñ
(ω
) is the frequency spectrum of a noise component of the received signal; and
{tilde over (F)}(ω
) is the frequency spectrum of the received signal with the noise spectrum projected out.
1 Assignment
0 Petitions
Accused Products
Abstract
A matched-phase noise filter includes an analog to digital converter (ADC) for receiving an analog composite, noise-dominated signal containing a signal of interest and producing a digital composite signal, an input/output port receiving the digital composite signal and providing a matched-phase signal, and a processor receiving the digital composite signal via the I/O port and generating the matched-phase signal. According to one aspect of the invention, the signal to noise ratio between the signal of interest and a noise component within the digital composite signal is increased by approximately an order of magnitude, based on an actual spectrum of the digital composite signal and an estimated spectrum of the noise component, and independent of the particular form of the signal of interest. A method of matched-phase noise filtering for improving the SNR of a noise dominated signal independent of the form of the signal is also described.
-
Citations
14 Claims
-
1. A matched-phase noise filter comprising a processor implementing the filter function:
-
wherein; φ
(ω
) is a set of phases;
{tilde over (P)}(ω
) is the frequency spectrum of a received signal;
Ñ
(ω
) is the frequency spectrum of a noise component of the received signal; and
{tilde over (F)}(ω
) is the frequency spectrum of the received signal with the noise spectrum projected out.
-
-
2. A method for filtering a noise dominated signal, comprising the steps of:
-
producing an actual spectrum of the noise dominated signal responsive to said noise dominated signal;
selecting an estimated spectrum representing noise in the noise dominated signal;
generating a desired spectrum indicative of a signal of interest based on said actual spectrum and said estimated spectrum and independent of the particular form of said signal of interest. - View Dependent Claims (3, 4, 5, 6, 7)
generating said desired spectrum indicative of said signal of interest based on said actual spectrum and said estimated spectrum and independent of the particular form of said signal of interest using the formula;
wherein; i=(−
1)½
;
φ
(ω
) is a set of phases;
{tilde over (P)}(ω
) is the frequency spectrum of said actual spectrum;
Ñ
(ω
) is the frequency spectrum of said estimated spectrum; and
{tilde over (F)}(ω
) is the frequency spectrum of said desired spectrum with the estimated spectrum projected out.
-
-
5. The method for filtering a noise dominated signal as recited in claim 2, further comprising the step of selecting said estimated spectrum from a collection of estimated spectra indexed according to a measurable parameter.
-
6. The method for filtering a noise dominated signal as recited in claim 5, wherein said measurable parameter is wind speed.
-
7. The method for filtering a noise dominated signal as recited in claim 2, wherein said producing step comprises:
-
receiving the noise dominated signal in the time domain;
converting the noise dominated signal in the time domain to the frequency domain thereby producing the actual spectrum of the noise dominated signal responsive to the noise dominated signal.
-
-
8. A method for filtering a noise dominated signal, comprising the steps of:
-
receiving a time series signal consisting mostly of noise and containing a signal of interest;
selecting a close estimate of the frequency dependence of the noise spectrum within said time series signal;
executing a Fast Fourier transform (FFT) of the time series signal so as to obtain input spectral amplitudes and phases of each frequency component of said time series signal;
orthogonalizing said input spectral amplitudes to the noise spectral amplitude so as to produce orthogonalized spectral amplitudes;
assigning input phases to the resultant orthogonalized spectral amplitudes so as to produce assigned phases; and
inverse FFT transforming the assigned phases and said orthogonalized spectral amplitudes back into the time domain to thereby produce a matched-phase signal;
wherein a signal to noise ratio (SNR) of said signal of interest with respect to said noise in said matched-phase signal is at least an order of magnitude greater than the SNR of said time series signal. - View Dependent Claims (9)
-
-
10. A filter comprising:
-
an input portion effective to receive an input signal having a spectrum {tilde over (F)}(ω
), where ω
is frequency, said input signal having a component Ñ
(ω
); and
a filtering portion, said filtering portion being effective to form a filtered signal {tilde over (F)}(ω
) such that;
- View Dependent Claims (11, 12, 13)
-
-
14. A matched-phase noise filter comprising a processor implementing a filter function, said filter function being proportional to:
-
wherein; φ
(ω
) is a set of phases;
{tilde over (P)}(ω
) is the frequency spectrum of a received signal; and
Ñ
(ω
) is the frequency spectrum of a noise component of the received signal.
-
Specification