Method and system for performing real-time spectral analysis of non-stationary signal

• US 10,371,732 B2
• Filed: 10/26/2012
• Issued: 08/06/2019
• Est. Priority Date: 10/26/2012
• Status: Active Grant

• Current Assignee: Avago Technologies General IP PTE Limited (Broadcom, Inc.), Keysight Technologies, Inc.
• Original Assignee: Keysight Technologies, Inc.
• Inventors: Al-Adnani, Adnan
• Primary Examiner(s): Islam, Mohammad K
• Application Number:
  US13/661,834
• Publication Number:
  US 20140122025A1
• Time in Patent Office: 2,475 Days
• CPC Class Codes: G01R 23/167 with digital filters
• CPC Class Codes: G01R 23/175 by delay means, e.g. tapped delay lines

Abstract

A method is provided for performing real-time spectral analysis of a non-stationary signal. The method includes sampling the non-stationary signal, using an observation window having a length short enough to approximate a stationary signal, to provide an initial set of sampled data, buffering the initial set of sampled data to obtain multiple buffered sets of sampled data, filtering the initial set of sampled data and the buffered sets of sampled data, using corresponding filter responses, to obtain multiple filtered sets of sampled data, and performing a chirp-z-transform (CZT) of the filtered sets of sampled data to provide a set of discrete Fourier transforms (DFT) coefficients. A total signal spectrum of the non-stationary signal is reconstructed using the set of DFT coefficients.

Claims

1. A method of performing real-time spectral analysis of a non-stationary signal by a spectrum analyzer, the method comprising:

sampling the non-stationary signal using an analog to digital converter, with a short observation window having a length short enough to approximate a stationary signal, to provide an initial set of sampled data;

buffering, in one or more buffers, the initial set of sampled data to obtain a plurality of buffered sets of sampled data, the plurality of buffered sets of sampled data emulating data sampled using a longer observation window based on the non-stationary signal sampled through the short observation window, improving overall spectral resolution of the non-stationary signal;

filtering the initial set of sampled data and the plurality of buffered sets of sampled data, using a corresponding plurality of filter responses, to obtain a plurality of filtered sets of sampled data;

performing a chirp-z-transform (CZT) of the plurality of filtered sets of sampled data to provide a set of discrete Fourier transforms (DFT) coefficients; and

reconstructing a total signal spectrum of the non-stationary signal using the set of DFT coefficients; and

displaying the reconstructed total signal spectrum of the non-stationary input signal on a spectrum analyzer display in real-time.

2. The method of claim 1, wherein the plurality of filter responses are derived from a single filter.

3. The method of claim 2, wherein buffering the initial set of sampled data comprises delaying the initial set of sampled data using a plurality of buffers, each buffer corresponding to one of the plurality of buffered sets of sampled data.

4. The method of claim 1, wherein each of the initial set of sampled data and the plurality of buffered sets of sampled data corresponds to a limited spectrum portion of the total signal spectrum of the non-stationary signal.

5. The method of claim 1, wherein spectral resolution is improved due to a relatively larger input data set, comprising the initial set of sampled data and the plurality of buffered sets of sampled data, mitigating frequency resolution error.

6. The method of claim 1, wherein performing the CZT of the plurality of filtered sets of sampled data comprises:

adding the plurality of filtered sets of sampled data;

performing a fast Fourier transform (FFT) on the added filtered sets of sampled data;

filtering the FFT sets of sampled data using an FFT of a complex chirp function;

performing an inverse FFT (IFFT) on the filtered FFT sets of sampled data; and

filtering the IFFT sets of sampled data using another complex chirp function to provide the set of DFT coefficients.

7. The method of claim 6, wherein filtering the initial set of sampled data comprises:

multiplying a filter coefficient for each sampled data point in the initial set of sampled data by a quotient of a complex number, based on a sampling frequency of the initial set of sampled data, and the number of initial and buffered sets of sampled data.

8. The method of claim 7, wherein filtering each of the plurality of buffered sets of sampled data comprises:

multiplying a filter coefficient for each sampled data point in each of the buffered sets of sampled data by the quotient of the complex number and the number of initial and buffered sets of sampled data.

9. The method of claim 1, wherein reconstructing the total signal spectrum of the non-stationary signal comprises evaluating the set of DFT coefficients in a z-plane, the set of DFT coefficients lying on circular or spiral contours beginning at an arbitrary point in the z-plane.

10. The method of claim 1, wherein a duration of the observation window is less than a period of intermittence or transience of the non-stationary signal.

11. A spectrum analyzer comprising:

an analog to digital converter (ADC) for receiving and sampling a non-stationary signal from a device under test (DUT), using a short observation window to approximate a stationary signal, to provide sampled signals;

a processing device programmed to;

filter the sampled signals using a first polyphase filter to provide first filtered sampled signals;

delay the sampled signals to provide delayed sampled signals;

filter the delayed sampled signals using a second polyphase filter to provide second filtered sampled signals;

perform chirp-z-transforms (CZTs) of the first and second filtered sampled signals to provide discrete Fourier transforms (DFT) coefficients; and

reconstruct a total signal spectrum of the non-stationary signal using the DFT coefficients; and

a display configured to display the reconstructed total signal spectrum of the input signal in real-time.

12. The spectrum analyzer of claim 11, wherein the processing device is further programmed to add the first and second filtered sampled signals before performing the CZTs.

13. The spectrum analyzer of claim 11, wherein performing the CZTs comprises:

performing a fast Fourier transform(FFT) on the added first and second sampled signals;

filtering the FFT sampled signals using an FFT of a complex chirp function;

performing an inverse FFT (IFFT) on the filtered FFT sampled signals; and

filtering the IFFT sampled signals using another complex chirp function to provide the set of DFT coefficients.

14. The spectrum analyzer of claim 11, wherein the first and second polyphase filters have corresponding first and second filter responses which are derived from a single filter.

15. The spectrum analyzer of claim 14, wherein reconstructing the total signal spectrum of the non-stationary signal comprises evaluating the DFT coefficients in a z-plane, the DFT coefficients lying on circular or spiral contours beginning at an arbitrary point in the z-plane.

16. The spectrum analyzer of claim 11, wherein a duration of the short observation window is less than a period of intermittence or transience, such that signal statistics of the non-stationary signal remain stationary throughout the observation window.

17. The spectrum analyzer of claim 11, wherein displaying the reconstructed total signal spectrum of the non-stationary signal enables zooming to an arbitrary narrow bandwidth of the sampled non-stationary signal, and further enables display at enhanced resolution across an entire bandwidth of the sampled non-stationary signal.

18. A spectral analyzer comprising:

an analog to digital converter (ADC) configured to receive and sample a non-stationary input signal from a device under test (DUT), and to provide first sampled data, wherein the sampling is performed using a short observation window having a length short enough to approximate a stationary signal;

a filter bank comprising a first polyphase filter configured to filter the first sampled data to provide first filtered sampled data, a second polyphase filter configured to filter second sampled data to provide second filtered sampled data, and a third polyphase filter configured to filter third sampled data to provide third filtered sampled data;

a first buffer configured to delay the first sampled data by a first delay to provide the second sampled data;

a second buffer configured to delay the second filtered sampled data by a second delay to provide the third sampled data;

an adder configured to add the first, second and third filtered sampled data to provide summed filtered data;

a chirp-z-transform (CZT) circuit configured to perform a CZT on the summed filtered data to provide frequency domain samples corresponding to the input signal, the frequency domain samples enabling reconstruction of a total signal spectrum of the input signal; and

a display configured to display the reconstructed total signal spectrum of the input signal in real-time.

19. The spectrum analyzer of claim 18, wherein the CZT circuit comprises:

a fast Fourier transform (FFT) circuit configured to perform an FFT operation on the summed filtered data to provide an FFT signal;

an FFT filter configured to filter the FFT signal to provide a filtered FFT signal;

an inverse FFT (IFFT) circuit configured to perform an IFFT operation on the filtered FFT signal to provide an IFFT signal; and

a IFFT filter configured to filter the IFFT signal to provide the frequency domain samples.

20. The spectrum analyzer of claim 18, wherein the first, second and third filtered sampled data emulate data sampled using a longer observation window based on the non-stationary signal sampled through the short observation window, thereby improving overall spectral resolution of the non-stationary signal.