The present invention relates to a device for analyzing a signal. The invention also relates to a platform comprising such a device and to a method of analysis.
For certain applications, it is desirable to perform real-time or short-time spectral analyzes on broadband signals, typically having a spectrum greater than or equal to 10 GHz, in order to characterize their intrinsic properties, such as the number of signals, their frequency or duration.
For this, it is known to carry out the digitization of the signal by using a broadband sample-and-hold device operating at a given sampling frequency associated with an analog-to-digital conversion device. For broadband signals, the performance of known sample-and-hold devices does not allow the simultaneous meeting of the Shannon criterion and obtaining a calculation speed sufficient for real-time processing. As a result, aliasing effects occur and make it impossible to distinguish certain signals, in particular to determine their frequency.
Because of these aliasing effects, it has been proposed to sample the signal several times in parallel but at different sampling frequencies and then to apply Fourier transforms on identical time windows comprising the sampled signals. The comparison of the spectra obtained makes it possible to find the signal.
In fact, since the frequency resolutions of the spectral analyzes are all different, the elimination of ambiguity between the signals in the comparison step is complex and can be erroneous. The reliability of these problems assumes a relatively slow method.
There is therefore a need for a device that is faster to determine the characteristics of a broadband signal.
To this end, a signal analysis device or a device for analyzing a signal is proposed that comprises N signal processing channels, each processing channel comprising a sampling device operating at a sampling frequency, an analog-to-digital converter operating at a conversion frequency, the conversion frequency being strictly greater than the sampling frequency, a filter interposed between the sampling device and the analog-to-digital converter, the filter being a filter having a cutoff frequency, and a calculator adapted to analyze the output signal of the analog-to-digital converter.

According to particular embodiments, the analysis device may comprise one or more of the following characteristics, taken separately or in any technically feasible combination:
on each processing channel, the difference between the conversion frequency
and the sampling frequency is strictly greater than 10%.
the cutoff frequency of each filter is greater than or equal to half of the largest
sampling frequency of all the processing channels.
the conversion frequency is strictly greater than the largest sampling frequency.
the components are identical between the processing channels, the identical
components being chosen from the filter and the analog-to-digital converter.
the analysis device is suitable for a signal having a spectral band greater than or
equal to 10 GHz.
each filter is a band-pass filter or a low-pass filter.
the analysis device further comprises a receiver designed to receive the signal
and a separator connected to the receiver and comprising N output ports,
wherein the separator separates the signal the receiver is designed to receive at
the N output ports, N being an integer greater than or equal to 1, each
processing channel being connected to a respective output port. In addition, the present description also relates to a platform, in particular an aircraft, comprising an analysis device as previously described.
The present description also describes a method for analyzing a signal by an analysis device comprising N signal processing channels, wherein the method comprises, on each processing channel, at least one sampling step at a frequency sampling device, to obtain a sampled signal, for filtering the signal sampled by a filter having a cutoff frequency, to obtain a filtered signal for converting the filtered signal by an analog-to-digital converter operating at a conversion frequency, the conversion frequency being strictly greater than the sampling frequency, to obtain a converted signal, and analysis of the signal converted by a calculator.
Other features and advantages of the invention will become apparent upon reading the following description of embodiments of the invention, given by way of example only and with reference to the drawings, wherein:
Fig. 1 shows a schematic representation of a platform comprising a device for
analyzing a signal;

Fig. 2 shows a schematic view of the analysis device of Fig. 1;
Fig. 3 shows an example of spectra that the analysis device is designed to
obtain;
Fig. 4 shows examples of spectra obtained at the end of a first step of an
implementation of an example of an analysis method;
Fig. 5 shows examples of spectra obtained at the end of a second step of an
implementation of an example of an analysis method;
Fig. 6 shows examples of spectra obtained after a third step of an
implementation of an example of an analysis method;
Fig. 7 shows examples of spectra obtained at the end of a fourth step of an
implementation of an example of an analysis method, and
Fig. 8 shows examples of spectra obtained after a fifth step of an
implementation of an example of an analysis method. A platform 10 is shown schematically in Fig. 1. The platform 10 is, for example, a vehicle. According to the proposed example, the platform is an aircraft. The platform 10 comprises systems 12 operating in real time ensuring the operation of the platform 10.
Three real-time systems 14 are shown in Fig. 1.
For the rest, it is assumed that one of the real-time systems is a signal analysis device 16.
The analysis device 16 is adapted to analyze the properties of an incident signal. The analysis device 16 is designed for a signal having a spectral band greater than or equal to 10 GigaHertz (GHz).
A signal with a spectral band greater than or equal to 10 GHz is considered a broadband signal.
The analysis device 16 comprises a receiver 18, a separator 20 and processing channels 22.
The receiver 18 is designed to receive the signal to be analyzed. The separator 20 has an input port 20E and N output ports 20S. N is an integer greater than or equal to 2. For example, the number N may be greater than 4.

The separator 20 is connected to the receiver 18. More specifically, the input port 20E of the separator 20 is connected to the output 18S of the receiver 18.
The separator 20 is adapted to separate the signal that the receiver 18 is adapted to receive at the N output ports 20S.
Each processing channel 22 is connected to a respective output port 20S of the separator 20.
The number of processing channels 22 is therefore equal to N.
Each processing channel 22 is suitable for processing a signal.
Each processing channel 22 comprises several components which are: a sampling device 24, an analog-to-digital converter 26, a filter 28 and a calculator 30.
In the case proposed, the sampling device 24 is a sample-and-hold device 24.
In the example described, it is assumed that the filter 28 and the analog-to-digital converter 26 are identical between the processing channels.
By "identical components" in this context, it is understood that both components have nearly the same characteristics as the manufacturing variations.
For a filter 28 formed by electronic elements, the variations result, for example, on the one hand from the intrinsic properties of the various elements used. Similarly, for an analog-to-digital converter 26, the variations are related to manufacturing variations, variations on the parasitic elements, and variations on the wiring.
The sample-and-hold device 24 is adapted to operate at a sampling frequency specific to the processing channel 22 considered.
Thus, the sampling frequency of the sample-and-hold device 24 of the processing channel 22 having the index i is denoted Fsi with the index i which is an integer between 1 and N.
In the example described, the sample-and-hold device 24 is designed to process a signal having a frequency excursion of several tens of GigaHertz.
In particular, the sample-and-hold device 24 has a low signal acquisition time with respect to the maximum frequency of the signal to be analyzed.
The filter 28 is interposed between the sample-and-hold device 24 and the analog-to-digital converter 26.
The filter 28 is adapted to filter an incident signal from the sample-and-hold device 24.
The filter 28 is an analog filter.

The filter 28 has a cutoff frequency Fc.
The cut-off frequency Fc of each filter 28 is greater than or equal to half of the largest sampling frequency Fsi of the set of processing channels 22.
According to the example described, the filter 28 is a band-pass filter.
In a variant, the filter 28 may be a low-pass filter.
It should also be noted that each filter 28 has attenuation at a frequency between Fso/2 and Fso compatible with the desired dynamics after spectral analysis.
The frequency Fso is the common conversion frequency at which each analog-to-digital converter 26 is designed to operate at a conversion frequency.
The conversion frequency Fso is strictly greater than the sampling frequency Fsi.
Preferably, the difference between the conversion frequency Fso and the sampling frequency Fsi is strictly greater than 10%, which makes it possible to reduce the architecture constraints on the filter 28.
Such an effect is even more favorable if the difference between the conversion frequency Fso and the sampling frequency Fsi is strictly greater than 20%, or even strictly greater than 50%.
In fact, the smaller the difference between the conversion frequency Fso and the sampling frequency Fsi, the more it becomes possible to use on each processing channel 22, filters 28 having rejection slopes with low steepness. Each filter 28 may then have a lower order, which allows consideration of simpler filters 28. In particular, the filter 28 may have fewer components, or components with lower performance.
The calculator 30 is adapted to analyze the digitized signal of the analog-to-digital converter 26.
The operation of the analysis device 16 is now described with reference to an exemplary implementation of a method of analyzing a signal, as well as to Fig. 3 to 8.
During a reception step, the receiver receives the signal to be analyzed.
As an illustration, the implementation of the method is illustrated for a limited spectrum signal. The extension to a wider spectrum signal is immediate but more difficult to represent.
The signal to be analyzed is the signal represented in Fig. 3.
Then, the signal to be analyzed is separated during a separation step on the N processing channels 22 using the separator 20.

On each processing channel 22, a plurality of operations are performed and are now detailed.
In a first operation, the signal is sampled by sample-and-hold device 24 at the sampling frequency Fsi of the sample-and-hold device 24.
This operation is interpreted as a periodization operation of the signal spectrum with a period related to the sampling frequency Fsi.
The spectrum of the signal obtained after implementation of this operation is described in Fig. 4 for the case of the first processing channel 22 (sampling frequency Fsi) and the second processing channel 22 (sampling frequency FS2).
Then, during a second operation, the filter 28 filters the sampled signal.
Since the cutoff frequency Fc of each filter 28 is greater than half the sampling frequency Fsi, the filter 28 selects only that portion of the spectrum that has been aliased onto the baseband.
Next, each analog-to-digital converter 26 scans each signal at the conversion frequency Fso.
The operations described after this operation are implemented by the calculator 30.
A fast Fourier transform operation is then applied to the converted signal.
The signals obtained after spectral analysis are shown in Fig. 5 for the first processing channel 22 and the second processing channel.
The knowledge of the sampling frequency Fsi of each processing channel 22 makes it possible to keep only the useful parts of the spectra obtained in the previous step.
The useful part of the spectrum on the first channel is the spectrum between 0 and half of the first sampling frequency Fsi.
More generally, the useful part of the spectrum on the ith channel is the spectrum between 0 and half of the ith sampling frequency Fsi.
Then, each spectrum is completed with zero values to complete the spectrum up to the conversion frequency Fso.
The spectra thus obtained for the first treatment channel 22 and the second treatment channel 22 are visible in Fig. 6.
The signal obtained previously serves as a basic pattern for reconstructing the spectrum between Fsi/2 and Fsi by symmetrizing the basic pattern. A pattern to be repeated is thus obtained.
Then the pattern to be repeated is repeated to reconstruct the signal.

The reconstructed signals thus obtained for the first processing channel 22 and the second processing channel 22 are visible in Fig. 7.
A search operation of the frequency zones where the reconstituted signals are superimposed, is then implemented.
The operation is shown diagrammatically in Fig. 8 which comprises three graphs, the first being that of Fig. 3 and the two following those of Fig. 7.
In the illustrated case, the superposition zone is in the rectangle 32 in dashed lines. The signal in the rectangle 32 corresponds to the signal to be analyzed which is the desired signal.
The method which has just been described makes it possible to carry out an analysis relating to a plurality of signals.
In addition, the method is easily reconfigurable to the extent that it is sufficient to change the sampling frequency Fsi to achieve a different operation of the method.
Also, the analysis device 16 is a reconfigurable digital broadband signal spectral analysis device.
The method may be easily implemented since the operations used are not complicated to optimize. For example, the fast Fourier transform may be easily optimized from the computational point of view.
In fact, the method is implemented on an analysis device 16 making it possible to simplify the digital processing that is performed subsequently, in particular by decoupling the multiple Fsi sampling frequencies used to digitize the signal and the frequency used to acquire the signals which are then analyzed using fast Fourier transforms.
The method does not impose the number of points on which the operations are to be implemented on each processing channel 22. The number of processing channels 22 may therefore be optimized depending on the desired application for the method.
In addition, it should be noted that the method is implemented by maintaining a width of the identical spectral channels between the processing channels 22.
Other variants of the analysis device 16 are conceivable. In particular, the signal to be analyzed may be unique (it is then divided by N), or may come from different antennas, and this analysis device 16 may be coupled to other signal analysis systems.

CLAIMS

1.Device for analyzing a signal (16), the analysis device (16) having N signal
processing channels (22), N being an integer superior or equal to 2, each processing channel
(22) comprising:
- a sampling device (24) operating at a sampling frequency (Fsi),
- an analog-to-digital converter (26) operating at a conversion frequency (Fso), the conversion frequency (Fso) being strictly greater than the sampling frequency (Fsi),
- a filter (28) interposed between the sampling device (24) and the analog-to-digital
converter (26), the filter (28) being a filter having a cutoff frequency (Fc), and
- a calculator adapted to analyze the output signal of the analog-to-digital converter
(26).
2. Analysis device according to claim 1, wherein, on each processing channel (22), the difference between the conversion frequency (Fso) and the sampling frequency (Fsi) is strictly greater than 10%.
3. Analysis device according to claim 1 or 2, wherein the cutoff frequency (Fc) of each filter (28) is greater than or equal to half of the largest sampling frequency (Fsi) of the set of processing channels (22).
4. Analysis device according to any one of claims 1 to 3, wherein the conversion frequency (Fso) is strictly greater than the largest sampling frequency (Fsi).
5. Analysis device according to any one of claims 1 to 4, wherein components are identical between the processing channels (22), the identical components being selected from the filter (28) and the analog-to-digital converter (26).
6. Analysis device according to any one of claims 1 to 5, wherein the analysis device (16) is designed for a signal having a spectral band greater than or equal to 10 GHz.

7. Analysis device according to any one of claims 1 to 6, wherein each filter (28) is a band-pass filter or a low-pass filter.
8. Analysis device according to any one of claims 1 to 7, wherein the analysis device (16) further comprises:

- a receiver (18) adapted to receive the signal,
- a separator (20) connected to the receiver (18) and having N output ports (20S), wherein the separator (20) separates the signal that the receiver (18) is designed to receive at the N output ports (20S),
- each processing channel (22) being connected to a respective output port (20S).
9. Platform (10), in particular an aircraft, comprising an analysis device (16)
according to any one of claims 1 to 8.
10. Method of analyzing a signal by an analysis device (16) comprising N signal
processing channels (22), N being an integer superior or equal to 2, the method comprising,
on each processing channel (22), at least one step of:
- sampling at a sampling frequency (Fsi) by a sampling device (24), to obtain a sampled signal,
- filtering the signal sampled by a filter (28) having a cutoff frequency (Fc), to obtain
a filtered signal,
- conversion of the filtered signal by an analog-to-digital converter (26) operating at
a conversion frequency (Fso), the conversion frequency (Fso) being strictly greater than the sampling frequency (Fsi), to obtain a converted signal and
- analysis of the signal converted by a calculator (30).