Method for detecting signals in a frequency-ambiguous digital receiver, and digital receiver implementing such a method

Abstract

A digital receiver comprising at least two reception pathways, the method carries out a digital inter-correlation of the signals obtained as output from at least two filters of different central frequencies and different ranks, the rank and the central frequency of the filters being chosen as a function of a determined frequency-wise search domain. For a determined search domain, the various sampling frequencies of the reception pathways are chosen so that the ambiguous frequencies resulting from the spectral aliasings vary as a monotonic function of the true frequency of the signals.

Claims

1. A method for detecting signals in a frequency-ambiguous digital receiver by aliasing of the frequency spectra, said receiver comprising at least two reception pathways, wherein, with a sampling frequency for said signals being specific to each pathway, said method comprises: determining a frequency-wise search band having a maximum frequency, which is less than or equal to half the smallest sampling frequency from among the sampling frequencies of said pathways, said sampling frequencies of said pathways inside said search band having the same ambiguity rank; determining the sampling frequencies of said pathways such that the aliased frequencies, which correspond to said pathways, vary as a monotonic function of the true frequency of said signals in said search band; digitally filtering said signals in said reception pathways in banks of filters that each have (i) the same central frequency, which is a multiple of a quantity 1/T, and (ii) the same width, which is equal to said quantity 1/T, Fech.sub.l and Fech.sub.m being respectively the sampling frequency of any pathway of order 1 and of any pathway of order m, Fech.sub.l/L=Fech.sub.m/M=1/T, L and M being integers, wherein at least one of the at least two reception pathways is of the order 1, wherein at least one of the at least two reception pathways is of the order m, and wherein T is a duration of an observation window; in said search band, the frequency aliasing of the signal in the at least one pathway of the order 1 giving a response in the filter of rank i, and the frequency aliasing of the signal in the at least one pathway of the order m giving a response in the filter of rank k+i or ik, carrying out a computation of inter-correlation between (i) the signal arising from the filter of rank i of central frequency Fech.sub.l/L of the at least one pathway of the order 1 and (ii) at least the signal arising from the filter of rank i+k or ik of the at least one pathway of the order m, of central frequency (i+k)Fech.sub.m/M or (ik)Fech.sub.m/M, i and k being integers; and carrying out the detection of the signals by comparing a power of the signal on output from the inter-correlation computation to a threshold.

2. The method as claimed in claim 1, wherein disparities, between the aliased frequencies corresponding to said pathways, are constant in said search band.

3. The method as claimed in claim 1, further comprising: displacing said frequency-wise search band by modifying the sampling frequencies of said pathways.

4. The method as claimed in claim 1, wherein said signals are signals of low peak power and of long pulse duration.

5. The method as claimed in claim 1, wherein at least two of the at least two reception pathways are connected to each antenna of an amplitude goniometer, the amplitude goniometer comprising several antennas.

6. The method as claimed in claim 5, further comprising: receiving the signals on the two reception pathways of the antenna; and performing, for a given antenna, the inter-correlation computation between the received signals, a result of said computation affording access to an amplitude of the signals, which is necessary for estimating their direction of arrival.

7. The method as claimed in claim 5, further comprising: receiving signals of two adjacent antennas; and performing an inter-correlation computation between the received signals.

8. The method as claimed in claim 1, wherein one of the at least two reception pathways is linked to each antenna of a frequency-ambiguous interferometer.

9. The method as claimed in claim 8, further comprising: extracting a relative phase of a signal based on a plurality of inter-correlation computations performed between a plurality of pathways taken pairwise, wherein each reception pathway is associated with a different antenna.

10. The method as claimed in claim 1, further comprising: detecting, in parallel with the determination of the frequency-wise search band, the determinations of the sampling frequencies, the digital filtering of said signals, and the computation of the inter-correlation, pulsed signals.

11. A digital receiver, comprising at least two reception pathways, wherein the digital receiver is configured to execute the method as claimed in claim 1.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) Other characteristics and advantages of the invention will become apparent with the aid of the description which follows, given in relation to appended drawings, in which:

(2) FIGS. 1, 2 and 3 show examples of use of a digital receiver respectively in an amplitude goniometer, in an interferometer and in association with a computational beamforming antenna;

(3) FIG. 4 shows an illustration of the manner of operation of a digital receiver according to the prior art;

(4) FIG. 5 shows an illustration of the variation of the ambiguous frequency obtained by digital spectral aliasing in a receiver;

(5) FIG. 6 shows an illustration of a particular frequency search domain used by the invention;

(6) FIG. 7 shows an illustration of the operating principle of a receiver implementing the detection method according to the invention;

(7) FIG. 8 shows an exemplary application of the invention to an interferometer;

(8) FIG. 9 shows an exemplary application of the invention to an amplitude goniometer.

DETAILED DESCRIPTION

(9) FIGS. 1, 2 and 3 illustrate through basic diagrams examples of use of digital receivers. These receivers can be associated equally well with amplitude goniometry antennas, interferometric antennas or else computational beamforming antennas.

(10) In the case of an array of amplitude goniometry antennas comprising several antennas as illustrated by FIG. 1, a current configuration consists in associating six antennas 101, 102, 103, 104, 105, 106 distributed in a hexagon so as to cover 360 in the horizontal plane, each antenna having an aperture of the order of 60 to 90. In such a device, the detection of the direction of arrival 20 of a signal is obtained on the basis of the amplitudes received on the various antennas.

(11) In the case of an interferometry antenna as illustrated by FIG. 2, a commonplace configuration consists in using four antennas 21, 22, 23, 24 each covering approximately 90 horizontally, irregularly spaced to form in a given plane an ambiguous interferometric base. A 360 coverage may be obtained for example by using four identical bases of four antennas, the bases being distributed along the sides of a square. In such a device, the direction of arrival of a signal is estimated in each 90 sector by utilizing the phase differences of the signal received on the four antennas of one and the same base.

(12) In the case of a computational beamforming antenna as illustrated by FIG. 3, several antennas 31 generally having an aperture of the order of 90 horizontally are installed in a given plane, the signals received by these various antennas being summed in a coherent manner after digitization so as to simultaneously form several directional beams covering the observation domain. This device also uses the phase of the signal received on the antennas to estimate the direction of arrival, by comparing the amplitude of the signal received in the beams formed in the adjacent directions. This solution is seldom used because of the volume of computations to be performed.

(13) For all these devices in particular, it is necessary to solve the problem of the detection and of the measurement in terms of frequency of the signals of long duration and of low peak power on the basis of a broadband digital receiver with multiple sub-sampling. Currently, the sensitivity of broadband receivers, be they analog or digital, is insufficient to detect these signals of long duration and low peak power.

(14) According to the invention, as will be described subsequently, specific means allowing the detection of the signals of low peak power and of long duration by carrying out an inter-correlation of the received signals sampled at different frequencies are integrated into a broadband digital receiver.

(15) FIG. 4 illustrates the operating principle of a digital receiver according to the prior art having a very broad reception band, typically from 2 to 18 GHz. The receiver comprises four reception pathways 1, 2, 3, 4 each having at input an analog-digital coder 41, 42, 43, 44. The four sampling frequencies Fech.sub.1, Fech.sub.2, Fech.sub.3, Fech.sub.4 of the coders are different, lying for example between 2 and 3 GHz, thus corresponding to the state of the art for analog coders. The four reception pathways can be linked to antennas which may or may not be different, according to the type of array of antennas that is used.

(16) In such a receiver, the various frequencies Fech.sub.1, Fech.sub.2, Fech.sub.3, Fech.sub.4 are chosen so that the ambiguity in the frequency measurement is removed with a sufficient margin of safety in the presence of noise, thus requiring a sufficient spacing between the various frequencies, typically of the order of a few tens of MHz. These various sampling frequencies must also be chosen so that their lowest common multiple is greater than the total analysis band in respect of the received signals. Moreover, because of Shannon's theorem, the instantaneous bandwidth being limited to a value of less than half the lowest sampling frequency, the sampling frequencies must be chosen to be as high as possible. Finally, it is practical to use a constant spacing F between the sampling frequencies since this makes it possible in particular to use simple algorithms to remove the distance-wise ambiguities.

(17) These constraints and facilities lead for example to choosing sampling frequencies as follows, M being an integer number:
Fech1=MF;Fech2=(M+1)F;Fech3=(M+2)F;Fech4=(M+3)F(1)

(18) For example, if the sampling frequency of the coders is of the order of 2 GHz, it is possible to choose, by taking M=52: Fech1=2080 MHz, i.e. 5240 MHz; Fech2=2120 MHz, i.e. 5340 MHz; Fech3=2160 MHz, i.e. 5440 MHz; Fech4=2200 Mhz, i.e. 5540 MHz.

(19) FIG. 5 illustrates the variation of the ambiguous frequency obtained by digital spectral aliasing as a function of the frequency of the input signal, for each of the four sampling frequencies of the example hereinabove, in a system of axes where the abscissae represent the true frequency of the received signals and the ordinates the ambiguous frequency detected. Four curves 51, 52, 53, 54 represent the ambiguous frequencies as a function of the true frequencies for the four sampling frequencies Fech1, Fech2, Fech3, Fech4.

(20) For each sampling frequency, this ambiguous frequency, forming an aliased frequency, can be written in the following manner:
Famb.sub.i=Ftruekamb.sub.iFech.sub.i if Ftrue>kamb.sub.iFech.sub.i(1)
Famb.sub.i=kamb.sub.iFech.sub.iFtrue if Ftrue<kamb.sub.iFech.sub.i(2)
where: Ftrue is the true frequency of the input signal of the receiver; Fech.sub.i is the sampling frequency on pathway i, i varying from 1 to 4; kamb.sub.i is the ambiguity rank, corresponding to the integer value of (Ftrue/Fech.sub.i)+.
kamb.sub.i is therefore equal to the integer value of (Fs/Fech.sub.i)+, Fs being the frequency of the received signal, at the input of the receiver.

(21) FIG. 5 shows in particular that the frequency measurement obtained after aliasing on the four reception pathways makes it possible to determine without ambiguity the sub-band 50 in which the frequency of the received signal is situated, the disparities between the aliased frequencies being a one-to-one function of the ambiguity rank, more precisely a disparity 59 characterizing a sub-band. Thus, knowing this sub-band 50 and the ambiguous frequency, the true frequency is deduced therefrom.

(22) Returning to FIG. 5, in practice, the spectral analysis 45, 46, 47, 48 may be performed for example by digital Fourier transform (DFT), or else with the aid of polyphase filters. On the basis of the spectral analysis, the detection and the estimation of the aliased frequency is performed 401, 402, 403, 404 on each pathway, followed by the estimation 49 of the true frequency of the received signal by ambiguity removal.

(23) To remove the ambiguity and preserve the relative phase between the pathways 1, 2, 3, 4, the same analysis resolution is used on the various pathways. For example, if a frequency resolution of 10 MHz is desired, it is possible to choose: For the sampling frequency Fech.sub.1=2080 MHz, a DFT on a number of points N1=208 For the sampling frequency Fech.sub.2=2120 MHz, a DFT on a number of points N2=212 For the sampling frequency Fech.sub.3=2160 MHz, a DFT on a number of points N3=216 For the sampling frequency Fech.sub.4=2200 MHz, a DFT on a number of points N4=220
thus corresponding to:
Fech.sub.1/N1=Fech.sub.2/N2=Fech.sub.3/N3=Fech.sub.4/N4=1/T(3)
T being the duration of the observation window, 1/T being equal to 10 MHz in this example.

(24) This mode of detection and estimation of the signal frequency as described hereinabove in relation to FIG. 5 is defeated when the signal-to-noise ratio is insufficient to allow detection 401, 402, 403, 404 in each elementary reception pathway.

(25) Moreover, the frequency of the input signal of the receiver being unknown, it is not possible to know the way in which the aliasings 51, 52, 53, 54 are performed and therefore the filters in which the signal is present on the various pathways 1, 2, 3, 4. This prohibits a priori the possibilities of recombining of the received signals between the various pathways aimed at increasing the signal-to-noise ratio, so as to allow detection. Such is the case in particular in the presence of long signals of low peak power.

(26) FIG. 6 illustrates a characteristic of the invention.

(27) According to the invention, on the basis of a determined and limited frequency domain 61 in which it is desired to seek to detect a continuous emission or a long pulse emission of low peak power, the various sampling frequencies are chosen in such a way that the disparities between aliased frequencies corresponding to the various reception pathways 1, 2, 3, 4 are constant in this determined frequency domain 61.

(28) This is obtained, if: Firstly, the ambiguity rank of the frequency obtained after aliasing is identical for all the sampling frequencies inside the frequency search domain. Stated otherwise, the search domain 61 is contained in one and the same ambiguity rank in relation to the sampling frequencies of the various pathways. In the example of FIG. 6, we must have: kamb.sub.1=kamb.sub.2=kamb.sub.3=kamb.sub.4=kamb, i.e.: integer value (Ftrue/Fech.sub.1)=integer value (Ftrue/Fech.sub.2)=integer value (Ftrue/Fech.sub.3)=integer value (Ftrue/Fech.sub.4) Secondly, in the interval corresponding to the frequency search domain, if the ambiguous frequency is a monotonic function of the true frequency of like direction of variation for the various sampling frequencies.

(29) For any true frequency value, Ftrue, contained in the frequency domain 61 defined by bounds F.sub.min, F.sub.max, we then have: For an increasing direction of variation, that is to say for the integer values of kamb: F.sub.minkambFech.sub.max and F.sub.max<(kamb+)Fech.sub.min For a decreasing direction of variation: F.sub.min(kamb+)Fech.sub.max and F.sub.max<(kamb+1)Fech.sub.min.

(30) Fech.sub.min is the smallest of the sampling frequencies of the various pathways. In the present example, Fech.sub.min is Fech.sub.1. Likewise, Fech.sub.max is the largest of the sampling frequencies and Fech.sub.max is Fech.sub.4.

(31) For example, by choosing the increasing direction of variation of the ambiguous frequency, it is possible to choose:
Fech.sub.maxF.sub.min/kamb.

(32) For a frequency domain 61 limited to F, it follows that:
FF.sub.maxF.sub.min, i.e.:
Fkamb(Fech.sub.minFech.sub.max)+Fech.sub.min/2.

(33) And by using relationship (3), we obtain for four sampling frequencies:

(34) $F\left(\frac{M}{2}-3\mathrm{kamb}\right)F.$
For example, if the low bound of the search domain is fixed at
F.sub.min=9 GHz,
it is possible to choose, having regard to the maximum accessible sampling frequencies of the order of 2 GHz, by taking as sampling frequency step size F=40 MHz: Kamb=4 Fech.sub.4=Fech.sub.max=(M+3)F the integer nearest to F.sub.min/Kamb=2.25 GHz, i.e. 5640 MHz=2.24 GHz
with:
M+3=56, i.e. M=53.

(35) Hence: Fech.sub.3=5540 MHz=2.2 GHz; Fech.sub.2=5440 MHz=2.16 GHz; Fech.sub.1=5340 MHz=2.12 GHz.

(36) And for a frequency search domain bounded by F580 MHz, we obtain F.sub.max=9.58 GHz.

(37) After having fixed the sampling frequencies in accordance with the foregoing, the disparity of the aliased frequencies between the various reception pathways 1, 2, 3, 4 in the search domain 61 is determined in a following step.

(38) In this domain 61, and as illustrated by FIG. 6, by retaining only the assumption of the monotonic increasing aliased frequency, we obtain: Ftrue=kamb Fech.sub.1+i.sub.1 Fech.sub.1/N1 Ftrue=kamb Fech.sub.2+i.sub.2 Fech.sub.2/N2 Ftrue=kamb Fech.sub.3+i.sub.3 Fech.sub.3/N3 Ftrue=kamb Fech.sub.4+i.sub.4 Fech.sub.4/N4
where i.sub.1, i.sub.2, i.sub.3, i.sub.4 represent the indices, or the ranks, of the filters 71, 72, 73, 74 of the pathways 1, 2, 3, 4 respectively, in which the input signal is aliased, N1, N2, N3, N4 having been defined previously. These filters are illustrated in the diagram of FIG. 7 presenting these four pathways. These filters 71, 72, 73, 74 are contained in banks of filters 701, 702, 703, 704. The spectral analysis can be obtained via a direct Fourier transform (DFT) or via polyphase filters. Stated otherwise, the banks of filters can be of the DFT or polyphase type for example.

(39) According to relationship (3) specifying that:
Fech.sub.1/N1=Fech.sub.2/N2=Fech.sub.3/N3=Fech.sub.4/N4=1/T
it follows that:
i.sub.2i.sub.1=kamb.F.T
i.sub.3i.sub.1=2kamb.F.T
i.sub.4i.sub.1=3kamb.F.T.

(40) For example, for the above example, if the spectral analysis resolution is fixed at 10 MHz, corresponding to T=100 ns, for kamb=4 and F=40 MHz:
i.sub.2i.sub.1=16
i.sub.3i.sub.1=32
i.sub.4i.sub.1=48.

(41) Knowing the rank i.sub.1 involved in the first bank of filters 701, the ranks i.sub.2, i.sub.3, i.sub.4 of the filters involved in the other banks 702, 703, 704 are deduced therefrom.

(42) Having thus identified the differences of index of the filters involved 71, 72, 73, 74, a following step consists in performing complex inter-correlations 75, 76, 77 of the signals arising from these filters between the various pathways.

(43) This inter-correlation is performed over a long time, typically of the order of 100 microseconds, in accordance with the type of inter-correlation described for example in patent application FR 1400514.

(44) It makes it possible not only to obtain a sufficient signal-to-noise ratio for detection, but also to extract the phase difference between the reception pathways, for example in the case of an interferometer, or else the amplitude difference between two adjacent antennas, in the case of an amplitude goniometer.

(45) In this process, the output of the filter of index i.sub.1 of pathway 1 will be correlated with the output of the filter of index i.sub.2 of pathway 2 and for example, for 4 reception pathways, the output of the filter of pathway 1 will also be correlated with the output of the filter i.sub.3 of pathway 3 and i.sub.4 of pathway 4, as illustrated by the example of FIG. 7.

(46) This process is performed for all the indices of filters whose central frequencies are contained in the search domain.

(47) In a following step 78, the results of the inter-correlations are thereafter compared in terms of amplitude with a threshold so as to ensure detection of the signals and identify the indices of the filters corresponding to the signal so as to estimate the frequency thereof. The amplitude and the phase of the inter-correlations are also stored at this level, for example so as to extract the direction of arrival of the signal.

(48) After having thus utilized a search domain 61, it is thus possible to choose a new domain by defining a new set of sampling frequencies.

(49) The invention has been described by way of example in respect of a digital receiver with four pathways. It applies more generally in respect of a receiver comprising at least two different pathways corresponding to different sampling frequencies Fech.sub.l, Fech.sub.m in banks of filters 701, 702, 703, 704 of like central frequency which is a multiple of a quantity 1/T, where 1/T=Fech.sub.l/L=Fech.sub.m/L, and of like widths 1/T=Fech.sub.l/L=Fech.sub.m/L, L and M being integer numbers, T being the duration of the observation window.

(50) Advantageously, the invention can in particular be applied to an interferometer or to a receiver with amplitude goniometry.

(51) FIG. 8 illustrates the application to an interferometer. The system illustrated by FIG. 7 is connected, in terms of reception, to the antennas 21, 22, 23, 24 of the device of FIG. 2. More precisely, coders 41, 42, 43, 44 each receive as input a signal arising from an antenna 21, 22, 23, 24, optionally via an interface which is not shown. In accordance with the above description, a first coder 41 samples at the frequency Fech.sub.1, a second coder at the frequency Fech.sub.2, a third coder at the frequency Fech.sub.3 and a fourth coder at the frequency Fech.sub.4, the frequencies preferably being different.

(52) The digital signals arising from the coders are thereafter processed in accordance with the method according to the invention, as output from spectral analyses 45, 46, 47, 48. The inter-correlation as described hereinabove is performed in particular between the received signal of the first antenna 21 with the received signals of the other antennas 22, 23, 24, the filter of index being associated with the first reception pathway linked to the first antenna 21. Stated otherwise, each reception pathway, of different sampling frequency, being associated with a different antenna, the relative phase of the signals is extracted on the basis of the various inter-correlation computations performed between the various pathways taken pairwise. The direction of arrival of the signals is obtained with the aid of this phase.

(53) The removals of ambiguity of the frequency measurement and of the interferometry phase measurement are thus performed in a single operation resulting from the inter-correlation products, as is described in relation to FIG. 7.

(54) FIG. 9 illustrates the application to an amplitude goniometer as illustrated by FIG. 1. In a particular embodiment, at least two reception pathways 1, 2 operating at two sampling frequencies are connected to each antenna 101, 102, 103, 104, 105, 106. For the sake of clarity, in FIG. 9, only two antennas 103, 104 are linked to reception pathways, each being linked to two reception pathways.

(55) On a given antenna 104, the inter-correlation is performed on two pathways 1, 2 of different sampling frequencies. This inter-correlation computation affords access to the measurement of amplitude of the signal necessary for estimating the direction of arrival. The removal of ambiguity in the frequency measurement is for example performed with the aid of the frequency-aliased signals obtained on the basis of the same signal sampled with the aid of two different frequency pairs. This signal may originate from one and the same antenna 104, or two adjacent antennas 103, 104 receiving it simultaneously. Coverage over 360 is obtained with the aid of 12 reception pathways for a goniometric device with 6 antennas, each antenna being linked to two pathways.