Frequency estimator for aeronautical communication

Abstract

A method and device are provided suited to estimating a frequency value for aeronautical communication between a first station and an airborne system moving in relation to the first station, the data being transmitted in the form of a succession of frames, a frame comprising at least one first header field of known data followed by one or more fields of unknown data, and having at least the following steps: Step 1: performance of a supervised correlation on the known data (300) of the header, and estimation of a first frequency range centered on a frequency ; Step 2: production of a blind correlation on at least all of the unknown data of a field of data, by looking for a correlation peak over the frequency range +/, as determined in step 1, and by retaining of the frequency corresponding to the correlation peak.

Claims

1. A method for estimating a frequency value for aeronautical communication between a first station and an airborne system moving in relation to the first station, data being transmitted in a form of a succession of frames, a frame comprising at least one first header field of known data followed by one or more fields of unknown data, at least one of the first station and the airborne system comprising a radiocommunication receiver, the method comprising at least the following steps being executed by the radiocommunication receiver: Step 1: performance of a supervised correlation on the known data of the header by the radiocommunication receiver, and estimation of a first frequency range centered on an estimated frequency having a precision of +/ by the radiocommunication receiver, and Step 2: production of a blind correlation on at least all of the unknown data of a field of data by the radiocommunication receiver, by looking for a correlation peak over the range centered on the frequency +/, as determined in step 1, and by retaining the frequency corresponding to the correlation peak, said retained frequency being used for transmitting data between the first station and the airborne system, wherein the frame is modulated with a modulation by change of phase comprising a field of pilots (302.sub.1) that is arranged before a field of data and in that it has at least the following steps: Step 1: performance of a supervised correlation by Fourier transform on the field of pilots (302.sub.1), in order to obtain a range centered on an estimated frequency having a precision of +/, and Step 2: use of the symbols of the modulated frame and use of the Fourier transform from the modulated symbols of the short frame, corresponding to the range centered on the frequency +/ defined in step 1, determination of the frequency corresponding to the maximum of the norm of the Fourier transform and deduction of the value of the Doppler effect therefrom.

2. The method according to claim 1, wherein the frame is a frame of DVB-S2 type using a DVB-S2 QPSK type modulation t.

3. The method according to claim 2, wherein it comprises at least the following steps: Step 1: estimation of the frequency , $=\mathrm{MaxIndex}\left({.Math.\mathrm{DFT}\left(R\left(k\right)*\mathrm{conj}\left(\mathrm{Ref}\left(k\right)\right)+\mathrm{BABG}\left(k\right)\right).Math.}^2\right)=\mathrm{MaxIndex}\left({.Math.\mathrm{DFT}\left({}^{*2**{}_f*t}*{\mathrm{rect}}_{90}\left(t\right)+\mathrm{BABG}\left(t\right)\right).Math.}^2\right)=\mathrm{MaxIndex}\left({.Math.{\mathrm{sinc}}_{90*}\left(f-{}_f\right)+\mathrm{BABG}\left(f\right).Math.}^2\right)$ where: conj( ) conjugate function of a complex number DFT: Fourier transform k: index of a symbol t: given instant .sub.f: Doppler frequency to be estimated rect.sub.90(t): rectangle function of length 90 sin c.sub.90*(f): cardinal sine function of width 90* Ref(k): reference symbols for the received symbols R(k) of the header where 1k90 BABG: additive white Gaussian noise Step 2: estimation of the value of the Doppler from the expression $=\frac{\mathrm{MaxIndexNDA}\left({.Math.\mathrm{DFT}\left({\left(R\left(k\right)+\mathrm{BABGEq}\left(k\right)\right)}^4\right).Math.}^2\right)}{4}=\frac{\mathrm{MaxIndexNDA}\left({.Math.\begin{array}{c}\mathrm{DFT}({}^{*2**4*{}_f*t}*{\mathrm{rect}}_{\mathrm{NDALength}}\left(t\right)+\\ \mathrm{BABGEq}\left(t\right))\end{array}.Math.}^2\right)}{4}=\frac{\mathrm{MaxIndexNDA}\left({.Math.{\mathrm{sinc}}_{\mathrm{NDALength}*}\left(f-4*{}_f\right)+\mathrm{BABGEq}\left(f\right).Math.}^2\right)}{4}$ where: NDALength: length over which the blind frequency estimate is produced, Rect.sub.NDALength(t): rectangle function of length NDALength, sin c.sub.NDALength*(f): cardinal sine function of width NDALength*, BABGEq(f): equivalent additive white Gaussian noise stemming from raising to the power of 4 the noisy signal SymbsQPSK+BABG, MaxIndexNDA: frequency belonging to the frequency range FreqRangeNDA for which the norm of the Fourier transform is at a maximum.

4. The method according to claim 3, wherein the frame is modulated by BPSK or QPSK modulation.

5. The method according to claim 4, wherein the value of NDALength is fixed at 8370 for an SNR of 0 dB.

6. The method according to claim 5, wherein the value of NDALength is equal to $\frac{8370}{{\mathrm{SNRLin}}^2},$ where SNRLin corresponds to the signal-to-noise ratio SNR expressed linearly.

7. The method according to claim 2, wherein the frame is modulated by BPSK or QPSK modulation.

8. The method according to claim 7, wherein the value of NDALength is fixed at 8370 for an SNR of 0 dB.

9. The method according to claim 8, wherein the value of NDALength is equal to $\frac{8370}{{\mathrm{SNRLin}}^2},$ where SNRLin corresponds to the signal-to-noise ratio SNR expressed linearly.

10. The method according to claim 2, wherein a frequency estimate is determined for each DVB-S2 frame constituting the communication.

11. A device for estimating a Doppler in an aeronautical communication system comprising at least a first station and an airborne system moving in relation to the first station, data being transmitted in a form of a succession of frames, a frame comprising at least one first header field of known data followed by one or more fields of unknown data, the device comprising at least: a radiocommunication receiver implemented with a first module configured to produce a supervised correlation on the known data of the header, and to estimating a first frequency range centered on an estimated frequency having a precision of +/, and the radiocommunication receiver further implemented with a second module configured to produce a blind correlation on at least all of the unknown data of a field of data, by looking for a correlation peak over the range of said estimated frequency +/, and by retaining the frequency corresponding to the correlation peak, said retained frequency being used for transmitting data between the first station and the airborne system, wherein the frame is modulated with a modulation by change of phase comprising a field of pilots (302.sub.1) that is arranged before a field of data and in that it has at least the following steps: Step 1: performance of a supervised correlation by Fourier transform on the field of pilots (302.sub.1), in order to obtain a range centered on an estimated frequency having a precision of +/, and Step 2: use of the symbols of the modulated frame and use of the Fourier transform from the modulated symbols of the short frame, corresponding to the range centered on the frequency +/ defined in step 1, determination of the frequency corresponding to the maximum of the norm of the Fourier transform and deduction of the value of the Doppler effect therefrom.

12. The device according to claim 11, wherein the frames are QPSK- or BPSK-modulated DVB-S2 data.

13. The device according to claim 11 wherein the first estimation module and the second module are produced using FPGA technology.

14. A device for estimating a Doppler in an aeronautical communication system comprising at least a first station and an airborne system moving in relation to the first station, data being transmitted in a form of a succession of frames, a frame comprising at least one first header field of known data followed by one or more fields of unknown data, the device comprising at least: a radiocommunication receiver implemented with a supervised correlation device for the known data of the header, and to estimating a first frequency range centered on an estimated frequency having a precision of +/, and the radiocommunication receiver further implemented with a blind correlation device for at least all of the unknown data of a field of data, by looking for a correlation peak over the range of said estimated frequency +/, and by retaining the frequency corresponding to the correlation peak, said retained frequency being used for transmitting data between the first station and the airborne system, wherein the frame is modulated with a modulation by change of phase comprising a field of pilots (302.sub.1) that is arranged before a field of data and in that it has at least the following steps: Step 1: performance of a supervised correlation by Fourier transform on the field of pilots (302.sub.1), in order to obtain a range centered on an estimated frequency having a precision of +/, and Step 2: use of the symbols of the modulated frame and use of the Fourier transform from the modulated symbols of the short frame, corresponding to the range centered on the frequency +/ defined in step 1, determination of the frequency corresponding to the maximum of the norm of the Fourier transform and deduction of the value of the Doppler effect therefrom.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) Other features and advantages of the present invention will be better understood upon reading the description that follows, which is provided by way of illustration and is no way limiting, to which the figures are appended, in which:

(2) FIG. 1 shows an example of a transmission system,

(3) FIG. 2A and FIG. 2B respectively show an illustration of the acceleration A and the variation in altitude Alt in the presence of an air pocket and the Doppler effect D and the variation in Doppler effect, induced curve V,

(4) FIG. 3 shows a DVB-S2 frame,

(5) FIG. 4 shows an example of synchronisation structure according to the prior art,

(6) FIG. 5 shows an example of the structure of the frequency estimator according to the invention, arranged at a DVB-S2 receiver.

DETAILED DESCRIPTION

(7) The example that follows will be given for the frequency estimation, bursty or otherwise, in the case of a DVB-S2 frame with modulation by means of QPSK (Quadrature Phase Shift Keying) phase change, in order to better explain the subject matter of the invention, and without limiting the scope thereof to the structure of the DVB-S2 frame. The invention can also apply for communications for which the frames used have a header or field of known data followed by a plurality of fields of unknown data. By way of example, the unknown data are BPSK (Binary Phase Shift Keying) modulated, or QPSK modulated. In the detailed example below, the system will work for signal-to-noise ratios of between 0 and 6 dB, for example, in order to comply with the known Arinc 791 norm.

(8) The method and the frequency estimator according to the invention have a first supervised frequency estimator and a second blind frequency estimator. The frequency estimator according to the invention is implemented within a DVB-S2 receiver that is positioned, by way of example, at the satellite station or the ground station and uses programmable circuit or FPGA (field-programmable gate array) technology, for example.

(9) The first frequency estimation module or first estimator implements an algorithm that produces a supervised correlation using a Fourier transform on the DVB-S2 header of 90 symbols.

(10) The algorithm estimates the frequency by computing the frequency corresponding to the maximum of the norm of the Fourier transform of the 90 symbols (R(k) where 1k90) that are received from the DVB-S2 header and correlated with the reference symbols (Ref(k) where 1k90) of this same header.

(11) The received symbols k are as follows:
R(k)=(Ref(k)+AWGN(k))*exp(i*2*.sub.f*k), where: Ref(k): reference symbols for the symbols R(k) BABG: additive white Gaussian noise .sub.f: Doppler frequency to be estimated.
The frequency estimation at the first estimation module or estimator is effected as follows:

(12) $=\mathrm{MaxIndex}\left({.Math.\mathrm{DFT}\left(R\left(k\right)*\mathrm{conj}\left(\mathrm{Ref}\left(k\right)\right)+\mathrm{BABG}\left(k\right)\right).Math.}^2\right)=\mathrm{MaxIndex}\left({.Math.\mathrm{DFT}\left({}^{*2**{}_f*t}*{\mathrm{rect}}_{90}\left(t\right)+\mathrm{BABG}\left(t\right)\right).Math.}^2\right)=\mathrm{MaxIndex}\left({.Math.{\mathrm{sinc}}_{90*}\left(f-{}_f\right)+\mathrm{BABG}\left(f\right).Math.}^2\right)$
where: conj( ): conjugate function of a complex number DFT: Fourier transform k: index of a symbol t: given instant f: given frequency .sub.f: Doppler frequency to be estimated rect.sub.90(t): rectangle function of length 90 sin c.sub.90*.sub.(f): cardinal sine function of width 90*.
The operating range is no more than +0.5*Rs, because it is a supervised estimator.

(13) At the conclusion of the first estimation module, exhibiting high resistance to noise, a frequency value is obtained.

(14) The second estimation module or second estimator is a blind frequency estimator with very high frequency resolution that will work on the following frequency range:
FreqRangeNDA=[;+]

(15) This frequency range over which the maximum peak is sought is determined by the precision of the supervised frequency estimator: it is therefore centred on and is 0.0025 Rs in normalised mode (2.5 kHz for an R.sub.s of 1 Mbaud), which corresponds to the worst-case precision of the supervised estimator at 0 dB. Thus, at 0 dB:
FreqRangeNDA=[0.0025;+0.0025]

(16) When the SNR is 0 dB, the blind algorithm uses all the symbols of the short QPSK frame with pilots, that is to say 8370 symbols. The algorithm involves, at 0 dB, using the Fourier transform for the 8370 QPSK symbols raised to the power of 4 in order to estimate the Doppler by computing the frequency f corresponding to the maximum of the norm of this Fourier transform. The Doppler corresponds to this estimated frequency divided by 4. If the received QPSK symbols R(k) are considered, where 1kNDALength:
R(k)=(SymbsQPSK(k)+BABG(k))*exp(i*2*.sub.f*k), where: SymbsQPSK=exp(i*(/4+n*/2)), where n=[0, 1, 2, 3], NDALength: length over which the blind frequency estimate is produced. NDALength is fixed at 8370 when the SNR is 0 dB, BABG: additive white Gaussian noise, .sub.f: Doppler frequency to be estimated.
The frequency estimate at the second estimation module or estimator is effected as follows:

(17) $=\frac{\mathrm{MaxIndexNDA}\left({.Math.\mathrm{DFT}\left({\left(R\left(k\right)+\mathrm{BABGEq}\left(k\right)\right)}^4\right).Math.}^2\right)}{4}=\frac{\mathrm{MaxIndexNDA}\left({.Math.\begin{array}{c}\mathrm{DFT}({}^{*2**4*{}_f*t}*{\mathrm{rect}}_{\mathrm{NDALength}}\left(t\right)+\\ \mathrm{BABGEq}\left(t\right))\end{array}.Math.}^2\right)}{4}=\frac{\mathrm{MaxIndexNDA}\left({.Math.{\mathrm{sinc}}_{\mathrm{NDALength}*}\left(f-4*{}_f\right)+\mathrm{BABGEq}\left(f\right).Math.}^2\right)}{4}$
where: Rect.sub.NDALength(t): rectangle function of length NDALength, Sin c.sub.NDALength*.sub.(f):cardinal sine function of width NDALength*, BABGEq(f): equivalent additive white noise stemming from raising to the power of 4 the noisy signal SymbsQPSK+BABG, MaxIndexNDA: frequency belonging to the frequency range FreqRangeNDA for which the norm of the Fourier transform is at a maximum.
The maximum range of operation is +0.125*R.sub.s, because it is a blind estimator that takes a signal raised to the power of 4 as its input.

(18) FIG. 5 schematically shows a structure for a frequency estimator according to the invention that is implemented, by way of example, at the DVB-S2 receiver of the ground station, which has two modules (a supervised correlation device, a blind correlation device) adapted to performing the following steps: firstly, rough supervised estimation of the frequency is effected on the DVB-S2 header of 90 symbols: a first frequency estimate having a precision of +2.5 kHz (for an Rs of 1 Msps) is then obtained; secondly, a blind correlation is effected on the 8370 symbols of the short QPSK frame over a range of +2.5 kHz from the rough frequency estimation. An estimation having a precision of 20 Hz is then obtained for the frequency error.

(19) The use of the blind estimator over a reduced frequency range allows the FER owing to the blind estimator to be divided by the ratio between the total range of operation of the estimator and this reduced range. Thus, in the present case, the range of operation of the estimator being +/125 kHz, the FER is divided by 50 (125/2.5=50). This allows a change from an FER higher than 10.sup.5 to an FER close to 10.sup.6 at 0 dB.

(20) The frequency estimator implemented according to the present invention allows an FER lower than 10.sup.5 to be attained at 0 dB under adverse aeronautical conditions. For signal-to-noise ratio values SNRs higher than 0 dB, it is possible to obtain the same level of performance while reducing the magnitude of the blind estimation.

(21) The steps of the method that just been described apply in the case where the length over which the frequency is estimated corresponds to

(22) $\mathrm{NDALength}=\frac{8370}{{\mathrm{SNRLin}}^2}$
(where SNRL in corresponds to the signal-to-noise ratio SNR expressed linearly), the performance (FER lower than 10.sup.5 (frequency error rate) under worst-case aeronautical Doppler conditions) being the same for SNRs typical of DVB-S2 QPSK (between 0 and 7 dB).

(23) Advantages

(24) The invention notably has the advantage of providing an estimate of the frequency very precisely over a very short estimation period, which allows very large variations in the frequency owing to the movement of the aeroplane to be followed. It can thus be used in totally bursty fashion, that is to say that for each DVB-S2 frame a frequency estimate is provided, the latter being independent of the estimate obtained over the previous frame. Thus, if the frequency estimate over a frame is erroneous, this does not impact on the other frames and the loss is limited to the frame for which the estimate is erroneous.