Method and device to correct the variation in the QDM of a navigation system

Abstract

Disclosed is a method to correct the drift in the phase shift (ΔΦ,θ) between two signals (S1) and (S2), said signals being generated by a signal having a reference frequency Fe or a corresponding period Ω, the drift in the phase shift resulting from a drift in frequency Fe or in period, according to a given relationship R((F,Ω), (θ, ΔΦ)). The method comprises at least the following steps:

Description

BACKGROUND OF THE INVENTION

[0001] 1. Field of the Invention

[0002] The present invention relates to a method and device to correct the variation or drift of the phase difference ΔΦ between two signals generated by a frequency signal Fe=2ΦΩ, a drift in the value of Fe prompting the drift in the value of the phase difference.

[0003] The invention can be applied especially to the correction of the drift in the value of the QDM (the code commonly used in navigation to designate the angle θ between a moving body and the magnetic north) in order to meet certain standards such as EUROCAE ED-22B, RTCA, ARINC MARK 2 AIRBORNE VOR RECEIVER 711-9. This drift results from a variation in the rotation frequency.

[0004] The invention can be applied for example but not exclusively, to air, sea or river navigation.

[0005] 2. Description of the Prior Art

[0006] FIG. 1 shows the principle of operation of a system used, on board a moving body or aircraft, for example an aircraft, to determine the direction of this moving body at a given place, as seen from a radio beacon. A continuous wave is sent in a cardioidal radiation pattern rotating at 30 rpm. At the same time, an omnidirectional antenna sends a reference signal at 30 Hz with frequency modulation. The on-board receiver detects the amplitude modulation at 30 Hz caused by the rotation of the pattern and the reference signal. The azimuth of the receiver seen from the station or VOR beacon1 is equal to the phase shift between the two signals E1 and E2. E1 is the modulation due to the rotation of the pattern and E2 is the reference transmitted.

[0007] A variation in the 30 Hz frequency prompts a linear drift in the phase difference and the beat introduced into the signal as can be seen in FIGS. 2 and 3.

[0008] FIG. 4 is an exemplary processing chain according to the prior art.

[0009] A standard digital VOR receiver technique consists in applying a separator filter to process two frequency bands, the AM modulated low-frequency band and the FM modulated high-frequency band, applying a frequency discriminator to recover the FM modulating frequency and comparing the phases of the two signals.

[0010] The composite signal VOR, after reception by a receiver 10, is separated by means of a separator 11 into an AM modulated low-frequency signal (LF) and an FM modulated high-frequency signal (HF).

[0011] The low frequency signal (LF) is sent to a linear filter 12 and then sampled 13 in order to determine its phase Φ1 by means of a Fourier transform.

[0012] The high frequency signal (HF) is sent first of all to a non-linear filter 14 and then sampled 16 in order to determine its phase Φ2 in implementing a Fourier transform. The non-linear signal has a frequency discriminator 15 which, in particular, has the function of recovering the FM modulating frequency.

[0013] Using the two phases Φ1 and Φ2, the value of the phase shift ΔΦ corresponding to the QDM (plus or minus the parameter of the phase-shift terms introduced by the filters) is determined for example by a computer programmed accordingly.

[0014] During the signal processing method, the operation of sampling the (LF) and (HF) signals by Fourier transform may generate spurious lines when the period of the analyzed signal is not a multiple of the temporal window of analysis.

[0015] To overcome this problem, there are known ways of using standard windows such as a generalized Hamming window to reduce the side lobes of the spectrum while widening the major lobe or the Kaiser lobe associating an infinite spectrum with a time signal with finite support. These two methods however have limited performance characteristics and disturb the signal-to-noise ratio for equal integration time.

[0016] It is also possible to use a DFT type filter bank method. In this case, the computation power needed is very great and leads to investment costs that are incompatible with certain applications.

[0017] FIG. 3 gives a view, in a referential system where the X-axis is the time axis and the Y-axis represents the drift in the QDM, of:

[0018] The curve (I) which corresponds to the value QDM(Fe) for the reference frequency Fe or the corresponding period Ω.

[0019] The curve (II) which corresponds to the drift in the QDM and the curve (III) which corresponds to a beat prompted by the drift in the value of the reference frequency Fe.

[0020] Hereinafter in the description, the term QDM designates the angle taken with respect to the magnetic north, also known as the bearing.

[0021] The invention consists especially of the use of a relationship linking the drift in the QDM with a variation in the reference frequency and a frequency estimator carefully positioned in the processing chain to correct this drift.

[0022] It can also use a trapezoidal sampling window, for example to carry out the Fourier transform.

SUMMARY OF THE INVENTION

[0023] The invention relates to a method to correct the drift in the phase shift (ΔΦ, θ) between two signals (S1) and (S2), said signals being generated by a signal having a reference frequency Fe or a corresponding period Ω, the drift in the phase shift resulting from a drift in frequency Fe or in period depending on a given relationship R((F,Ω), (θ, ΔΦ)).

[0024] The invention comprises at least the following steps:

[0025] determining the value of the phase shift (ΔΦ, θ) from the phase Φ1 of the signal S1 and the phase Φ2 of the signal S2,

[0026] estimating Fest the value of frequency Fe, or the corresponding period Ω,

[0027] bringing this value (ΔΦ, θ) into an interval of given values [QDMmin, QDMmax] containing the value QDM(Fe) obtained for the reference value Fe taking account of the values ΔΦ and Fest and of the given relationship R((F,Ω), (θ,ΔΦ)).

[0028] According to one embodiment, the method comprises, for example, a step of processing by Fourier transform using a sampling window whose shape is substantially trapezoidal.

[0029] The invention also relates to a method to correct a drift in the phase shift (ΔΦ, θ) between two signals, S1 having a phase Φ1 and S2 having a phase Φ2, the two signals being generated by a signal with a frequency Fe or with the corresponding period, the drift in the phase shift resulting from the drift in Fe or the drift in the period according to a given relationship R((F,Ω), (θ, ΔΦ)). The method comprises at least one step to determine at least one of the phases (Φ1, Φ2) by Fourier transform using a substantially trapezoidal sampling window.

[0030] The invention also relates to a device for the correction of the drift in the phase shift (ΔΦ,θ) between two signals (S1) and (S2), said signals being generated by a signal with a reference frequency Fe, the drift being linked to the drift in frequency Fe according to a given relationship R((F,Ω), (θ, ΔΦ)). The device comprises at least:

[0031] a separator of the signals S1 and S2, means to determine the value of the phase Φ1 and Φ2 for each of the signals and means to determine the value of the phase shift ΔΦ,

[0032] an estimator of the frequency Fe or of the corresponding period positioned after the separator,

[0033] a device comprising, in memory, the relationship R((F,Ω), (θ, ΔΦ)) receiving the values ΔΦ and the estimated frequency or the estimated corresponding period, adapted to bringing the value of ΔΦ to a value within an interval of given values [QDMmin, QDMmax] containing the value QDM(Fe) obtained for the reference value Fe (or period).

[0034] According to one embodiment, the device comprises for example means to process the signals (S1) and/or (S2) by Fourier transform, the sampling window having a substantially trapezoidal shape.

[0035] The estimator to estimate the frequency value may be of the Prony's estimator type.

[0036] The invention also relates to a VOR receiver comprising a device for the correction of a drift in the phase shift (ΔΦ, θ) between two signals (S1) and (S2), said signals being generated by a signal having a reference frequency Fe, the drift being linked to the drift in frequency Fe or to the drift in period according to a given relationship R((F,Ω), (θ, ΔΦ)), comprising at least:

[0037] a separator of the signals S1 and S2, means to determine the value of the phase Φ1 and Φ2 for each of the signals and means to determine the phase shift value ΔΦ,

[0038] an estimator of the frequency Fe or of the corresponding period positioned after the separator,

[0039] a device comprising, in memory, the relationship R((F,Ω), (θ, ΔΦ)) receiving the values ΔΦ and the estimated frequency or the estimated corresponding period, and adapted to bringing the value of ΔΦ to a value within an interval of given values [QDMmin, QDMmax] containing the value QDM(Fe) obtained for the reference value Fe (or period).

[0040] The methods and devices according to the invention can be applied for example to reducing the variation of the QDM of a navigation system or radio navigation system.

[0041] The method and device according to the invention can be used especially to achieve compliance with stringent standards, such as RTCA and EUROCAE standards, or to approach them as closely as possible. The invention also improves the limit of detection as compared with the prior art devices.

BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The invention will be understood more clearly from the following description, given by way of an illustration that in no way restricts the scope of the invention, of an example designed to reduce the variation in the QDM of a VOR radio navigation system where:

[0043] FIG. 1 shows the principle of operation in VOR mode,

[0044] FIG. 2 shows a view, in a frequency-degree (radian) reference system, of the linear drift in the phase shift as a function of a frequency drift,

[0045] FIG. 3 shows the drift and the beat resulting from the drift in the frequency value (time-amplitude pattern in degrees of phase shift),

[0046] FIG. 4 gives a schematic view of an exemplary processing chain for the signal received by the VOR according to the prior art,

[0047] FIG. 5 shows an exemplary processing chain for the signal according to the invention,

[0048] FIG. 6 gives a schematic view of the QDM obtained without the frequency estimator (prior art) or with the frequency estimator,

[0049] FIG. 7 gives a schematic view of a window of analysis of the signal having a trapezoidal shape,

[0050] FIG. 8 shows the spectra expressed in dB for different sampling windows,

[0051] FIGS. 9 and 10 show the attenuation of the phase anomalies obtained in implementing the trapezoidal window,

[0052] FIGS. 11 and 12 show the correction of the QDM obtained by implementing the method according to the invention.

MORE DETAILED DESCRIPTION

[0053] In order to obtain a clearer understanding of the principle implemented in the invention, the following description, given by way of an illustration that in no way restricts the scope of the invention, relates to the correction of the variation of the QDM in a system of navigation or radio navigation following a drift in the rotation frequency. This procedure can be used especially to meet the above-mentioned standards or to approach them to the utmost possible extent.

[0054] Without departing from the context of the invention, the different variants of implementation of the method and the device according to the invention can be applied advantageously to any system where a drift in the reference frequency (or in the corresponding period) will lead to a variation or a deviation of a phase difference ΔΦ between two signals.

[0055] FIG. 5 gives a schematic view of an exemplary embodiment for the processing chain according to the invention.

[0056] The VOR modulation received by an aircraft is expressed by the following expression (1):

K 0 +k 3 cos(ωt)+k1 cos(Ωt−θ)+k2 cos(332Ωt−msin(Ωt))

[0057] The composite VOR signal is received by a VOR receiver 20 according to a method known to those skilled in the art. It is then transmitted to a separator 21 so as to obtain a first low frequency (LF) signal S1 corresponding to the reference frequency Fe, equal for example to 30 Hz, or to the corresponding period Ω and a second high frequency (HF) signal S2.

[0058] The signal (LF) having a sinusoidal shape with a variable phase cos (Ωt −θ) is first of all filtered by a linear filter 22 and then sampled 23 by means of a microprocessor or a digital computer by implementation of a discrete Fourier transform or DFT to obtain the phase value Φ1 with, for example, Φ1=(Ωt −θ)−L, L being the delay introduced by the filter. A part of this signal is routed at output of the linear filter 22 towards a frequency estimator 24 in order to determine the estimated value Fest of the reference frequency Fe or again the estimated value Ω est of the corresponding period Ω.

[0059] The linear filters are, for example, half-band filters introducing a constant phase shift. The filters as well as the separator are for example digital devices known to those skilled in the art.

[0060] The signal (HF) with a sinusoidal shape, modulated at 332 times the reference frequency Fe equal to 30 Hz and containing the reference phase (cos(332Ωt−msin(Ωt)), is filtered through a non-linear filter 25 and then sampled by means of a Fourier transform 26 in order to obtain the phase value Φ2 corresponding to the value Ωt plus or minus the parameter I of the delay introduced by the discriminator, with Φ2=Ωt−I, for example.

[0061] Using the two values of the phases Φ1 and Φ2, the method determines the value of the phase shift ΔΦ, for example according to a computation known to those skilled in the art, implemented by means of a computer.

[0062] The processing chain also has a device 27 containing, in memory, the relationship R((F,Ω), (θ, ΔΦ)) linking the drift in bearing to the drift in the reference frequency or again to the drift in the corresponding period Ω. The control logic 27 receives the value of the estimated frequency Fest or the value of the estimated period Ωest and the value of the bearing θ or of the phase shift ΔΦ, which makes it possible, on the basis of the relationship R((F,Ω), (θ, ΔΦ)), to bring the drift in angle (θ,ΔΦ) to a value within an interval of given values [QDMmin, QDMmax] containing, for example, the value QDM(Fe) obtained by the reference frequency Fe or around a given value such as the value QDM(Fe).

[0063] The value of the angle θ is computed, for example, from the value of the phase shift ΔΦ in correcting it by means of a value corresponding to the values of the delays L and I introduced by the linear and non-linear filters.

[0064] This correction can also be made after the drift in angle is brought to a value within the above-mentioned interval of given values.

[0065] In this exemplary embodiment, the relationship R((F,Ω), (θ, ΔΦ)) linking the drift in the value of the reference frequency Fe and the drift of the value QDM is a linear relationship as described in FIG. 2. In this FIG. 2, the X-axis represents the value of the frequency F varying in an interval [29.5 Hz; 30.5 Hz] around the reference value Fe=30 Hz and the Y-axis represents the value of the drift in the QDM expressed in radians or in degrees.

[0066] The boundary values QDM max and QDM min are, for example, chosen as a function of the standard to be met, for example one of the standards referred to here above.

[0067] The relationship R((F,Ω), (θ, ΔΦ)) can be set up in the course of preliminary trials in which, for different frequency values varying around a given reference value Fe, the phase shift value is recorded in order to establish a database ((F, Ω), (ΔΦ, θ)).

[0068] The Fourier transforms, the computation of the phase shift as well as the correction in the value of the phase shift are achieved by example by means of a microprocessor or computer programmed according to a method known to those skilled in the art.

[0069] The frequency estimator 24 is positioned at the end of the processing chain, especially after the filter. This frequency estimator is for example a Prony's estimator described in the IEEE document Signal Processing Letters, Vol. 3, No. 2, February 1996 “Instantaneous Frequency Estimation Using Linear Prediction with Comparisons to the DESAs”.

[0070] FIG. 6 shows a time-(degree or radian) pattern with the correction made in the value of the QDM in implementing the steps of the method according to the invention. The curve (III) corresponds to the value of the drift in QDM obtained without the use of a frequency estimator. It is brought (curve IV) towards the value of the QDM obtained for the reference value Fe represented by the straight line (I).

[0071] The digital example that follows is given for a reference signal Fe substantially equal to 30 Hz in the context of the VOR application.

[0072] In this case, the output of the linear filter 22 is a 30 Hz sinusoidal form for which the amplitude of the spectrum of the unwanted harmonics that compose it can be attenuated by an 8th-order Butterworth filter placed before the frequency estimator 24.

[0073] The VOR signal is sampled at 26 560 Hz and the input of the Butterworth filter is a sinusoidal form sampled at 207.5 Hz. Therefore, only seven samples represent a signal period. Among these seven samples, only four are used to estimate the frequency. The frequency or the period are generally estimated and a control logic corrects the QDM when it remains stable and within standards, for example (29.55 Hz; 30.45 Hz).

[0074] The experimental precision of the value of the estimated frequency is 10−4 without energy losses to eliminate static error.

[0075] By acting in this way, it is possible to comply with the standards described here above and to remain within the tolerance of variation in QDM of ±0.4° with respect to the value QDM(Fe) despite a variation in the frequency Fe=30 Hz of ±1.5%. This corresponds for example to an interval of values QDMmin=(QDM(Fe)−0.4°) and QDMmax=(QDM(Fe)+0.4°).

[0076] According to one mode of implementation of the method, the method uses a trapezoidal or substantially trapezoidal window to perform the Fourier transform of the signal in order to limit the initial phase and final phase effects in the spectral analysis. The amplitude of the signal is thus reduced linearly on the frequency period to be analyzed. The principle consists in bringing the frequency curves closer to each other at the position of the discontinuity in a sort of impossible curve.

[0077] An exemplary trapezoidal window is shown in FIG. 7 in a standardized (time or number of samples)-amplitude diagram. It can be used to attenuate or erase the appearance of high frequencies due to the periodization when the signal is different from a multiple period of the window of analysis. The edges of the signal are attenuated at the beginning and/or at the end of the window of temporal analysis.

[0078] FIG. 8 has shown the spectrum expressed in dB respectively for a rectangular window (RECT), a triangular window (TRIA) and a trapezoidal window (TRAP) according to the invention.

[0079] This figure shows the small difference between the spectrum of the trapezoidal window and that of the rectangular window in terms of energy, width of the major lobe and width of the side lobe. The trapezoidal type window is used to approach the gain obtained by a rectangular type window.

[0080] The X-axis corresponds to the field of the frequencies and the Y-axis to the standardized amplitude of the lobe expressed in dB. The vertical axis corresponds to the middle of the lobe.

[0081] The conventional approach is to multiply the sample signal by a rectangular window which is the reference window. This approach has the drawback of bringing out anomalies in the signal or phase leap referenced 30 in FIGS. 9 and 10.

[0082] The use of a trapezoidal window as described in FIG. 7 attenuates this observed anomaly 30 as can be seen in FIG. 10.

[0083] In these two FIGS. 9 and 10, the X-axis corresponds to the time domain or to the number of sampling points at a given sampling frequency and the Y-axis to the amplitude of the standardized signal.

[0084] The parameters of the window, for example the time on which the window shows an attenuation at the beginning and at the end, is equal for example to 2.5 times the period of the signal to be sampled. The trapezoidal type of shape can be shown by means of two straight-line portions D1 and D2 connected by a straight line D3, the slope of the straight lines D1 and D2 being computed to obtain the above-mentioned attenuation for example.

[0085] In the application to the correction of the value of the phase shift or QDM at the level of a VOR receiver, the energy is a major parameter because it directly conditions the sensitivity of the receiver. The choice of the sampling window results from a compromise by which it is possible to obtain high separating power for the window, minimize the loss of energy and obtain a high signal-to-noise ratio.

[0086] The use of a trapezoidal window also means that an initial offset can be applied to the amplitude (a sort of raised trapezoid) so as to reduce the energy loss to the minimum necessary.

[0087] Since this attenuation can reduce the total signal-to-noise ratio of the receiver, the goal is to minimize it in order to maintain the same integration time, compatible with the desired reception level on the one hand and the response time of the equipment on the other hand.

[0088] The following table brings together the characteristics of the different windows.

		Width of the	Coherent gain
	Side lobe	slope at 3 dB	(standardized)
Window	(dB)	(BINS)	energy loss
Rectangle	−13	0.89	1
Triangle	−27	1.28	0.5
Hamming	−43	1.30	0.54
Hanning (α = 1)	−23	1.64	0.64
Gaussian (α = 2.5)	−42	1.33	0.51
Kaiser Bessel (α = 2.0)	−46	1.43	0.49
Tukey (α = 0.25)	−14	1.01	0.88
Trapezoid	−14	1.06	0.80

[0089] The analysis of this table shows that, in practice, apart from the Tukey windows, the energy loss is in the range of 3 dB.

[0090] Most of these windows may be likened to the Hamming window by modifying the parameters of the equation.

[0091] The aim is to minimize the Gibbs phenomena for all the derived windows of Hanning and optimize the time frequency product for the Gaussian windows.

[0092] FIGS. 11 and 12 give a schematic view of the correction made in the value of QDM by applying the method according to the invention.

[0093] FIG. 11 shows the value of QDM obtained for a frequency value of 30.3 Hz using a trapezoidal window as described in FIG. 7 to carry out the Fourier transform in order to reduce the beat.

[0094] FIG. 12 shows the value of the QDM (VI) obtained for a frequency of 30.3 Hz by using a frequency estimator and a trapezoidal window for the Fourier transform.

Claims

1. A method to correct the drift in the phase shift (ΔΦ,θ) between two signals (S1) and (S2), said signals being generated by a signal having a reference frequency Fe or a corresponding period Ω, the drift in the phase shift resulting from a drift in frequency Fe or in period, according to a given relationship R((F,Ω), (θ, ΔΦ)), the method comprising at least the following steps: determining the value of the phase shift (ΔΦ, θ) from the phase Φ1 of the signal S1 and the phase Φ2 of the signal S2, estimating Fest the value of frequency Fe, or the corresponding period Ω, bringing this value (ΔΦ, θ) to a value within an interval of given values [QDMmin, QDMmax] containing the value QDM(Fe) obtained for the reference value Fe in taking account of the values ΔΦ and Fest and of the given relationship R((F,Ω),(θ, ΔΦ)).

2. A step of processing by Fourier transform according to claim 1, using a sampling window whose shape is substantially trapezoidal.

3. A method according to one of the claims 1 or 2, wherein the value of the reference frequency Fe is substantially equal to 30 Hz, or the value of the corresponding period is substantially equal to 2π*30 Hz, the signal S1 is an LF signal filtered by a linear filter before it is sampled by Fourier transform, and the signal S1 is at least partly used to estimate the frequency Fe or the period Ω, the signal S2 is a HF signal filtered by a non-linear filter before it is sampled by Fourier transform.

4. A method according to claim 3, wherein the trapezoidal shape of the window is chosen to obtain, at a frequency of 30 Hz or a corresponding period, an attenuation on 2.5 times the value of the period of the signal to be sampled S1 or S2.

5. A method according to one of the claims 1 to 4, wherein the relationship R((F, Ω), (θ, ΔΦ)) is a linear function.

6. A method according to one of the claims 1 to 5, wherein the interval of tolerable values for the step c) is equal to [QDM(Fe)−0,4; QDM(Fe)+0,4°] for a frequency value substantially equal to 30 Hz or a corresponding period substantially equal to 2π*30 Hz and a variation of +/−1,5%.

7. A method to correct a drift in the phase shift (ΔΦ, θ) between two signals, S1 having a phase Φ1 and S2 having a phase Φ2, the two signals being generated by a signal with a frequency Fe or with the corresponding period, the drift in the phase shift resulting from the drift in Fe or the drift in the period according to a given relationship R((F,Ω), (θ, ΔΦ)), the method comprising at least one step to determine at least one of the phases (Φ1, Φ2) by Fourier transform using a substantially trapezoidal sampling window.

8. A device for the correction of the drift in the phase shift (ΔΦ,θ) between two signals (S1) and (S2), said signals being generated by a signal with a reference frequency Fe, the drift being linked to the drift in frequency Fe according to a given relationship R((F,Ω), (θ, ΔΦ)), the device comprising at least: a separator of the signals S1 and S2, means to determine the value of the phase Φ1 and Φ2 for each of the signals and a means to determine the value of phase shift ΔΦ, an estimator of the frequency Fe or of the corresponding period positioned after the separator, a device comprising, in memory, the relationship R((F,Ω), (θ, ΔΦ)) receiving the values ΔΦ and the estimated frequency or the estimated corresponding period, adapted to bringing the value of ΔΦ to a value within an interval of given values [QDMmin, QDMmax] containing the value QDM(Fe) obtained for the reference value Fe (or period).

9. A device according to claim 8, comprising means for processing the signals (S1) and/or (S2) by Fourier transform, the sampling window having a substantially trapezoidal shape.

10. A device according to claim 8 or 9, wherein the device to estimate the frequency value is of the Prony's estimator type.

11. A VOR receiver comprising a device for the correction of a drift in the phase shift (ΔΦ, θ) between two signals (S1) and (S2), said signals being generated by a signal having a reference frequency Fe, the drift being linked to the drift in frequency Fe or to the drift in period according to a given relationship R((F,Ω), (θ, ΔΦ)), comprising at least: a separator of the signals S1 and S2, means to determine the value of the phase Φ1 and Φ2 for each of the signals and a means to determine the value of phase shift ΔΦ, an estimator of the frequency Fe or of the corresponding period positioned after the separator, a device comprising, in memory, the relationship R((F,Ω), (θ, ΔΦ)) receiving the values ΔΦ and the estimated frequency or the estimated corresponding period, and adapted to bringing the value of ΔΦ to a value within an interval of given values [QDMmin, QDMmax] containing the value QDM(Fe) obtained for the reference value Fe (or period).

12. Application of the method according to one of the claims 1 to 7 or of the device according to one of the claims 8 to 11 to reducing the variation in the QDM of a navigation system or radio navigation system.