WO2021099071A1

WO2021099071A1 - Method for processing a signal implemented by a receiving device of a first transmitting site with a view to removing an interfering signal, corresponding receiving device and corresponding computer program.

Info

Publication number: WO2021099071A1
Application number: PCT/EP2020/080101
Authority: WO
Inventors: Olivier Rousset; Cyril Bernier
Original assignee: Enensys Teamcast
Priority date: 2019-11-18
Filing date: 2020-10-27
Publication date: 2021-05-27
Also published as: FR3103335A1; FR3103335B1

Abstract

The invention relates to a method for processing a signal, which is implemented by a first receiving device of a first transmitting site, comprising: receiving (221) a combined signal comprising: - an interfering signal, corresponding to a first signal emitted by a first emitting device of said first transmitting site, via a transponder, and - at least one desired signal, corresponding to at least one second signal emitted by a second emitting device of said at least one second transmitting site, via said transponder; aligning (222) said first signal with said combined signal; and subtracting (222) said first signal aligned with said combined signal. According to the invention, the method comprises a prior calibrating step comprising: - receiving (211) a reference signal emitted by said second emitting device, via said transponder, - measuring (212) the phase noise associated with the transmission of said reference signal, said alignment (222) of said first signal and said combined signal taking into account said phase noise.

Description

DESCRIPTION

TITLE: Process for processing a signal implemented by a reception device of a first broadcasting site to remove an interfering signal, corresponding reception device and computer program.

1. Field of the invention

The field of the invention is that of digital communications.

More precisely, the invention relates to transmissions between several broadcasting sites, via a transponder.

Even more precisely, the invention proposes a solution making it possible to cancel, or at the very least to reduce, an interfering signal received at one of the broadcasting sites, corresponding to a signal transmitted by a transmitter of the broadcasting site, received and retransmitted by the transponder.

The invention finds applications in any transmission system implementing a transponder, for example a satellite, in particular in broadcasting networks according to the DVB-S, DVB-S2, or DVB-S2X standard (in English "Digital Video Broadcasting - Satellite ”, in French“ digital television broadcasting - satellite ”), or other existing or future standards.

2. Prior art

In order to optimize the spectral efficiency for transmissions between different broadcasting sites, the “carrier-in-carrier” ^® technique has been developed.

As illustrated in FIG. 1, such a technique allows a pair of modems (modulator-demodulator), communicating via a transponder, to transmit on the same transmission frequency F1 and to receive on the same frequency of reception F2. Such an embodiment is also called a “point-to-point” mode.

According to the example of FIG. 1, a first transmission device is considered, comprising a modulator, and a first reception device, comprising a demodulator, located on a first broadcasting site 11. A second transmission device is also considered. transmission, comprising a modulator, and a second reception device, comprising a demodulator, located on a second broadcasting site 12. The transmission and reception devices of the first broadcasting site, respectively of the second broadcasting site, can be two. separate equipment, or form a single transmission-reception equipment. The transmission and / or reception devices of each site broadcast communicate via a transponder 13, for example a satellite.

Such a transponder 13 can therefore receive on the one hand a first signal SI transmitted by the first transmission device at the frequency F1, and on the other hand a second signal S2 transmitted by the second transmission device at the frequency F1.

The transponder 13 can amplify the combination of the signals received, and retransmit this combination SI + S2 to the reception devices of each of the broadcasting sites 11, 12 at the frequency F2. This signal is called a combined signal hereinafter.

In order to be able to demodulate the second signal transmitted by the transmitting device of the second broadcasting site 12, the receiving device of the first broadcasting site 11 must remove the component of the combined signal received corresponding to the first signal transmitted by the transmitting device. of the first broadcast site 11. Symmetrically, in order to be able to demodulate the first signal transmitted by the transmission device of the first broadcast site 11, the reception device of the second broadcast site 12 must remove the component of the combined signal received corresponding to the second signal transmitted by the transmission device of the second broadcasting site 12.

To do this, each broadcast site 11, 12 implements an echo cancellation technique (or “anti-echo”, in English “echo canceller”).

FIG. 1B illustrates an example of an echo cancellation technique implemented at the level of the first broadcast site 11.

The first transmission device comprises a modulator 111 which generates the first baseband signal SI. This first signal is transposed to the frequency F1 (“up-converter” 112) and transmitted to the transponder.

The first reception device receives the combined signal on the frequency F2 coming from the transponder and transposes it into baseband (“down-converter” 114). The combined signal obtained after passing through the radiofrequency channel is denoted by S1 '+ S2'.

An estimate 113 of the offset provided to the first signal SI by the radiofrequency channel is also determined, so as to obtain an estimate of the first received signal SI ′.

The term shift designates here, and throughout the rest of the document, the deformations in time, frequency, and / or amplitude brought to the first signal SI by the radiofrequency channel, ie by all the elements of the radiofrequency chain ("up- converter "112, up channel, transponder 13, down channel and" down converter " The estimate of the first received signal S1 ′ is subtracted 115 from the combined signal S1 ′ + S2 ′, so as to cancel the echo and obtain an estimate of the second signal S2 ′ received by the first reception device.

The second signal S2 ′ can then be demodulated by the demodulator 116 of the first reception device.

However, for echo cancellation to be effective, it is necessary for the estimate of the offset provided to the first signal SI by the radiofrequency channel to be precise.

In particular, in order to correctly estimate the offset brought to the first signal SI by the radiofrequency channel, it is important to know the static and dynamic deformations brought by the radiofrequency channel. It is therefore necessary to correctly estimate: the phase and amplitude error and the associated variations, the frequency offset error and the associated variations, the time shift and the associated variations.

However, current techniques do not make it possible to fully estimate the offset provided to the first signal SI by the radiofrequency channel.

There is therefore a need for a new echo cancellation technique which is simple to implement and efficient.

3. Disclosure of the invention

The invention proposes a solution allowing at least two modems to communicate via a transponder, in the form of a method for processing a signal, implemented by a reception device of a first site. broadcast, said first reception device, comprising: receiving a combined signal comprising: an interfering signal, corresponding to a first signal transmitted by a transmission device of the first broadcasting site, said first transmission device, to destination of at least one reception device of at least one second broadcasting site, called second reception device, via a transponder, and at least one desired signal, corresponding to at least one second signal transmitted by a transmission device from said at least one second broadcast site, called second transmission device, to the first reception device, via the transponder, aligning the first signal with the combined signal, delivering a first aligned signal, subtracting the first aligned signal from the combined signal, delivering an estimate of said at least one desired signal.

According to the invention, such a method comprises a preliminary calibration step comprising: the reception of a reference signal transmitted by the second transmission device, via the transponder, the measurement of the phase noise associated with the transmission of the signal from reference.

Furthermore, according to the invention, the alignment of the first signal on the combined signal takes account of the phase noise.

Thus, the proposed solution makes it possible to improve the estimation of the offset provided by the radiofrequency channel to one of the signals transmitted by taking account of the phase noise measured in the transmission system. The "alignment" aims to correct this shift, i.e. the distortions in time, frequency, and / or amplitude brought by the radiofrequency channel.

In this way, it is possible to estimate more precisely the “interfering” component of the received combined signal, and to eliminate it, or at least reduce it, before demodulation.

Indeed, the inventors have demonstrated that phase noise has an impact on the performance of echo cancellation.

It is recalled that phase noise is a short-term instability, mainly coming from the oscillators of a radiofrequency chain, and corresponds to rapid variations (> 1 Hz) in frequency. Phase noise is actually a slight frequency modulation of the carrier.

As the phase noise varies little over time, it can be considered that the initial measurement of the phase noise, implemented for example during the installation of the transmission / reception devices, is representative of the phase noise of the chain. radio frequency between the first broadcast site and the second broadcast site, via the transponder. In fact, the phase noise is mainly linked to the quality of the reception head (LNB converter or “Low Noise Block”) of the reception device and changes little during operation.

According to a particular embodiment, the proposed solution allows at least two modulator-demodulators to transmit on the same frequency F1 and to receive on the same frequency F2, which allows a gain in terms of spectral efficiency. Furthermore, we note that it can be difficult to extract the phase noise from the received signal, knowing that other physical phenomena add degradations, such as Gaussian noise, intermodulations linked to non-linearities of the radiofrequency chain, etc.

Thus, according to a particular embodiment, the reference signal is a sinusoid. In this way, when the first transmitting device transmits a reference signal of sinusoidal type (“sinus” mode, or CW, for “continuous wave”), the second receiving device can make a measurement of the phase noise of any phase. the radiofrequency chain, and vice versa.

One advantage of this mode is that it allows the reference signal to stand out from the Gaussian reception noise. Moreover, this mode is not sensitive to the non-linearities of the system. The invention therefore proposes a clever solution for measuring phase noise. In another embodiment, the invention relates to a reception device of a corresponding reception site.

Such a device for receiving a signal from a first broadcasting site, called a first receiving device, comprises: receiving means configured to receive a combined signal comprising: an interfering signal, corresponding to a first signal transmitted by a device transmission from the first broadcast site, said first transmission device, to at least one reception device from at least one second broadcast site, called second reception device, via a transponder, and at least one desired signal, corresponding to at least a second signal transmitted by a transmitting device from said at least one second broadcasting site, said second transmitting device, to the first receiving device, via said transponder, alignment means configured to align the first signal with the combined signal, providing a first aligned signal, subtracting means configured to subtract the first aligned signal from the combined signal, d providing an estimate of said at least one desired signal.

According to the invention, the reception means being configured to first receive a reference signal transmitted by the second transmission device, via the transponder, the first reception device also comprises phase noise measuring means configured to measure the phase noise associated with the transmission of the reference signal, and the alignment means are configured to align the first signal with the combined signal taking into account the phase noise.

Such a reception device is in particular suitable for implementing the method for processing a signal described above. It can in particular be integrated into a transmission device of the broadcasting site. This reception device may of course include the various characteristics relating to the method for processing a signal according to the invention, which may be combined or taken in isolation.

In particular, such a device comprises at least one processor, operatively coupled to a memory, configured to implement the various steps of the method for processing a signal according to the invention.

The invention also relates to one or more computer programs comprising instructions for implementing a method for processing a signal as described above when this or these programs are executed by at least one processor.

The invention also relates to an information medium readable by a computer, and comprising instructions of a computer program as mentioned above.

4. List of figures

Other characteristics and advantages of the invention will emerge more clearly on reading the following description of a particular embodiment, given by way of simple illustrative and non-limiting example, and the appended drawings, among which:

[Fig IA]

[Fig IB] Figures IA and IB illustrate the general principle of an echo cancellation technique in a point-to-point mode;

[Fig 2] FIG. 2 illustrates the main steps implemented by a device for receiving a first broadcast site according to one embodiment of the invention;

[Fig B] Figure 3 shows the general principle of the alignment step of Figure 2;

[Fig 4] Fig. 4 shows a block diagram of the frequency correction according to one embodiment of the invention;

[Fig 5] Figure 5 shows a block diagram of the correction in time, phase and / or power according to one embodiment of the invention;

[Fig 6] FIG. 6 illustrates the mask of the phase noise according to an embodiment of the invention;

[FIG. 7] FIG. 7 illustrates an exemplary implementation of a reception device on a programmable circuit of the FPGA type; [Fig 8] FIG. 8 illustrates an example of architecture for the FIFO of FIG. 7;

[Fig 9] FIG. 9 illustrates an example of architecture for the NCO of FIG. 7;

[Fig 10] FIG. 10 illustrates an example of architecture for the correlator of FIG. 7;

[Fig 11] FIG. 11 illustrates an exemplary architecture for the frequency correction module of FIG. 7;

[Fig 12] FIG. 12 illustrates an exemplary architecture for the time, phase and / or amplitude correction module of FIG. 7;

[Fig 13] Figure 13 shows the simplified structure of a receiving device according to a particular embodiment;

[Fig 14] FIG. 14 illustrates the implementation of the invention in a point-to-multipoint mode.

5. Description of an embodiment of the invention

5.1 General principle

The invention is placed in the context of communications between at least two broadcasting sites, via a transponder.

The general principle of the invention is based on taking into account the phase noise associated with the radiofrequency channel (“upconverter”, up channel, transponder, down channel, “downconverter”), to optimize the performance of a cancellation technique. echo implemented at at least one of the broadcasting sites.

FIG. 2 illustrates the main steps implemented by a device for receiving a first broadcasting site, called the first reception device, in a communication system comprising two broadcasting sites, such as the one illustrated in FIG. -medium rare). It is noted that, symmetrically, the same steps can be implemented by a reception device of the second broadcasting site. It can also be noted that the proposed solution can be applied to point-to-multipoint mode, as described below.

As illustrated in FIG. 2, the first reception device first of all implements a calibration step 21, comprising:

the reception 211 of a reference signal transmitted by a transmission device belonging to the second broadcasting site, called the second transmission device, via the transponder,

- measurement 212 of the phase noise associated with the transmission of the reference signal. This calibration step can be implemented once during the installation of the transmission / reception devices, or it can be repeated. It can be implemented periodically or on an ad hoc basis, for example following a change in the propagation channel between the two broadcasting sites, via the transponder.

Once the phase noise associated with the transmission of the reference signal is measured, the first receiving device can process useful signals.

Thus, the first reception device implements:

the reception 221 of a combined signal comprising: an interfering signal, corresponding to a first useful signal transmitted by a transmission device of the first broadcasting site, called the first transmission device, intended for the reception device of the second site broadcast, said second receiving device, via a transponder, and a desired signal, corresponding to a second signal transmitted by the second transmitting device, to the first receiving device, via the transponder,

- alignment 222 of the first signal on the combined signal, delivering a first aligned signal, taking into account the measured phase noise,

the subtraction 223 of the first signal aligned to the combined signal, delivering an estimate of the desired signal.

5.2 Example of implementation

The general principle of the alignment step 222 of FIG. 2 is described below with reference to FIG. 3.

As previously indicated, the first receiving device receives (221) a combined signal Sc comprising an interfering signal and the desired signal. The first reception device also knows the first useful signal SI transmitted by the first transmission device.

The first receiving device then seeks to align (222) the first signal SI on the combined signal Sc. The first aligned signal SI 'can then be subtracted from the combined signal Sc, so as to obtain an estimate of the desired signal S2'.

According to at least one embodiment, the alignment 222 implements a first synchronization step 31, called coarse synchronization.

In particular, such a coarse synchronization step implements: a correlation between the combined signal Sc and the first signal SI shifted in time and / or in frequency, for at least two shift values in time and / or in frequency, a determination of a correlation peak, an estimate of a coarse time and / or frequency offset between the combined signal and the first signal.

For example, for each time and / or frequency shift, the value of the amplitude of the correlation between the combined signal Sc and the first signal SI shifted in time and / or frequency is stored in a table. Then the table of amplitude values as a function of time and / or frequency is analyzed in order to detect the maximum value, corresponding to a correlation peak between the combined signal Sc and the first signal SI. The time and / or frequency values associated with this correlation peak make it possible to determine a coarse time and / or frequency shift between the combined signal and the first signal.

The correlation is expressed for example in the following form:

with L corresponding to the length of the correlation.

By way of example, the following parameters are chosen for coarse synchronization 31: the length of the correlation is chosen to be greater than 1000 samples, for example equal to 10,000 in order to have better precision, the increment step of the frequency shift is equal to 10kHz, the frequency offset test range is equal to +/- 1MHz, the time offset increment step is equal to 10 ns (depending on the transmission speed, in English "baudrate", here for a baudrate> 50MHz), the time offset test range is equal to 5 ms.

This coarse synchronization step 31 can be implemented once during start-up, or else repeated, for example following the detection of a loss of synchronization.

Optionally, several iterations of the coarse synchronization algorithm in time and / or in frequency can be implemented, by refining the search range of the offset in time and / or in frequency in order to have an increment step of the offset. in frequency less than 10 kHz, for example of the order of 100 Hz. It is noted that, according to a particular embodiment, the first signal carrying a header not obtained from a scrambling sequence (non "scrambled" header), the correlation step is deactivated during reception of the header. of the first signal.

According to another particular embodiment, the first signal is constructed from a scrambling sequence distinct from that used to construct the second signal.

These two variants, which can be implemented independently or jointly, make it possible to limit the risk of false detections, in particular when the signals S1 and S2 transmitted by each of the transmission devices resemble each other / are correlated.

For example, if we look at the context of the DVB-S2 standard, such a header is of the “PLHEADER” type and the scrambling sequence of the “PL Scrambler” type.

Once the coarse time and / or frequency offset between the combined signal and the first signal has been determined, fine timing algorithms based on frequency servo using a digital phase locked loop (DPLL) and / or a recursive adaptive filter (FIR) can be implemented to follow the variations of the combined signal and to maintain the alignment of the first signal with the combined signal.

Thus, according to at least one embodiment of the invention, the alignment 222 also implements: a frequency correction of the first signal, implementing a phase locked loop 32, a correction in time, phase and / or gain (ie amplitude) of the first frequency corrected signal, implementing adaptive filtering 33.

Figure 4 shows a block diagram of the frequency correction, used to fine-tune the frequency synchronization, and to track and correct the frequency shifts between the first signal and the combined signal.

According to the example illustrated in FIG. 4, the frequency correction implements at least one iteration of the following steps: frequency shift 41 of the first signal SI according to a set point D /, conjugate complex multiplication 42 of the combined signal Sc with the first signal SI shifted in frequency according to the setpoint D /, low-pass filtering 43 of the result of the multiplication, estimation 44 of the phase shift between the combined signal Sc and the first signal shifted in frequency from the signal obtained at the output of the low-pass filtering, error correction 45 delivering an instruction D / update. In particular, the frequency shift step is implemented by an NCO oscillator (in English "numerically-controlled oscillator", in French "oscillator controlled by digital input").

For the first iteration (initialization), the setpoint D / corresponds to the frequency offset obtained at the output of the coarse synchronization step 31. For the following iterations, the setpoint D / is that obtained at the output of the error correction 45 .

The low-pass filtering 43 uses, for example, a low-pass filter with an adjustable cut-off frequency. In particular, such a low-pass filter makes it possible to filter the phase noise by allowing frequencies below a cut-off frequency determined during the calibration step to pass. As explained below, such a cutoff frequency is determined from the phase noise of the radiofrequency chain.

The complex conjugate multiplication 42, associated with the low-pass filtering 43, makes it possible in particular to carry out a correlation between the combined signal and the first transmitted signal delayed, then corrected in frequency.

According to the example illustrated in FIG. 4, the estimation 44 of the phase shift between the combined signal Sc and the first signal shifted in frequency implements an arctangent calculation, and the error correction 45 implements a type corrector "Proportional, Integral, Derivative" (PID) in order to update the setpoint Dί. According to a particular embodiment, the coefficients of the PID corrector also depend on the cutoff frequency determined during the calibration step.

By way of example, seeking to obtain a residual error of the order of 1 Hz when the frequency servo algorithm has converged, the following parameters are chosen for the phase locked loop 32: proportional gain of the PID: Kp = le ^-2 , integral gain of the PID: Ki = 1, gain derived from the PID: Kd = 300.

In particular, the inventors have observed that the residual noise of the phase locked loop degrades the performance of the adaptive filter. It is therefore sought, according to at least one embodiment, to optimize the coefficients of the phase-locked loop in order to reduce this level of residual noise.

After the frequency correction, or simultaneously with the frequency correction, a correction in time, phase and / or gain (amplitude) of the first signal is implemented. FIG. 5 presents a block diagram of the correction in time, phase and / or power of the first signal, implementing an adaptive filtering 33. Such adaptive filtering makes it possible to closely follow the variations in time, in phase and / or in power.

For example, such an adaptive filtering 33 implements an algorithm belonging to the group comprising: an algorithm of LMS type (in English “Least Mean Square Error”, in French algorithm for minimizing the mean square error), an algorithm of type NLMS (in English "Normalized Least Mean Square Error", in French algorithm for minimizing the normalized mean square error "), an RLS type algorithm (in English" Recursive Least Square ", in French recursive least squares algorithm).

According to the example illustrated in FIG. 5, the correction in time, phase and / or gain implements an algorithm of the minimization of the mean square error type, implementing at least one iteration of the following steps: adaptive filtering of the first signal frequency corrected, implementing an adaptive filter 51 with finite impulse response, delivering said first aligned signal, updating 52 of the coefficients of the adaptive filter, the updating of the coefficients being implemented after the step of subtraction 223 of the first signal aligned with the combined signal, delivering an estimate of the desired signal (E = S2 '), and multiplication of the estimate of the desired signal by a convergence parameter m.

Noting:

51 the first signal emitted by the first transmission device (interfering signal to be deleted by the first reception device),

52 the signal transmitted by the second transmission device (desired signal to be demodulated by the first reception device,

Sc the combined signal received by the first reception device, carrying the first signal SI and the second signal S2 distorted by the propagation channel (including in particular the transponder),

Y the signal at the output of the adaptive filter 51;

E the error signal, corresponding to the desired signal S2 ', we have:

Sc = (51 * H1 + S2 * H 2) + b, with b a Gaussian white noise and H corresponding to the temporal response of the propagation channel, Y = W * SI, with W corresponding to the coefficients of the adaptive filter 51,

E = Sc - Y

If we consider an algorithm of the LMS or NLMS type, the coefficients of the adaptive filter 51 can be expressed in the following form:

W [n + 1] = W [n] + m E [n] .Sl [n], with m corresponding to the adaptation step, also called the convergence parameter.

If we consider an RLS type algorithm, the coefficients of the adaptive filter 51 can be expressed in the following form:

W [n + 1] = W [n] + K [n] .E [n]

l corresponding to the adaptation step.

By way of example, the following values can be chosen for the adaptation step: for the LMS algorithm: m = 0.000003; for the NLMS algorithm: m = 0.0002; for the RLS algorithm, l = 1 - 0.000001.

In particular, an RLS type algorithm can be favored for a low length of the adaptive filter, for example less than 100 coefficients.

According to a particular embodiment, the convergence parameter m is determined from the residual phase noise obtained during the calibration step, as described below.

5.3 Calibration

The following describes, in relation to FIG. 6, the measurement of the phase noise implemented during the calibration step 21, and its use. Indeed, as indicated above, the calibration step 21 makes it possible to determine a cutoff frequency for the low-pass filtering 43 and for the error correction 45, used for the frequency control, as well as a noise of residual phase used to determine the convergence parameter m of the adaptive filter.

A) Determination of phase noise

For example, it is considered that the reference signal S2 _ref emitted by the second transmission device during the calibration step is a sinusoid. In this case, the phase noise f (ΐ) can be expressed as follows:

S2 _re f (t) = VOcos (2nF2t + 0 (t)) with:

VO the amplitude of the sinusoid,

F2 the frequency of the sine wave

4> (t) the instantaneous variation of the phase, also called phase noise

The spectrum of such a signal must therefore show a line at the frequency F2, of amplitude VO. The dispersion around this line represents the mask of the phase noise.

On reception of the reference signal, the first reception device can determine a phase noise mask 61, centered on the carrier frequency used by the downlink transponder. It is recalled in fact that, as the reference signal passes through the transponder, it is transmitted on a frequency F1 by the second transmission device, but received on a frequency F2 by the first reception device.

For example, the determination of a phase noise mask implements the measurement of the power of the reference signal received on at least one frequency belonging to a narrow frequency band (ie of the order of a few kilohertz) centered on the carrier frequency used by the downlink transponder. In other words, different measurements are made of the power of the reference signal received by the first reception device (611, 612, 613, etc.) for different frequencies close to the frequency F2.

It can be seen in FIG. 6 that the template 61 makes it possible to break down the narrow frequency band into two parts: a first part 62 corresponding to the frequency band located around the central frequency F2, corresponding to a greater power of the received signal. These are fairly slow variations that can be compensated for; a second part 63, remote from the central frequency F2, corresponding to the residual phase noise. This second part corresponds to the phase noise that an adaptive filter, used to compensate for the variations in time, phase and / or amplitude of the received signal, can compensate. These instantaneous variations have an impact on the reactivity of the recursive algorithm.

We can then estimate a cutoff frequency Fc associated with the first part 62 of the template 61 which can be applied a low-pass filter to filter the phase noise and to optimize the performance of the system. Indeed, this first part corresponds to the noise phase that the frequency servo algorithm, used to compensate for the variations in frequency of the received signal, can compensate.

For example, the cutoff frequency is estimated from the standard deviation associated with the measurements of the power of the reference signal received on said narrow band. Indeed, even if the mask of the phase noise does not correspond exactly to a Gaussian distribution, the inventors have shown that a calculation of the standard deviation makes it possible to obtain a correct estimate of the cutoff frequency Fc.

The cut-off frequency thus determined, for example of the order of a few hundred hertz, can then be used by the frequency servo algorithm making it possible to compensate for the frequency variations of the received signal.

It is also possible to estimate the residual phase noise associated with the second part 63 of the template 61, associated with the reference signal received outside the frequency band defined by the carrier frequency used by said downlink transponder plus or minus the frequency of cut-off (+/- Fc and +/- 1MHz of the central frequency F2).

For example, the residual phase noise measurement (residual_PN, in dB) is carried out by measuring the power of the carrier at the frequency F2, denoted Power_CW, and by comparing it with the power of the signal in the narrow band around the frequency F2, [F2 - Fc - 1 MHz; F2 - Fc] U [F2 + Fc \ F2 + Fc + 1 MHz], noted Power_residual:

Residual_PN = 10 * logl0 (Power_CW / Power_Residual)

The residual phase noise thus determined, for example of the order of -30dB to -50dB, can then be used to adjust the parameters of the adaptive filter used to compensate for the variations in time, phase, and / or amplitude of the received signal.

B) Use of phase noise

Once the cutoff frequency and the residual phase noise have been determined, it is possible to use these parameters to adjust the frequency servo implemented by the phase locked loop 32 and to adjust the coefficients of the adaptive filter 51 described. previously.

More precisely, the inventors have demonstrated that the algorithm which makes it possible to carry out the frequency slaving must be more or less fast as a function of the phase noise.

Thus, according to at least one embodiment, the proposed algorithm is capable of following the slow variations of the phase noise and of filtering the instantaneous variations in order to optimize the frequency control. To do this, a cutoff frequency associated with the first part 62 of the template 61 is determined, as explained above, and the low-pass filter 43 (ie phase filter) of the phase-locked loop 222 is adjusted with this cutoff frequency value.

Recall that the phase filter is an exponential weighted average function expressed in the form:

where: a is a filter parameter (a <1) for example a = 1. e ^{~ 6} ;

X _n is the complex conjugate product of the signal combined with the transmitted signal delayed and then corrected in frequency.

For a continuous system, the Laplace transform of such a transfer function is written:

By expressing the filtering parameter in the form a = ka ₀ , where a ₀ is the maximum filtering constant ("1) and k is an adjustment parameter in the interval [1; l / a ₀ [, ^we can show that, knowing the cutoff frequency of the phase filter, the adjustment parameter k can be written: wk = - F - a ₀ (V2 - 1 + w) with:

F _c w ⁼ T ^r ec ~ h where Fc is the cutoff frequency of the phase filter and F _ech is the sampling frequency.

Thus, the determination of the cut-off frequency Fc from the phase noise makes it possible to determine the adjustment parameter k itself used to determine the filtering parameter of the low-pass filter 43.

Also to optimize the frequency control, the coefficients of the PID error corrector 45 can be determined as a function of the cutoff frequency.

It is considered by way of example that the coefficients Kp, Ki and Kd are initialized to values Kp ₀ , Ki ₀ and Kd ₀ , adjusted for a nominal cut-off frequency of 300 Hz.

To keep the setting of the PID error corrector regardless of the cutoff frequency of the phase filter, the coefficients Kp, Ki and Kd are modified according to a band factor b such as: Kr = bKr ₀

Kί = bKί ₀ Kd = bKά o

„300 with: b = -

^ FC

The inventors have also demonstrated that the recursive algorithm based on an adaptive filter can be optimized by taking into account the phase noise.

Indeed, as indicated previously, if we consider an algorithm of the LMS or NLMS type, the coefficients of the adaptive filter 51 can be expressed in the following form:

W [n + 1] = W [n] + m E [n]. SI [n], with m corresponding to the adaptation step.

The larger this adaptation step, the faster the algorithm is to converge, but the greater the residual noise. In order to obtain better performance of the adaptive filtering algorithm, the phase noise value can enter into the calculation of the convergence parameter m according to the following formula:

with M a constant allowing the measurement of the phase noise to be weighted on the convergence parameter m. For example, the constant M is of the order of 0.8.

It is also possible to express the parameter l as a function of the phase noise if we consider an RLS type algorithm.

5.4 Example of implementation

An example of the implementation of a reception device according to at least one embodiment of the invention, according to which the echo cancellation technique is implemented on a FPGA type programmable circuit (in English “Field Programmable Gâte Arrays”).

For the sake of simplification of the description, we always place ourselves in the context of a communication system comprising two broadcasting sites, such as that illustrated in FIG. 1A, and more precisely at the level of the reception device of the first broadcasting site.

The complex samples corresponding to the received combined signal Sc (coming from the analog digital converter 71) and the complex samples corresponding to the first signal SI coming from the modulator are first of all interpolated (decimation filters 721, 722) at twice the frequency of sampling (2.F _ech ). For example, IQ_COMBINE denotes the interpolated complex samples corresponding to the received combined signal Sc and IQJNTF the complex samples corresponding to the first signal SI.

An attempt is then made to remove the interfering signal (SI received) from the combined signal Sc. The suppression of the interfering signal of the combined signal Sc implements, according to this example, two processing chains: a chain 73 for processing the data stream, at the frequency 2.F _ech , operating in real time, a chain 74 for seeking corrections in frequency, time, phase and / or amplitude, which can operate by data block.

According to the example illustrated in FIG. 7, the correction search chain 74 comprises a controller 741 making it possible to manage the chaining of the coarse synchronization, frequency slaving and adaptive filtering algorithms, and to check their convergences.

At the end of the elimination of the interfering signal from the combined signal Sc, an estimate of the second signal S'2 is obtained, at the frequency 2. F _ech . The signal S'2 obtained can be interpolated (interpolation filter 75) at the frequency F _ech , before being converted into an analog signal in a digital-to-analog conversion module 76 and transmitted to the demodulator 77 of the first reception device. According to this implementation, the demodulation function is therefore not implemented in the FPGA. The second estimated signal S2 ′ is thus reconverted into a radiofrequency signal before entering the demodulator.

In particular, an external memory 78, for example of the DDR (“Double Data Rate”) type, can be added. A DDR driver 79 as well as FIFOs internal to the FPGA at the level of writing and reading can be associated with this memory.

The two processing chains 73 and 74 are described in more detail below.

The data stream processing chain 73 makes it possible to determine the offset to be given to the first signal transmitted by the transmission device of the first broadcasting site in order to be able to align it with the combined signal and then subtract it from the combined signal.

According to the example illustrated in FIG. 7, the first signal SI is processed by different modules: a FIFO 731, making it possible to delay the complex samples of the signal SI, for example up to 280 ms. For example, we denote by IQ_delay the complex samples at the output of FIFO 731; an NCO 732, making it possible to provide a frequency correction to complex delayed samples. For example, we denote by IQ_NCO the complex samples at the output of the NCO 732; a complex filter 733, making it possible to provide a correction in phase and / or in amplitude at delayed and frequency corrected complex samples. For example, we denote by IQ_FIR the complex samples at the output of filter 733; a subtracter 734 making it possible to subtract the first aligned signal (IQ_FIR samples) from the combined signal (IQ_COMBINE samples). For example, we denote by IQ_DES the complex samples at the output of the subtracter 734.

FIG. 8 illustrates an example of architecture for FIFO 731.

Such a FIFO 731 comprises, for example, means making it possible to shift the complex samples by an integer value, as well as means making it possible to shift the complex samples by a fractional value, in order to precisely follow the variations in time.

For example, the means making it possible to shift the complex samples by an integer value use a memory 7311 (FIFO IN, FIFO OUT). The means for shifting the complex samples of a fractional value implement an interpolation filter, for example of the 5 ^th Lagrange minimum order.

Note that the integer shift implemented by the memory 7311 and the fractional shift implemented by the Lagrange filter 7312 must be synchronized.

Figure 9 illustrates an example architecture for the NCO 732.

Such an NCO 732 comprises for example a phase accumulator 91, a phase to sine / cosine conversion 92, then a complex multiplication 93 for the frequency offset.

For example, the phase to sine / cosine conversion 92 uses a memory (LUT table) or a Cordic (in English “COordinate Rotation Digital Computer”, in French “digital calculation by rotation of coordinates”).

The complex filter 733 is for example a finite impulse response filter of at least 15 coefficients, making it possible to correct the amplitude, the phase and / or the residual error in time. The coefficients can be updated almost instantaneously in order to follow the rapid variations of the combined signal.

The subtracter 734 allows for its subtraction of complex samples. It does not use the resources of the FPGA in terms of DSP (in English “Digital Signal Processing”) and of BRAM (in English “Block Random Access Memory”).

The chain 74 for searching for frequency, time, phase and / or amplitude corrections makes it possible to estimate the shifts to be applied to the signal SI in order to align it with the combined signal Sc. According to the example illustrated in FIG. 7, the processing chain 74 implements three different modules: a correlator 741, making it possible to estimate a coarse offset in time and / or in frequency; a frequency correction module 742, making it possible to follow the variations in frequency; a time, phase and / or amplitude correction module 74S, making it possible to adjust the coefficients of the filter 733 in order to rapidly follow the variations in time, phase and / or amplitude.

The algorithms implemented by these different modules can be managed by a controller 744, for example implementing a state machine making it possible to clock the sequences and to check the alignment and convergence of the algorithms.

FIG. 10 illustrates an example of architecture for the correlator 741.

Such a correlator 741 takes as input the complex samples of the combined signal (IQ_COMBINE), as well as the complex samples of the first signal SI after passing through the FIFO 7S1 and the NCO 732 (IQJMCO).

The objective is to synchronize these two signals in time and / or in frequency, in order subsequently to be able to follow the variations linked in particular to the movement of the satellite type transponder.

The correlator 741 implements a first step of coarse synchronization in time and / or in frequency (reference SI in relation to FIG. 3), making it possible to shift in time and / or in frequency the first signal thanks to the detection of a peak correlation. This first synchronization step is based on a central correlation 101c, ie comparing a complex sample of the signal combined at time t (IQ_COMBINE (t)) with a complex sample of the first signal SI after passing through the FIFO 731 and the NCO 732 at the same time t (IQ_NCO (t)). The result of the central correlation can be stored in a memory 102c.

Optionally, the correlator 742 can implement a second synchronization step, called fine synchronization.

This second synchronization step is based on the one hand on a “before” correlation lOlu (or “under” correlation), ie comparing a complex sample of the signal combined at time t -1 (I Q_CO MBINE (tl)) with a complex sample of the first signal SI after passing through the FIFO 731 and the NCO 732 at the instant t (IQ_NCO (t)), and on the other hand on an “after” correlation 101o (or “over” correlation), ie comparing a complex sample of the signal combined at time t +1 (IQ_COMBINE (t + l)) with a complex sample of the first signal SI after passing through the FIFO 7S1 and the NCO 7S2 at time t (IQ_NCO (t)).

The result of the "before" correlation can be stored in a memory 102u and the result of the "after" correlation can be stored in a memory 102o.

The difference obtained between the "before" correlation and the "after" correlation is transmitted to the controller 744, which can thus update the integer and fractional time delay values and supply them to the FIFO 7S1. This difference therefore makes it possible to adjust the fine offset and to adjust the Lagrange filter.

Controller 744 can also provide NCO 7S2 with the D / update setpoint.

During the second synchronization step, the central correlation makes it possible to verify that the algorithm is still locked on the correlation peak detected during the first synchronization step.

FIG. 11 illustrates an example of architecture for the frequency correction module 742.

The fine frequency correction, as well as the monitoring of the variation of the frequency, is based on a digital PLL comprising four modules in addition to the NCO 732. These modules implement the steps of FIG. 4. The same references are therefore used as those of figure 4:

a conjugate complex multiplication 42 between the complex samples of the combined signal (IQ_COMBINE) and the complex samples of the first signal SI after passing through the FIFO 731 and the NCO 732 (IQ_NCO);

a low-frequency filtering module 43 of the result of the multiplication, making it possible to limit the overall calculation noise. For example, the cut-off frequency is between 100Hz and 10kHz, with a nominal value at 300Hz);

a calculation of arctangent 44 of the result of the filtering, making it possible to estimate the phase shift between the combined signal and the first signal SI after passing through the FIFO 731 and the NCO 732;

a PID-type error correction module 45 making it possible to correct the phase shift and update the setpoint D /.

The controller 744 can update the smoothing parameter of the low pass filter 43 and the coefficients of the PID 45, and supply the NCO 732 with the setpoint D / thus updated.

FIG. 12 illustrates an example of architecture for the time correction module, phase and / or amplitude 743, based on an LMS type algorithm.

The LMS algorithm is based on a calculation of the coefficients of the filter 733 from a minimization of the quadratic error. Such an algorithm takes as input the complex samples of the first signal SI after passing through the FIFO 731 and the NCO 732 (IQ_NCO) and the complex samples obtained after subtraction of the first aligned signal from the combined signal (IQ_DES), corresponding to the estimate of the desired signal.

The complex samples of the estimate of the desired signal IQ_DES (corresponding for the algorithm to the error) are multiplied 121 by the convergence parameter m. The result is then multiplied by the samples IQ_NCO of the first signal SI corrected in time and in frequency in order to be able to update the coefficients W of the filter 733.

Returning to FIG. 7, the decimation 721, 722 and interpolation 75 filters are described below.

The first decimation filter 721, which makes it possible to transform the complex samples of the combined signal coming from the ADC 71 from a fixed frequency to a frequency at 2.F _ech depending on the transmission speed (“baudrate”) can be chosen. sufficiently selective to avoid aliasing of noise in the useful band.

The second decimation filter 722 which takes the complex samples from the first transmitting device (internal modulator) requires little stress because the signal is much cleaner (since the first transmitting and receiving device are located at the same site of broadcast, and possibly belong to the same equipment). It can be satisfied with a lower order compared to the first decimation filter 721.

Finally, the interpolation filter 75 from 2.F _ech to the fixed frequency of the DAC 76 has little constraint because the signal leaving the DAC 76 can be directly demodulated by the demodulator 77, for example an S2X demodulation chip.

According to at least one embodiment, implemented for example in the reception device illustrated in FIG. 7, the main steps of synchronization, frequency correction and adaptive filtering, implement the following specificities: for synchronization: use of sequences d 'separate scrambling for different modems; deactivation of the correlation during the header of the first signal (interfering signal); minimization of the number of correlations by adapting the time / frequency adjustment step; once the coarse synchronization has been achieved, the correlation can be used to follow the slow variations of the transponder by comparing the “before” correlation at time t-1 and the “after” correlation at time t + 1; an offset management system in the DDR external memory is provided so that the echo is always between t-1 and t + 1. The management of the shift in the memory can be done at the same time as a shift of the coefficients of the correlator; for frequency correction: use of a low pass filter before the arctangent function in order to limit the artefacts of this function; adjusting the parameters of the phase locked loop as a function of the cutoff frequency of the low pass filter determined from the measurement of the phase noise; for adaptive filtering: use of a finite impulse response filter; use of an LMS algorithm with a limited filter length (eg 15 coefficients); the LMS algorithm can work in blocks in order to limit the resources required for all the parallel calculations. We can reduce the size of the block in order to make the algorithm more responsive. This is a response time / FPGA resource compromise.

5.5 Receiving device

An example of implementation of the invention with an FPGA has been described above.

FIG. 13 more generally illustrates the simplified structure of a reception device according to one embodiment of the invention.

As illustrated in FIG. 13, a first device for receiving a first broadcasting site comprises a memory 131 (comprising for example a buffer memory) and a processing unit 132 (equipped for example with at least one processor, FPGA, or DSP), controlled or pre-programmed by an application or a computer program 133 implementing the method for processing a signal according to one embodiment of the invention.

On initialization, the code instructions of the computer program 133 are for example loaded into a RAM memory before being executed by the processing unit. 132. The processing unit 132 implements the steps of the processing method described above, according to the instructions of the computer program 133.

To do this, according to one embodiment, the processing unit 132 is configured to: receive at least one reference signal transmitted by at least one second transmission device from a second broadcast site, via the transponder, measure the phase noise associated with the transmission of the reference signal, receiving a combined signal comprising: an interfering signal, corresponding to a first signal transmitted by a first transmission device from the first broadcasting site, called the first transmission device, to destination of at least one reception device of the second broadcast site, said second reception device, via the transponder, and at least one desired signal, corresponding to a second signal transmitted by the second transmission device, intended for the first receiving device, via the transponder, aligning the first signal on the combined signal taking into account the phase noise, delivering a first aligned signal, subtracting the first aligned signal from the combined signal, delivering an estimate of said at least one desired signal.

5.6 Variants

A point-to-point mode has been described above implementing two broadcasting sites, each comprising a transmission device and a reception device. In particular, each broadcasting site implements a reception device as described above. It is therefore a symmetrical "full-duplex" transmission system, where the two modems, which receive the combination of the two signals (SI transmitted by the first transmission device and S2 transmitted by the second transmission device ), extract the signal they want to demodulate (S2 for the first receiving device and SI for the second receiving device) by subtracting the transmitted signal (SI for the first transmitting device and S2 for the second transmitting device ).

If the suppression is perfect and if the two modems transmit on exactly the same band, 50% of the spectral band can be used for other applications.

As a variant, the proposed solution can be applied in point-to-multipoint mode.

In the case of point-to-multipoint mode, a main modem (also called “Hub”), located on a main broadcast site, supplies a multitude of secondary modems (also called “end-user”). The transmission remains a “full duplex” transmission, but asymmetric.

As for the point-to-point mode, it is possible to gain spectral efficiency by using the same transmission frequencies for the different transmission devices and the same reception frequencies for the different reception devices.

FIG. 14 illustrates the architecture of such a point-to-multipoint mode by simplifying it to a main modem 141 communicating with three secondary modems 142, 143, 144 via a transponder 145. The application case conventional is of the order of a few tens to a few hundred modems.

In point-to-multipoint mode, it is generally possible for the secondary modems to directly demodulate the signal transmitted by the main modem, because its carrier Cm has a higher power than all the other carriers.

Thus the reception device according to at least one embodiment can be implemented only at the level of the main broadcast site for this point-to-multipoint mode, without being implemented at the level of the secondary broadcast sites.

According to other variants, in order to improve the operation of the echo cancellation technique, it is also possible: to add a gain amplifier (AGC) at the input of the reception device to limit false detections, and / or to limit the power difference between the carriers of the signals transmitted in the point-to-point mode, and / or to limit the difference in bandwidth between the carriers of the signals transmitted and / or to operate in the linear zone of the 'transponder amplifier.

Claims

1. A method of processing a signal, implemented by a reception device of a first broadcast site, said first reception device, comprising: the reception (221) of a combined signal comprising: an interfering signal, corresponding to a first signal transmitted by a transmission device from said first broadcast site, called first transmission device, to at least one reception device from at least one second broadcast site, called second reception device , via a transponder, and at least one desired signal, corresponding to at least a second signal transmitted by a transmitting device from said at least one second broadcasting site, called second transmitting device, to said first receiving device, via said transponder, aligning (222) said first signal with said combined signal, providing a first aligned signal, subtracting (223) from said first signal aligned with said combined signal, providing an estimate of said at least one desired signal, characterized in that said method comprises a prior calibration step comprising: receiving (211) a reference signal transmitted by said second transmission device, via said transponder, measuring (212) of the phase noise associated with the transmission of said reference signal, and in that said alignment (222) of said first signal on said combined signal takes into account said phase noise.

2. Method according to claim 1, characterized in that said phase noise measurement implements: determining a phase noise mask, centered on the carrier frequency used by said downlink transponder, estimating d 'a cutoff frequency associated with said template.

3. Method according to claim 2, characterized in that the determination of a phase noise mask implements the measurement of the power of the reference signal received on at least one frequency belonging to a narrow frequency band centered on the frequency. carrier frequency used by said downlink transponder.

4. Method according to claim 3, characterized in that said cut-off frequency is estimated from the standard deviation associated with said measurements of the power of the reference signal received on said narrow band.

5. Method according to any one of claims 2 to 4, characterized in that said measurement of the phase noise also implements the estimation of a residual phase noise, associated with said reference signal received outside the band. frequency defined by the carrier frequency used by said downlink transponder plus or minus the cutoff frequency.

6. Method according to any one of claims 1 to 5, characterized in that said alignment implements a first synchronization step, called coarse synchronization, comprising: a correlation between the combined signal and the first signal shifted in time and / or in frequency, for at least two offset values in time and / or in frequency, a determination of a correlation peak, an estimate of a coarse offset in time and / or in frequency between the combined signal and the first signal .

7. Method according to claim 6, characterized in that, said first signal carrying a header not obtained from a scrambling sequence, said correlation is deactivated during the reception of said header.

8. Method according to any one of claims 1 to 7, characterized in that said first signal is constructed from a scrambling sequence distinct from that used to construct said at least one second signal.

9. Method according to any one of claims 1 to 8, characterized in that said alignment also implements: a frequency correction of said first signal, implementing a phase locked loop (32), a time correction , phase and / or gain of said first frequency corrected signal, implementing adaptive filtering (33).

10. The method of claim 9, characterized in that said frequency correction implements at least one iteration of the following steps: frequency shift (41) of the first signal according to a set point D /, conjugate complex multiplication (42) of said signal. combined with the first signal shifted in frequency according to the setpoint D /, low-pass filtering (43) of the result of the multiplication, estimation of the phase shift (44) between said combined signal and the first signal shifted in frequency from the signal obtained at the output of the low-pass filtering, error correction (45) delivering a D / update instruction.

11. The method of claim 10, characterized in that said low-pass filtering implements a low-pass filter allowing frequencies lower than said cut-off frequency to pass according to any one of claims 2 to 4.

12. Method according to any one of claims 10 and 11, characterized in that said error correction implements a corrector of the “Proportional, Integral, Derivative” (PID) type, the coefficients of which depend on the cutoff frequency according to. any one of claims 2 to 4.

13. Method according to any one of claims 9 to 12, characterized in that said correction in time, phase and / or gain implements an algorithm of the quadratic error minimization type, implementing at least one iteration of following steps: adaptive filtering of the first frequency corrected signal, implementing an adaptive filter, delivering said first aligned signal, updating the coefficients of said adaptive filter, said updating of said coefficients being implemented after said step of subtracting said said first signal aligned with said combined signal, and multiplying said estimate of said at least one desired signal by a convergence parameter m.

14. The method of claim 13, characterized in that said convergence parameter is determined from said residual phase noise according to claim 5.

15. A computer program comprising instructions for implementing a method according to any one of claims 1 to 14 when this program is executed by a processor.

16. Device for receiving a signal from a first broadcast site, said first receiving device, comprising: receiving means configured to receive a combined signal comprising: an interfering signal, corresponding to a first signal transmitted by a device. transmission from said first broadcast site, called first transmission device, to at least one reception device of at least one second broadcast site, called second reception device, via a transponder, and at least one desired signal, corresponding to at least a second signal emitted by a transmitting device from said at least one second broadcast site, said second transmitting device, to said first receiving device, via said transponder, alignment means configured to align said first signal with said combined signal, delivering a first aligned signal, subtracting means configured to subtract said first aligned signal from said combined signal, providing an estimate of said at least one desired signal, characterized in that, said receiving means being configured to previously receive a reference signal transmitted by said second transmission device, via said transponder, said first reception device also comprises phase noise measuring means configured to measure the phase noise associated with the transmission of said reference signal, and in that said means for alignment are configured to align said first signal with said combined signal taking into account said noise d th phase.