The present invention relates to a method for processing a signal formed by a pulse sequence comprising at least one repeating unit formed of at least one pulse, said pattern being repeated in the pattern signal with a repetition period.

The invention lies in the field of signal processing, in particular digital signals, and finds particular application in the field of characterization of radar pulse sequences.

In effect, a periodic or cyclo-periodic signal consists of pulse patterns which are repeated, is generally characterized by parameters which the pattern repeat period, or PRM and the pulse repetition intervals, or PRI signal. In practice, the characterization of a radar signal intercepted in a listening time slot allows in particular to identify the transmission mode of the signal intercepted or track pulse sequences from the same transmission.

In known manner the pulse repetition intervals are used to characterize such a pulse sequence.

However, PRI values ​​are distorted if random loss pulse signal, a phenomenon also known under the name of urban sprawl, which occurs in practice.

The invention aims to improve the characterization of the temporal repetition periods of the pulses in a pulse sequence, in particular in the case of random losses.

For this purpose, in a first aspect, the invention provides a method of processing a signal formed by a pulse sequence comprising at least one repeating unit formed of at least one pulse, said pattern being repeated in the signal with a pattern repetition period. The method comprises the steps of:

- estimation of the pattern repetition period of said signal,

- calculating a phase sequence according to a date of arrival of each pulse relative to a selected reference arrival time and the estimated pattern repetition period,

- from said phase sequence calculated estimate of at least one phase value and an associated standard deviation, said phase value being associated with a representative phase time of the repeating unit;

- obtaining and operating signal characterization parameters using the estimated phase values.

Advantageously, the method of the invention allows to characterize the temporal characteristics repetitions of a sequence of pulses in a robust manner to random pulse losses.

The proposed method applies for pulse to successive or nested sequences pulse trains, the repetition period of pulses of a pulse train that can be assigned or not a random noise or pseudo-random inducing fluctuation of the repetition period relative to a mean value.

The method according to the invention may have one or more of the following characteristics, taken separately or in any technically acceptable combination.

It comprises, after the step of calculating a phase sequence, a step of constructing a histogram phases, and extraction of one or more spike (s) of the histogram, each peak being associated with a phase time of the pulse sequence.

The step of estimating a phase associated with a time of the pulse sequence, said time being associated with a peak of the histogram phases, comprises an extraction of a subset of arrival dates around said peak, and an accurate estimate of the phase value associated with said time and a variance associated with the estimated phase value.

When the pulse sequence comprises a plurality of interleaved pulse trains, each pulse train having an associated phase time accurately estimating the phase associated with time comprises a vector sum of unit vectors, each vector having a orientation equal to a calculated phase value, with the pattern of said previously estimated signal repetition period, from a time of arrival of said subset, said vector summation providing a resultant vector.

The phase value associated with said estimated time is equal to the argument of said resultant vector.

When the pulse sequence comprises a plurality of successive pulse trains, each pulse train having an associated phase time accurately estimating the phase associated with a currently processed further comprises an estimate of a period repeating units corresponding to said time from the subset of arrival dates considered.

The estimate of the signal of the pattern repeat period comprises a first coarse estimate of a first estimated value of the pattern repetition period, and a second refined estimate of a second estimated value of the pattern repeat period.

When the step of obtaining a sequence of phases for each pulse arrival date t n , is applied a period equal to the estimated folding pattern repetition period according to the following formula:

Where
férieure of real number x, t 0 is the reference arrival date and λ the estimated pattern repetition period.

According to another aspect, the invention relates to a device for processing a signal formed by a sequence of successive pulses, adapted to implement a method as briefly described above.

This device includes at least one computing processor adapted to implement:

- an estimation unit of the pattern repetition period of said signal,

- a calculating unit, based on a date of arrival of each pulse relative to a selected reference arrival time, and the estimated pattern repetition period, a sequence of phases,

-a estimation unit, from said sequence of calculated phases, at least one phase value and an associated standard deviation, said phase value being associated with a representative phase time of the repeating unit;

- an obtaining unit and operation of the digital signal characterization parameters using the estimated phase values.

According to one embodiment the device further comprises a structural unit of a phase histogram, and extracting one or more spike (s) of the histogram, each peak being associated with a phase of the time of pulse sequence.

According to another embodiment, the invention concerns a computer program comprising software instructions that, when implemented by a programmable device, implement a method for processing a signal formed by a sequence pulse as briefly described above.

According to another aspect, the invention relates to an information recording medium, comprising software instructions for performing a method of processing a signal formed by a pulse sequence such as briefly described above, when these instructions are executed by a programmable device.

Other features and advantages of the invention emerge from the description which is given below, with illustrative and non limitative, with reference to the appended figures, in which:

- Figure 1 is a schematic view of a sequence of pulses with interleaved pulse trains;

- Figure 2 is a block diagram of the main steps of a method according to an embodiment of the invention;

- Figure 3 schematically illustrates a number of repetition periods of a digital signal and an associated phase histogram;

- Figure 4 is a diagram showing functional blocks of a programmable device adapted to implement the invention.

FIG 1 schematically illustrates a digital signal S having a pulse sequence 10, consisting of pulses \ l 2 , ... l n received in temporal succession. Each pulse I k is schematically illustrated on a time axis.

The pulse sequence 10 is a pattern of repeated sequence, and comprises, in the simplified example illustrated, four periodic bit streams interleaved.

Of course, in practice, the actual digital signals treated include any number of nested P signals.

In the example of Figure 1, the pattern 12 is composed of four successive pulses, to 4 , and pattern repetition period λ.

Differences between arrival on the time axis, known by the abbreviation DTOA ( "difference of time of arrived") between two successive pulses are used to calculate the values of interval between successive pulses, denoted PR respectively PRI 4 .

Each of the pulse trains overlapping in the pulse sequence 10 consists of pulses are repeated with a repetition period λ, equal to the pattern repetition period of the pulse sequence 10.

In other words, each pulse being T p is made up of pulses

The reference arrival time t 0 is chosen arbitrarily, for example equal to the date of arrival of the first pulse of the pulse sequence under consideration.

The random loss of a subset of the pulse induces an error on the calculation of PRI values k of a sequence of received pulses.

It is proposed to calculate the phase values for each pulse, representative of the difference P ,, for each pulse between its arrival date and a record date periodized in relation to the pattern repeat period. In particular, P to P 4 are illustrated on Figure 1 for the example considered.

The phase values ​​are calculated as follows.

From a sequence of arrival dates {t k }, we compute the phases, with a λ given period of folding with respect to a reference arrival time t 0 by the following formula:

Where \ _x] is the integer part lower (or default) of the real number x.

The phase values 9 k greater than or equal to 0 and strictly less than λ.

The phase values thus calculated have a better robustness to random losses of pulses that DTOA values measured between successive pulses, when the respective values of reference arrival time t 0 and period λ are known or previously determined.

The main steps of an embodiment of the signal processing method according to the invention will be described below with reference to Figure 2.

This process applies in the case of a pulse sequence formed of a plurality of pulse trains P interleaved with P an integer greater than 1, each pulse train forming a phase time m (i) of the pulse sequence. Such a pulse sequence is also known as "P stagger times." For example, Figure 1 illustrates a stagger signal to 4 times.

The method of the invention also applies to the case of a pulse sequence formed of a plurality of pulse trains P that follow one another in time, each of the pulse trains having a pattern repeat period associated, equal to the pulse repetition period of said pulse train.

It is not necessary to know the type of sequence in advance, the method allows to deduce.

In a case of application, the pulse sequence is in most affected, voluntarily being created by the designer of the radar, a noise called jitter in English, the pulse repetition period fluctuating randomly or pseudo-around an average value.

Each time m (i) has an associated phase value, the pulses of the pulse train T, associated with m (i) having a repetition period λ which is the same as the repetition period associated with the sequence of pulses.

In the embodiment described, there are arrival dates: {t k , n} l≤k≤ pulses of a pulse sequence to be processed in a given time reference. These arrival dates are for example stored in a memory of a computing device adapted to implement the invention.

During a first step 20 of estimating the pattern repetition period, placing in work in the embodiment described, a pattern repeat period value, simply referred to as PRM, the pulse sequence, denoted A is calculated.

This calculation comprises two sub-steps, a first sub-step 22 of the first estimate, also known as rough estimate, for obtaining a first estimated value of the PRM A, followed by a second sub-step 24 second estimate, called also fine estimate, for obtaining a second estimated value, a, of the PRM.

The first sub-stage 22 implements any known method of estimating the pattern repeat period. For example, the pattern estimation methods are described in the "New technologies for the deinterleaving of repetitive sequences," HK Mardia et al, published in IEE Proceedings F Vol 136 No. 4, pp. 149-154, in August 1989. This estimate 22 is produced for example by the methods of histograms differences in dates of arrival (or DTOA). Alternatively, any other method for estimating the pattern repeat period, using a Fourier transform or a wavelet transform may be used. However, as in this first sub-step 22 the accuracy of the estimate is not sought, it is best to

At substep 24 of second estimate of the PRM, is calculated for each stored arrival date, an integer index associated according to the first estimated value A g "of the PRM:

k(i) = (EQ 2)

g

Then performs the following calculation:

The variance associated with this estimate is given

Where (a) ~ ^ = - ^ - (EQ 4)

n — n

n being the number of pulses of the pulse sequence, and σ is the standard deviation of the measurement time t.

The standard deviation σ is the measuring instrument of arrival dates. He is known for each instrument, to the conditions of measurements interception.

The above mentioned calculations apply in particular for the modeling of arrival dates according to a linear model based on the values ​​of k (i), in other words for a sequence of periodic pulses of period λ:

t i = t 0 + A - k (i) and + (EQ 5)

The above calculations apply for pulse sequences P trains nested, possibly noisy by a non-cumulative noise.

In the case of sound effects by non-cumulative noise, each arrival can be written as follows:

t i = (i - I) + A m (i) + v i (EQ 6)

Where v i is a wrist-talented center type σ ν and φ Euge) the highlighted phase du moment m (i).

Alternatively, the pulse sequence received is affected by a noise (jitter) Cumulative:

t i = t + At the _ l + φ + v i (EQ 7)

Where v i is a centered Gaussian white noise standard deviation σ v and φ the phase of the pulse train of the sequence considered.

In this embodiment, applying the following calculation to the second estimate of the PRM:

X = ^!L (EQ 8)

n - l

The variance associated with this estimate is given by:

n being the number of pulses of the sequence and σ being the standard deviation of the measurement time t. This standard deviation quadratically combines the standard deviation of measurement of time t and the standard deviation σ v of the law jitter.

Returning to Figure 2, step 20 of estimating the PRM is followed by a step 30 of calculating a phase sequence, by calculation according to formula (EQ1), taking into account the arrival reference t 0 selected PRM and the estimated value X obtained in step 20.

One thus obtains a sequence of phases: [9 k , 1 (tk p ) ^o ) (EQ 10)

Where t k p is the date of arrival of the kth pulse time of the m (p), and the values 0 (t k _) are calculated using equation (EQ 1) with the repetition period value

pattern X m (p) estimated.

The estimate of the end stage m (pX , o associated with the time m (p) is given by:

fim(p to = arg[S(p)] (EQ 1 1 )

Where arg [S (p)] is the argument of the vector (p).

The variance associated with this estimate is given by:

Where
is the norm of the vector sum S (p) and ln (x) is the natural logarithm of x.

Returning to Figure 2, step 70 accurate estimate of the time phase of m (p) is followed by a step 80 for characterizing the initial pulse sequence by the recording of a summary comprising parameters characteristic of the sequence:

- In the case where the pulse sequence treated consists of interleaved pulse trains are summarized comprises:

or you name moments P;

o The calculated values ​​of the pattern repetition period To, and the associated variance;

o For each time, the estimated phase value fi m (pX , o , and the associated variance.

In the case where the pulse sequence treated consists of successive pulse trains, the summary comprises:

or you name moments P,

o Each time:

■ the calculated value of the pattern repeat period At m (p) , the associated variance,

■ the estimated phase value fi m (pX , o , and the associated variance.

These summaries to characterize the pulse sequence processed are stored, and are operable in an operating step 90 for various applications, for example comparison with pre-recorded signal characteristics for identifying the signal transmission mode intercepted.

Of course, the summaries are usable for any other application processing on the radar pulses.

The process of the invention is implemented by a programmable device, which can be either integrated in a radar reception device, is independent and used for further processing.

A programmable device 100, as shown schematically in Figure 4, adapted to implement the invention, typically a computer, includes a CPU of Treatment 1 10, or CPU, capable of executing computer program instructions when the device 100 is turned on. The device 100 also includes information storage means 120, such as registers or memories, capable of storing executable code instructions for the implementation of programs comprising code instructions able to implement the methods according to the invention.

In one embodiment, the software instructions for implementing the method of processing a signal formed by a pulse sequence such as briefly described above are stored on an information recording medium, e.g. an external memory, a mass storage memory, an external drive. When the information recording medium is connected to the programmable device, dispositf is adapted to execute these software instructions, as explained above.

Optionally, the programmable device 100 includes a screen 130 and an input 140 through an operator command, for example a keyboard, optionally an additional means score 150, such as a mouse for selecting graphical elements on the screen 130.

The various functional blocks 1 10 and 150 of the device 100 described above are connected via a communication bus 160.

In variant not shown, the programmable device 100 is in the form of programmable logic components, such as one or more FPGAs (English Field-Programmable Gate Array), or in the form of dedicated integrated circuits of the ASIC type ( English application-specific integrated circuit).

Advantageously, the proposed method allows to treat any type of pulse sequence, whether nested or successive pulse trains.

The invention allows processing signals having a plurality of repeating units of periods and phases for each pattern repetition period.

Advantageously, the proposed method allows to obtain reliable abstracts including in the case of random losses of pulses, and in addition, the level of reliability of each estimate is quantifiable through the associated estimation variance is also calculated and stored.

CLAIMS

1. - A method of processing a signal formed by a pulse sequence comprising at least one repeating unit formed of at least one pulse, said pattern being repeated in the pattern signal with a repetition period, characterized in that it comprises the following steps:

- estimate (20) of the pattern repeat period of said signal,

- calculating (30) a sequence of phases as a function of a date of arrival of each pulse relative to a selected reference arrival time and the estimated pattern repetition period,

- from said sequence of phases calculated, estimation (60, 70) of at least one phase value and an associated standard deviation, said phase value being associated with a representative phase time of the repeating unit;

- obtaining (80) and operating (90) of the signal characterization parameters using the estimated phase values.

2. - Method according to claim 1, comprising, after the step of calculating (30) a phase sequence, a construction step (40) of a phase histogram, and extracting one or more peak (s) of the histogram, each peak being associated with a phase time of the pulse sequence.

3. - The method of claim 2, wherein the step of estimating a phase associated with a time of the pulse sequence, said time being associated with a peak of the histogram phases, comprises a extraction a subset of arrival dates around said peak, and an accurate estimate of the phase value associated with said time and a variance associated with the estimated phase value.

4. - Method according to claim 3, wherein when the pulse sequence comprises a plurality of interleaved pulse trains, each pulse train having an associated phase point, the accurate estimation of the associated phase when comprises a vector sum of unit vectors, each vector having a direction equal to a calculated phase value, with the pattern repetition period of said signal having been estimated from an arrival of said subset, said vector summation providing a resultant vector.

5. - Method according to claim 4, wherein the phase value associated with said estimated time is equal to the argument of said resultant vector.

6. - The method of claim 3, wherein when the pulse sequence comprises a plurality of successive pulse trains, each pulse train having an associated phase time accurately estimating the phase associated with a when treated further comprises an estimate of a pattern repetition period corresponding to said time from the subset of arrival dates considered.

7. -Procédé according to any one of claims 1 to 6, wherein the estimate of the signal of the pattern repeat period comprises a first coarse estimate of a first estimated value of the pattern repetition period, and second refined estimate of a second estimated value of the pattern repeat period.

8. A process according to any one of claims 1 to 7, wherein in the obtaining of a phase rotation step for each pulse arrival time t n , is applied a refolding period equal to the estimated pattern repetition period according to the following formula:

higher real number x, t 0 is the reference arrival date and λ the estimated pattern repetition period.

9. A device for processing a signal formed by a pulse sequence comprising at least one repeating unit formed of at least one pulse, said pattern being repeated in the pattern signal with a repetition period, characterized in that it comprises at least one computing processor adapted to implement:

- an estimation unit of the pattern repetition period of said signal,

- a calculating unit, based on a date of arrival of each pulse relative to a selected reference arrival time, and the estimated pattern repetition period, a sequence of phases,

-a estimation unit, from said sequence of calculated phases, at least one phase value and an associated standard deviation, said phase value being associated with a representative phase time of the repeating unit;

- an obtaining unit and operation of the digital signal characterization parameters using the estimated phase values.

10.- Device according to claim 9 further comprising a structural unit of a phase histogram, and extracting one or more spike (s) of the histogram, each peak being associated with a phase time the pulse sequence.