DE102008017974A1 - Temporally fluctuating signal adjusting method for preprocessing radar data for Doppler-evaluation, involves converting remaining time differences into corresponding set of sampling intervals and adjusting signals in specific measure - Google Patents

Abstract

The method involves determining a pulse repetition interval from available time stamps. Process data is extracted from the available time stamps. A new histogram is formed over sampling values and a maximum counted value is stored, till an optimum pulse repetition interval is located with a retrieval of overall maximum of a maximum counted value of the histograms for the respective flow. Remaining time differences are converted into a corresponding set of sampling intervals and fluctuating signals are adjusted in a specific measure. Independent claims are also included for the following: (1) an adjustment device for signals (2) a storage medium for computer program data with stored program data for performing a method for adjustment of temporally fluctuating signals.

Description

For the application Imaging techniques in radar technology such as synthetic aperture radar (SAR) is the coherent one Processing of signals indispensable. Prerequisite for this is the correct evaluation of the phase along identical distance ranges the recorded pulses. This can be the signals impressed Take advantage of Doppler frequencies and create a transverse resolution. In bistatic Geometries where the radar transmitter and receiver are spatially separated are, in addition the problem that both parties are not on a common timebase To fall back on can. in this connection The challenge arises, a synchronization between transmitters and receiver to establish.

In the measuring practice are manifold Types of receiver systems known, which are used to receive emitted signals. These can then be divided according to how a measurement recording made becomes. Common are manually set times, to which a Measurement begins. Another possibility is to take measurements each time a solid Time interval. To are i. d. R. quartz watches are present in digital recording devices, which allow such a recording mode.

A third alternative is the amplitude-dependent measurement triggering, too here with the help of digital recording devices. In such systems will a temporary or even an unlimited measurement recording started at the time at which the strength of the incoming signal reaches or exceeds a previously set amplitude level. Digital recording devices is mean that she is an original continuous signal in time discrete sections (the so-called. Sampling interval) and save it with the associated triggering times. Only this third alternative offers locally separate radar systems the possibility, to perform a synchronization by means of the directly received signal.

These However, the approach has weak points that are more or less less impact on the temporal stability of recorded signals. For one, it is vulnerable across from temporal fluctuations, the structure of the receiver apparatus itself be evoked. Here arise depending on the measuring principle temporal errors of the order of magnitude of several sampling intervals. For records with few multiples the Nyquist frequency can Therefore, phase errors arise that have a desired or expected coherence of the Destroy signals. In addition, the signal recording by the use of different Local oscillators be imprinted a temporal drift, which is deposited in the Timestamps are reflected.

The Inaccuracies are due to the magnitude of the pulse repetition interval not obvious in the millisecond to microsecond range but in a high-frequency sampling in one to two digits Nanosecond a significant deterioration of coherence. To the others, the detection times depend on the steepness of the Signal rise edge and the achieved amplitude level of each Signal. This results in the data acquisition a delayed measurement triggering for weakening Signals connected with a shortened Recording of the useful signal in the direction of the beginning of the recording window. Both effects lead together to a temporal shift of the pulses, which behaves erratically and therefore is unpredictable.

to Processing of the received signals come in accordance with the prior art so far two known methods into consideration. These are for one the correlation and the use of CFAR algorithms (CFAR = Constant False Alarm Rate) to the digital record of the Signals.

Correlation:

This Approach has found widespread use in signal processing. He exploits the fact that temporally adjacent signals a high similarity to each other. The position of a signal to be shifted, at the maximum relative to the previous signal, the correlation value becomes, thereby marks the optimal position. After application of pulse to pulse form a new signal data matrix that i. d. R. an improvement the timing of the raw data causes.

unfavorable Way you can after this application, both a successive trend and erratic Shifts occur. The latter effect arises primarily in incomplete recorded signals. This is the automatic character of the Approach lost, because a manual readjustment is necessary. In addition, after passing through the correlation algorithm, the coherence of the recorded signals are not guaranteed.

CFAR algorithms:

basis for this Procedure is the consideration a constant false alarm rate. This does not become a global threshold but agreed to the medium noise adapted dynamic Detection level of the incoming signals. Advantage over one Fixed amplitude threshold is that in attenuating Signals a time shift of the measurement triggering time results and so ideally the signal strength is tracked.

This Method can be used at the same time in the radar measurement, if the technical requirements are met. On the other hand exists i. d. R. the possibility the procedure only in the later Apply signal processing to the data.

The Applying the CFAR algorithms put a note of that respective signal-to-noise ratio of recorded signals to an adaptive change to execute the detection threshold can. This can happen in cases necessary in which the receiver noise is constant in the course of a measurement is, however, attenuates the useful signal to be recorded or strengthened. Since the variation of this threshold relative to the signal-to-noise ratio is not must be known, lets the optimum level for each single measurement triggering do not predict beyond doubt and there remains a temporal blur. One Another disadvantage is the increased apparative effort, for the implementation this method in contrast to a measurement triggering with a constant trigger threshold is required.

The solution using correlation or CFAR algorithms look exclusively at the present signal information and leave the also recorded Timestamp information without in front.

The The object of the invention is to provide fluctuating signals from a transmission source continuously or in sections with a constant Pulse repetition frequency emitted and from a receiver device were taken, reliable precise in time align so that the originally coherent behavior of the radiated Signals is restored.

The The object is achieved by a method according to claim 1 and a device according to claim 2 solved.

The sees solution according to the invention as input variables for the process, the amplitude-triggered signal data, the associated time stamp marks the measurement triggering and the sampling interval used. For an indefinite period First measurement without filing time stamps is in advance of the procedure one Extraction of temporally identical recording windows by means of manual set amplitude thresholds to subsequently the Release dates win every signal window. The procedure starts with a (section by section) estimate of the pulse repetition interval from the present timestamps. Subsequently, will extracted from the total amount of timestamp of that part, the to go into the further processing. Starting from the beginning certain pulse repetition interval is in an optimization loop of these timestamp values after their artificial coarsening gradually changing Pulse repetition interval deducted. This is for each loop a new histogram over the lines (= samples) formed and the maximum count stored. The optimal pulse repetition interval is reached exactly when the global maximum of the maximum counts of all calculated histograms was found. Finally is the pulse repetition interval found from all timestamps subtracted, the remaining time differences in the corresponding Number of sampling intervals converted and the signals in this Extent shifted.

By the invention makes it possible recorded signals with a constant pulse repetition frequency, the were recorded in temporally limited recording windows, and across from the transmission signal temporal shifts (also of unknown origin) have to align so that there is a coherence of the signal phases can be produced. In contrast to the previously known solutions, it is now necessary only the exclusive Use of timestamp values to align the corresponding ones Signal data.

The method thus leads in all cases to a significant improvement in the time alignment of recorded signal data, which experienced a time error in the measurement triggering during recording. A relative deterioration of alignment with the original data set can no longer occur if the boundary conditions are met. It is only necessary for the application of the method to have the knowledge of the signal data to be aligned, the associated time stamps for each measurement triggering and the size of the sampling interval used. A knowledge about what the signal is not required for the procedure.

One Another advantage is the ability of the procedure, also incomplete correctly move recorded signals. This is above all For signals of importance, in which the signal rising edges, the crucial for the amplitude detection are, could not be recorded.

The new method can be implemented after implementation in a program on a standard computer real-time, as he with his extensions in the 2 is shown because the functions are kept simple in the algorithm used. Another reason for a low computation time is due to the fact that in the process only an optimization of the recorded time stamps and not on the signal data, which require a significantly larger volume of data to perform.

specially for records that recorded in bistatic radar measurements with moving transmitter or receiver The new process will change the real lighting conditions played correctly. In particular, happens by the application a transfer of the Record of an amplitude-triggered representation in the same signal increases can be found in one line, in a distance-dependent representation, in the same times (= distances) in one line are. This will be an important step in preprocessing to bistatic SAR processing.

Hereinafter, the invention with reference to the 1 to 12 explained in detail.

• I. Amplitudengetriggerte Signaldaten xt, die in digitaler Form in einer Datenmatrix vorliegen, bestehend aus den Dimensionen Abtastwerte und Pulsnummer (3)
• II. Die Anzahl der aufgenommenen Pulse N, entnommen aus Dimensionen von I.
• III. Die Anzahl der Abtastwerte je Puls Nsam, entnommen aus Dimensionen von I.
• IV. Die Größe des Abtastintervalls dsam in Sekunden
• V. Der zeitliche Beginn jedes aufgenommenen Pulses trig in Sekunden mit einer zeitlichen Auflösung mindestens in der Größe des Abtastintervalls (4)
• VI. Die Größe des eingestellten Pretriggers ptsam in Anzahl von Abtastintervallen

The following input variables are required for the procedure:

• I. Amplitude-triggered signal data xt, which are present in digital form in a data matrix, consisting of the dimensions of samples and pulse number ( 3 )
• II. The number of recorded pulses N, taken from dimensions of I.
• III. The number of samples per pulse Nsam, taken from dimensions of I.
• IV. The size of the sampling interval dsam in seconds
• V. The beginning of each recorded pulse trig in seconds with a temporal resolution at least the size of the sampling interval ( 4 )
• VI. The size of the set pretrigger ptsam in number of sampling intervals

• VII. Mit dp wird die erforderliche Anzahl an Pulsen festgelegt, die mindestens erreicht wer den muss, um daraus die Existenz einer in den Daten auftretenden PRI zu erkennen. Alternativ bietet sich eine Festlegung in Abhängigkeit von der Gesamtzahl N aufgenommener Pulse in Prozent an.
• VIII. Mit tkrit wird entschieden, ob die Auswahl des Optimerungsbereichs über die kontinuierlichste Messauslösung oder über das stärkste Signal in einem Signalblock erfolgt.
• IX. Mit nv ist der Jitterbereich der Zeitstempel in Vielfachen des Abtastintervalls dsam an zugeben, um gerundete Werte für eine Auflösungsverringerung der Zeitstempelwerte zu erhalten. Hintergrund ist die in Schritt 12 des Algorithmus benötigte Erhöhung der Zählwerte in der Histogrammauswertung.
• X. Mit ch wird entschieden, ob eine Interpolation fehlender Zeitstempel und ein Auffüllen der Signaldaten mit Nullvektoren erfolgen soll.
• XI. Mit NPRIx wird die maximal zulässige Anzahl von Pulsen vereinbart, die in einem Block fehlen dürfen (sog. Burstfehler) und die der Algorithmus für die Verarbeitung toleriert. Bei Überschreitung des Schwellwertes NPRIx (z. B. an den Rändern des Datensatzes) wird derjenige Datenbereich im Algorithmus weiterverwendet, der die Bedingung vollständig erfüllt.
• XII. Mit tsbit wird die Höhe der Triggerschwelle definiert, falls bei der Aufnahme der Daten ein Pretrigger eingestellt wurde.

In addition, manual control parameters must be specified:

• VII. Dp defines the required number of pulses that must be reached at least to recognize the existence of a PRI appearing in the data. Alternatively, a determination as a function of the total number N of pulses taken is given in percent.
• VIII. With tcrit it is decided whether the selection of the optimization range is made via the most continuous measurement triggering or via the strongest signal in a signal block.
• IX. With nv, the jitter region is to indicate the timestamps in multiples of the sample interval dsam to obtain rounded values for a resolution reduction of the timestamp values. The background is the increase in the counts in the histogram evaluation required in step 12 of the algorithm.
• X. With ch it is decided whether an interpolation of missing timestamps and a filling of the signal data with zero vectors is to take place.
• XI. NPRIx is used to set the maximum number of pulses that may be missing in a block (so-called burst error) and that the algorithm tolerates for processing. If the threshold value NPRIx is exceeded (eg at the edges of the data set), the data area in the algorithm which completely fulfills the condition is used further.
• XII. The height of the trigger threshold is defined with tsbit, if a pretrigger was set during the recording of the data.

The required process steps are described below:

Calculation of the occurring in the record Pulse repetition intervals from timestamp values:

• 1. Rounding the timestamp trig to an integer Multiples of the recording window Nsam · dsam
• 2. Application of a one-dimensional median filter with a window length of five Values for calculating a coarser one Timestamp vector trigh
• 3. Formation of elementary (= adjacent) differences of trigh for the ranges first to penultimate and second to last value and subsequent subtraction. By finding those subtraction results that are greater than twice the length of the recording window, after the addition of 1, all the pulse numbers of the discontinuities sps are obtained in trigh. In this lot are still all jumpstones len, which come from both a change in the PRI and missing pulses.
• 4. Formation of elementary differences of sps and finding remaining ones Jump points whose minimum size is the value exceeds dp (parameter VII). Addition of 1 results in the number of different PRIs in the timestamp record in variable spsn.
• 5. To avoid incorrect specification of occurring PRIs in the variable spsn, which is characterized by a small temporal variation timestamp values, a histogram of the element-wise differences formed by trigh and those counts extracted, which is greater than dp are. Exist in the histogram directly adjacent PRIs whose Cause in the rounding in step 1, these become one combined PRI.

• 6. Extraktion der Zeitstempelwerte trignPRI aus Zeitstempelvektor trig und Daten xt mit j-ter konstanter PRI nach Vorgaben aus Schritt 5.
• 7. Aufbau der Zeitdatenbasis durch Bestimmung der PRI aus Zeitstempelwerten trignPRI mittels Berechnung des Medians der elementweisen Zeitdifferenzen, Subtraktion dieser PRI von den Zeitstempelwerten trignPRI und abschließende Subtraktion des Medians zur Bildung der Zeitstempeldifferenzen trigp. Dieser stellt den Startvektor für die nachfolgenden Schritte dar (5 und in der Nahbereichsdarstellung durch Anwendung der Modulo-Funktion 6).

The information about the jumps sps and spsn is used to determine the run variables and the time and data ranges for j-constant PRIs in the subsequent loop algorithm (steps 6.-25.), Which is run through j. If j = 1, only one PRI is available and the complete data set is taken over to the next step 6. For j> 1, the data record with the associated time stamps is divided at the jump points of the PRI and processed sequentially in blocks.

• 6. Extraction of the timestamp values trignPRI from timestamp vector trig and data xt with j-constant PRI according to the specifications from step 5.
• 7. Construction of the time database by determining the PRI from timestamp values trignPRI by means of calculating the median of the element-wise time differences, subtracting this PRI from the timestamp values trignPRI and finally subtracting the median to form the timestamp differences trigp. This represents the start vector for the subsequent steps ( 5 and in the near-field representation using the modulo function 6 ).

• 8. Detektion fehlender Zeitstempel durch Untersuchung, ob Zeitunterschiede in trigp größer als das 1.5-fache der PRI existieren. Anschließend erfolgt die Auswahl von Pulsblöcken mit kontinuierlicher Messauslösung unter Berücksichtigung eines Schwellwertes dp für die Mindestgröße eines Blocks.
• 9. Auswahl und Extraktion eines Zeitstempelblocks nach dem Kriterium der längsten kontinuierlichen Messauslösung oder alternativ nach maximaler Signalstärke des mittleren Pulses eines Blocks durch Entscheidungsvariable tkrit (Parameter VIII).
• 10. Subtraktion des Medians, Anwendung der Modulo-Funktion mit Bereich [–0.5·PRI, +0.5·PRI] zur Eliminierung von Ausreißern und abschließende Umrechnung in dimensionslose Abtastwerte mittels Division durch das Abtastintervall dsam zum Erhalt von df1 (7).

Extraction of a range df1 from timestamp values trigp at which the optimization is performed:

• 8. Detection of missing timestamps by examining whether time differences in trigp greater than 1.5 times the PRI exist. Subsequently, the selection of pulse blocks with continuous measurement triggering taking into account a threshold dp for the minimum size of a block.
• 9. Selection and extraction of a time stamp block according to the criterion of the longest continuous measurement triggering or alternatively to maximum signal strength of the mean pulse of a block by decision variable tcrit (parameter VIII).
• 10. Subtraction of the median, use of modulo function with range [-0.5 · PRI, + 0.5 · PRI] to eliminate outliers and final conversion into dimensionless samples by division by sampling interval dsam to obtain df1 ( 7 ).

Determine a remaining, PRI-dependent trend in the extracted data df1:

• 11. Calculation of the sign vorz and the Slope of the trend of extracted timestamp df1 by education the difference of the medians from the third and second quarters of the vector. It has proved to be practical, the extent D of this slope to the To increase value 1, by the value D = 0 in the possible limit case to avoid. Then D is multiplied by a factor of 3 to get a Reserve for to maintain the loop pass in step 12.
• 12. In a loop, starting with D / 3 as a slope, a straight line function bP is formed, which is subtracted from the time stamp differences df1 and then rounded to a selected multiple of the sampling interval nv (parameter IX). This rounding factor nv is device-dependent and must be set to the fluctuation range of the jitter ( 8th ) to obtain a sufficient number of counts in the following histogram formation. Subsequently, a histogram with the length D is formed over all rows, the maximum histogram value is determined and stored together with the corresponding pitch c. For the next loop pass, the slope bP is incrementally changed in the direction of the sign vorz. This process is repeated until, as a termination condition, the difference between the current and the last detected optimum slope c is less than D. This is the optimal correction slope found ( 9 ).
• 13. Will be more than an optimum after completion of step 12 found, is an initial Selection required. This is done by determining the elemental Differences, stored in variable h, on the basis of which possible jumps be found between the ranges of values of c. Are jumps existent, so may after supplementation of h by 1 as the first value and the length of c as the last value the element-wise difference the maximum with associated position and value to be calculated in vector h. This is clearly the largest contiguous area Optima [c (1) c (end)].
• 14. Now the loop from step 12 is repeated with an increased accuracy of 10 according to the new limits from step 13, so that in c2 all slopes of the maximum values of the histo are stored.
• 15. If c2 is not just one value, there are possible count differences between the optima find and the optimal range analog to determine step 13. Finally, the optimal straight line function with median of c2 as slope, vorz as sign of the slope and c (1) is set as a point of reference and stored as a PRIgang.

Insert missing timestamp and padding the data gaps with zero vectors in trignPRI to ensure coherence in Choice of ch = 1 (parameter X):

• 16. Determination of the number of missing pulses d by element-wise subtraction of the time stamp data, followed by division through the PRI and rounding. By finding all values with d> 1, all can Pulse numbers dn are determined after which at least one pulse is missing.
• 17. Extraction of the area from the time stamps, the considering from NPRIx (parameter XI) the tolerance limit for missing pulses at both edges of the timestamp vector.
• 18. Paste the missing timestamp times through interpolation between the already existing time stamps and incorporation of zero vectors to the corresponding places in the signal data set, which is Updated versions of trignPRIn and xt revealed.

Determination of the optimal timestamp vector:

• 19. Subtraction of step 7 PRI and the timestamp median of trignPRIn to calculate trigpp.
• 20th Division of all Timestamp differences trigpp through the sampling interval dsam and Subtract the calculated correction step PRIgang from step 15 for obtaining sampling differences df, which no longer show a trend.

Calibration of all sampling differences the earliest Arrival time:

• 21. Determination of the pulse Npuls with the smallest Sample in the optimized range of df.
• 22. Subtract the sample of Npuls from all sample differences df.
• Shifting the sampling differences of all pulses to a set Pretrigger to consider:
• 23. finding the sampling point sp in the pulse Npuls, in which the manually set trigger threshold tsbit (parameter XII) exceeded becomes.
• 24. Correction of all the sampling differences df by the pretrigger ptsam (parameter VI) and the displacement sp occurring in Npuls, so that the final form of df arises ( 10 ).

Alignment of the pulses:

• 25. Application of the correction vector df determined in step 24 to the signal data by shifting each pulse by the number of sampling intervals stored in df. For this purpose, a subtraction with subsequent application of a modulo function is performed. The size of the recording window Nsam · dsam remains unchanged. In order to avoid that due to the modulo function sampling points from the end of a pulse appear again at the beginning of the recording window, the resulting sample jumps of each pulse are logged by means of a maximum interrogation. Subsequently, the surplus part is set to zero. The result is a matrix of fully aligned pulses xt ( 11 ).

The Steps 6-25 be for go through each PRI found in the record once.

to Control of the processing level is the functioning algorithm in the programming language of Mathworks Matlab attached below (filename: profilausrichtung_npri_patent.m).

• trigh

Zeitstempel nach Anwendung eines Medianfilters

• sps

Pulsnummern der Sprungstellen zwischen wechselnden PRIs

• spsn

Anzahl unterschiedlicher PRIs im Zeitstempeldatensatz (= j)

• trignPRI

Vektor mit Zeitstempeln, die eine konstante PRI aufweisen (einer von j Blöcken)

• trigp

Zeitstempeldifferenzen, Entstehung nach Subtraktion einer PRI von trignPRI

• df1

Vektor mit Auswahl der Abtastwerte, an denen die Optimierung durchgeführt wird

• vorz

Steigung des Trends der extrahierten Zeitstempel df1

• D

Steigung zwischen den Medianen des zweiten und drittem Viertel von df1

• bP

Geradenfunktion, deren Steigung in der Optimierungsschleife verändert wird

• c

detektierte optimale Steigungen in erster Optimierungsschleife

• h

elementweise Differenzen zwischen den Werten der optimalen Steigungen c

• c2

detektierte optimale Steigungen in zweiter Optimierungsschleife

• PRIgang

errechneter optimaler Korrekturgang zur Eliminierung des Zeitstempeltrends

• d

Vektor mit Anzahl von fehlenden Pulsen im Datensatz

• dn

Pulsnummern, nach denen mindestens ein Puls fehlt

• trigpp

Zeitstempeldifferenzen nach Einfügung fehlender Pulsankunftszeiten

• df

Korrekturvektor, um den alle Signaldaten verschoben werden

• Npuls

Pulsnummer mit kleinstem Verschiebungswert nach Subtraktion der PRI

• sp

erster Abtastpunkt im Puls Npuls, der die Triggerschwelle überschreitet

In addition to the input variables I-VI and control parameters VII-XII, further essential parameters that occur in the course of the algorithm are summarized in the following list:

• trigh

Timestamp after applying a median filter

• sps

Pulse numbers of jumps between changing PRIs

• spsn

Number of different PRIs in the timestamp data record (= j)

• trignPRI

Vector with timestamps that have a constant PRI (one of j blocks)

• trigp

Timestamp differences, origin after subtraction of a PRI from trignPRI

• df1

Vector with selection of the samples on which the optimization is performed

• ex

Slope of the trend of the extracted timestamp df1

• D

Increase between the medians of the second and third quarters of df1

• bP

Line function whose slope is changed in the optimization loop

• c

detected optimal slopes in the first optimization loop

• H

element-by-element differences between the values of the optimal slopes c

• c2

detected optimal slopes in the second optimization loop

• PRIgang

Calculated optimal correction gear to eliminate the timestamp trend

• d

Vector with number of missing pulses in the record

• dn

Pulse numbers after which at least one pulse is missing

• trigpp

Timestamp differences after insertion of missing pulse arrival times

• df

Correction vector around which all signal data are shifted

• Npuls

Pulse number with lowest shift value after subtraction of the PRI

• sp

first sampling point in the pulse Npuls, which exceeds the trigger threshold

Claims (6)

Method, time-fluctuating signals, the from a transmission source with constant or only in sections constant pulse repetition frequency and emitted by a receiver device were so aligned that the originally coherent behavior the radiated signals is restored, the method includes: a) detecting the amplitude-triggered signal data with the corresponding ones Timestamp marks of the measurement release and the sampling interval used; b) determine the pulse repetition interval from the present time stamps; c) Extraction of the process data from the total amount of timestamps; d) starting from the beginning certain pulse repetition interval, in an optimization loop, from the specific timestamp values, according to which artificial coarsening, a gradually changing one Deduct pulse repetition interval, wherein for each loop pass a new histogram over the samples are formed and the maximum count is stored, until the optimal pulse repetition interval with the finding of the global Maximums of the maximum counts all calculated histograms are found, e) of the so found Pulse repetition interval, subtracted from all timestamps, the remaining ones Time differences in the corresponding number of sampling intervals to convert and move the signals to that extent. Correction device for signals, at least existing from conventional construction stages of a computer system with input and output means for Signals for execution A method according to claim 1, characterized by, signal acquisition stage, Memory units, data preparation and post-processing modules, input unit for control parameters, Processing stage with a program control for signal processing of time-fluctuating signals, according to the method of claim 1 and an output unit for signal data display. Storage medium for Computer program data with program data stored therein, wherein the majority of program data when controlling a computer, causing the computer to carry out the method of claim 1. Use of the method according to claim 1, for preprocessing of radar data for the Doppler evaluation in a synthetic aperture radar (SAR) and in the signal analysis parasitically recorded radar signals, the one radar receiver system in a multistatic geometry. Use of the method according to claim 1, for the determination the optimal pulse repetition interval, or the optimal pulse repetition frequency, which the timestamp is the cheapest timed to be in a comparison with the originally sent Pulse repetition interval, the relative temporal drift between transmitter source and receiver system to determine. Use of the method according to claim 1, for coherent summation of received signals from a remote signal source were radiated to opposite classical methods with a low number of recorded Signals an increase the signal-to-noise ratio to effect and thus faster identification of the signal parameters to reach.