WO2014155038A1

WO2014155038A1 - Separating an unknown radar signal out from multiple unknown radar signals

Info

Publication number: WO2014155038A1
Application number: PCT/GB2014/000113
Authority: WO
Inventors: James Peter RICHES
Original assignee: UK Secretary of State for Defence
Current assignee: UK Secretary of State for Defence
Priority date: 2013-03-26
Filing date: 2014-03-25
Publication date: 2014-10-02
Anticipated expiration: 2015-09-26
Also published as: GB201305481D0; GB201405269D0; GB2512497B; WO2014155038A8; GB2512497A

Abstract

There is provided a method of separating an unknown radar signal (Src_A) out from multiple unknown radar signals (SRC_A, Src_B) and a radio signal separator (10) configured to perform the method. The method comprises receiving the multiple unknown radar signals at spatially different antennas (RX1, RX2T RX3), and separating the unknown radar signal out from the multiple unknown radar signals based upon the time difference of arrival (ΤDOA) of the unknown radar signal at the spatially different antennas

Description

SEPARATING AN UNKNOWN RADAR SIGNAL OUT FROM MULTIPLE

UNKNOWN RADAR SIGNALS

Technical Field of the Invention

The invention relates to a method and a radar signal separator for separating an unknown radar signal out from multiple unknown radar signals.

Background to -the Invention

There is a desire to separate unknown radar signals out from one another, so that the unknown radar signals can be individually analysed and identified. In dense radar environments with multiple radar transmitters, it can be difficult to separate different radar signals out from one another.

A radar signal typically comprises a train of pulses that are transmitted by a radar transmitter. The identification of which pulses relate to which radar transmitters, so that the radar signal of any particular radar transmitter can be reconstructed by selecting just the pulses relating to that particular radar transmitter, is a first step in identifying unknown radar transmitters.

Since the radar signals are unknown, it is not possible to separate a given radar signal out from other radar signals by searching for known characteristics of the given radar signal, for example in a manner similar to that done for separating known GPS system signals out from one another.

A known technique for separating unknown radar signals out from one another comprises separating radar pulses according to their radio frequency. However, maritime navigation radars may make up the majority of signals and these typically operate on very similar radio frequencies. Other known techniques include separating radar pulses based upon the pulse width and/or bearing if available. Other approaches have been to use artificial intelligence techniques and graph theory. Another known technique comprises the use of Pulse Repetition Interval (PRI) to separate radar signals from one another, by assuming that radar pulses which occur at fixed time intervals are likely to belong to the same radar signal, or at least to belong^" to a smaller group of radar signals than all of the radar signals that are present. However, the radar pulses of a radar signal are sometimes jittered in time, which can make detection of the jittered radar signal more difficult.

US Patent No 5, 583,505 discloses a receiver that uses PRI techniques and which is designed to cope with jittered pulses, however when many different radar signals are present, separation of individual radar signals even when using these techniques becomes difficult, particularly when different radar signals have similar PRIs.

It is therefore an aim οΐ the invention to provide an improved way of separating unknown radar signals from one another.

Summary of the Invention

The Time Of Arrival (TO A) for a radar signal at an antenna is a function of the relative distance from the radar transmitter to the antenna. Therefore a radar signal from a particular radar transmitter will arrive at spatially different antennas at different times. The time difference between a radar signal arriving at one antenna and the same radar signal arriving at another antenna is referred to herein as a Time Difference Of Arrival (TDOA).

When radar signals are sent from different radar transmitters that have different positions to one another, the radar signals normally have different TDOA's from one another, due to the differing radar transmitter positions. Therefore, it is possible to distinguish different radar signals from one another based upon the TDOA of the radar signals at spatially different antennas.

Therefore, according to a first aspect of the invention, there is provided a method for separating an unknown radar signal out from multiple unknown radar signals. The method comprises receiving the multiple unknown radar signals at spatially different antennas, and separating the unknown radar signal out from the multiple unknown radar signals based upon the time difference of arrival (TDOA) of the unknown radar signal at the spatially different antennas.

Preferably, the signals received from the spatially different antennas are compared to one another by performing a cross-correlation between them to determine a correlation peak. The correlation peak designates the TDOA of radar pulses within the received signals, the radar pulses responsible for producing the correlation peak. The parts of the received signals that correlate with one another at this correlation peak may then be extracted to construct the radar signal that produced the radar pulses.

The cross-correlation may produce multiple correlation peaks, and each peak may be reviewed for constructing a radar signal from the parts of the received signals that correlate with one another at the correlation peak. Since signals transmitted from different transmitters will have different TDOA's at the spatially different antennas, each correlation peak should correspond to a particular transmitter, and therefore a particular signal, such as a radar signal.

Since radar signals generated from spatially different sources are separated from one another based upon the fact that the.y have different. TDOAs, rather than that they have any particular values of TDOA, the received signals recorded at the spatially different antennas do not need to be time-synchronised with one another. This is because time offsets affect all of the TDOAs, and do not typically cause one TDOA to match another TDOA when the TDOAs would not have matched one another had the received signals been all recorded at exactly the same time. Accordingly, the method can be easily implemented without any requirement for complex and/or expensive time synchronisation methods between the received signals.

Advantageously, the receiving of the multiple unknown radar signals at spatially different antennas may comprises receiving a first RF time segment taken from a first antenna of the spatially different antennas; and receiving a second RF time segment taken from a second antenna of the spatially different antennas. The first RF time segment is a recording of the RF signal received at the first antenna during a first time segment, and the second RF time segment is a recording of the RF signal received at the second antenna during a second time segment. Hence, only two spatially different antennas are required to separate an unknown radar signal out from multiple unknown radar signals. Whilst two different radar transmitters could theoretically produce the same TDOA at the_. two antennas if the two different transmitters were both located in positions producing the same TDOA, the vast majority of transmitter positions will produce different TDOA's, and so this is considered unlikely to occur in practice. Even if this does occur, a particular signal that is identified as having a particular TDOA, will normally comprise many less radar signals to be separated than all the radar signals in the first and second RF time segments. In such a case, the method may further comprise additional processing steps to separate the unknown radar signal out from the particular signal, such that the separation of the unknown radar signal from the multiple unknown radar signals is based upon the known techniques, as well as being based upon the TDOA of the unknown radar signal at the first and second antennas.

The calculation of the TDOA may comprise calculating a cross-correlation between the first RF time segment and the second RF time segment; and identifying a peak in the calculated cross-correlation, the peak designating the TDOA of the unknown radar signal at the first and second antennas.

The first and second time segments preferably overlap one another in time so that the cross- correlation can identify a true TDOA between the first and second segments rather than just an aliased TDOA, which may occur when later radar pulses of a radar signal correlate to earlier radar pulses of the radar signal, if there is very little variation between the radar pulses of the radar signal.

If every individual pulse emitted from the radar transmitter is unique, for example if the radar signal contains jittered pulses, then there are unlikely to be any significant peaks in the correlation that correspond to aliased TDOAs, improving the efficacy of the method.

The TDOA of the unknown radar signal at the first and second antennas may be a timeshift of the second RF time segment from the first RF time segment at which the first RF time segment and the second RF time segment correlate with one another. The separating the unknown radar signal out from the multiple unknown radar signals may further comprise calculating an intersection of the first RF time segment and the second RF time segment when the second RF time segment is shifted from the first RF time segment by the timeshift; and outputting the calculated intersection as the unknown radar signal.

The intersection operation outputs the signal parts that are present in both the first RF time segment and the second RF time segment when the second RF time segment is shifted from the first RF time segment by the timeshift, and may for example be calculated by multiplying the first RF time segment and the second RF time segment when the second RF time segment is shifted from the first RF time segment by the timeshift.

The second antenna is sufficiently remote from the first antenna to give measurable differences between the time of arrival of a pulse at the first antenna and the time of arrival of the pulse at, the second antenna. Whether a time difference is measurable or not, and therefore, how remote the first and second antennas need to be, will clearly depend upon the sampling rates of the first and second RF time segments. For example, the higher the sampling rate, the closer together the first and second antennas could be positioned.

To help avoid non-radar transmitters from generating correlation peaks, the frequency range of each of the first and second RF time segments may be restricted to cover expected radar signal frequencies, prior to the step of calculating a cross-correlation between the first RF time segment and the second RF time segment. Then, there will be less correlation peaks that need to be reviewed.

Advantageously, the receiving the multiple unknown radar signals at spatially different antennas may comprise receiving a third RF time segment taken from a third antenna of the spatially different antennas. Then, the calculating a cross-correlation between the first RF time segment and the second RF time segment comprises calculating the cross-correction between the first RF time segment, the second RF time segment and the third RF time segment. Therefore, a correlation peak on the three dimensional surface resulting from the cross-correlation will designate a TDOA of the unknown radar signal at the first and third antennas, in addition to the TDOA between the first and second antennas. The TDOA between the first and second antennas may also be determined.

The TDOA of the unknown radar signal at the first and third antennas may be a timeshift of the third RF time segment from the first RF time segment at which the third RF time segment correlates with the first and second RF time segments, the second RF time segment being shifted from the first RF segment by the timeshift of the second RF time segment from the first RF time segment.

The step of calculating an intersection may comprise calculating the intersection of the first RF time segment, the second RF time segment, and the third RF time segment, when the second RF time segment is shifted from the first RF time segment by the timeshift of the second RF time segment from the first RF time segment,. and the third RF time segment is shifted from the first RF time segment by the timeshift of the third RF time segment from the first RF time segment.

Using three RF time segments from three spatially different antennas helps to separate radar signals that have the same TDOA to both the first and second antennas, for example when two different radar signal transmitters both lie on a hyperboloid surface that has its foci at the first and second antennas, as will be apparent to those skilled in the art.

Advantageously, a position from which the unknown radar signal originates may be determined based upon the timeshift of the second RF time segment from the first RF time segment, the timeshift of the third RF time segment from the first RF time segment, the relative times at which the first, second, and third RF time segments were taken from the first, second, and third antennas respectively, and the positions of the first, second, and third antennas.

The relative times at which the RF time segments were taken from the antennas may be determined according to respective timestamps associated with each one of the RF time segments, or the antennas may all be configured to begin recording at exactly the same time as one another, for example by receiving a start command from a transmitter that is equidistant from all of the antennas.

The method may further comprise identifying a further peak in the calculated cross- correlation, the further peak designating a TDOA of a further unknown radar signal at the first and second antennas. A further intersection may be calculated to separate the further unknown radar signal from the received signals. According to a second aspect of the invention, there is provided a radar signal separator for separating an unknown radar signal out from multiple unknown radar signals. The radar signal separator is configured to receive the multiple unknown radar signals from spatially different antennas; and separate the unknown radar signal out from the multiple unknown radar signals based upon the time difference of arrival (TDOA) of the unknown radar signal at the spatially different antennas.

The radar signal separator may be configured to receive the multiple unknown radar signals from spatially different antennas by receiving a first RF time segment taken from a first antenna; and receiving a second RF time segment taken from a second antenna that is remote from the first antenna, and may be further configured to separate the unknown radar signal out from the multiple unknown radar signals based upon the TDOA of the unknown radar signal at the spatially different antennas by:

- calculating a cross-correlation between the first RF time segment and the second RF time segment;

- identifying a peak in the calculated cross-correlation, the peak designating a timeshift of the second RF time segment from the first RF time segment at which the first RF time segment and the second RF time segment correlate with one another;

- calculating an intersection of the first RF time segment and the second RF time segment when the second RF time segment is shifted from the first RF time segment by the timeshift; and

- outputing the calculated intersection as the unknown radar signal.

The radar signal separator may be configured to perform the methods discussed in relation to the first aspect. The phrase "spatially different antennas" refers to antennas that are at different positions to one another.

Preferably the computational burden of the method is reduced by converting the radar signals (or more typically the RF segments of interest) from a detailed waveform (e.g. a raw waveform or an envelope, which may include sampling at or near to radar frequencies of interest) to a simplified representation. The simplified representation preferably contains information on the timing of radar pulses. This typically includes the leading and/or trailing edges of the (each) waveform, and may substantially comprise a list of the leading and/or trailing edges Of pulses in the (each) waveform. Optionally the leading edges are used, optional l y the trai ling edges are used, and preferably both are used. A lternativel y or additional ly it may include information on the mid-point of the radar pulses, or any other representative timing indicator (fourth rising edge of each pulse, three-quarter point of each pulse etc). It should not matter if a consistent offset is included and so the simpl ified representation may include the timings of each radar pulse plus a constant (E.g. 1 ms) although in practice there should be no particular benefit in doing so.

Performing the cross-correlation on the si mplified representation enables a reduction in the computat ional burden of the operation, as compared to performing the cross-correlation on a raw waveform. This may enable the processing method to omit the step of recording a raw waveform in the first place. Due to the high frequencies used in radar transmissions, recording raw waveforms is technically chal lenging, and cross correlating the raw waveforms qu ickly (e.g. in real time) requires considerable processing power which is expensive, so the use of a s impl ified representation may reduce the cost, and perhaps also the power requirement of an associated computer processor.

When the cross correlation is performed on a s implified representation, a peak in the identified output is used to identify a TDOA of one of the signals, and this is used to identify leading edges and/or trail ing edges of pulses in the signals (or RF segments of interest) which have a TDOA that match the identified TDOA, and to construct one of the original signals ( i.e. in this case a simplified representation thereof including information on the leading and/or trailing edges of pulses in the radar s ignal ) which can be used to identify the type of radar transmitter and/or the status of the radar transmitter. Techniques for identifying the type of radar transmitter and/or its status are known in the art and are outside the scope of th is document, however as an example a pulse repetition interval, or indeed a whole pattern known to be included in certain types of radars, or radars in certain operating modes may be identified in the reconstructed radar signal.

Therefore advantageously in the method of the first aspect the first and second RF time segments may each comprise a simplified representation of respectivel y associated radio frequency waveform time segments, each simplified representation including summary information including the timings of radar pulses of the respectively associated radio frequency waveform time segments (the timings being at least relative timings - i.e. relative to other pulses in the same RF time segment). More specifically the first and second R F time segments may each include information on the ti mings (typicall y the relative ti m i ngs) of radar pulses of ( i.e. that were derived from) respectively associated radio frequency waveforms, and substantially exclude such associated radio frequency waveforms ( i.e.

omitting waveform sampling data at or near the wavelength of the carrier frequency of the radar signals). Optionally this is preceded by the step of extracting information on the timings of radar pulses of a radio frequency waveform of the first and second RF time segments, or alternatively the simpl ified representation is received and accordingl y processed.

Brief Description of the Draw ings

Il lustrat ive embodiments of the invention will now be described by way of example only and with reference to the accompanying drawings, in which:

Fig. 1 shows a schematic diagram of two unknown radar transmitters and two spatial ly different antennas according to first and second illustrative embodiments of the invention;

Fig. 2 shows a timing diagram of a first radar signal emitted from a first one of the unknown radar transmitters of Fig. 1 ;

Fig. 3 shows a timing diagram of a second radar signal emitted from a second one of the unknown radar transmitters of Fig. 1 ;

Fig. 4 shows a timing diagram of the radar signals received at a first one of the spatial ly different antennas of Fig. 1 ;

Fig. 5 shows a timing diagram of the radar signals received at a second one of the spatial ly different antennas of Fig. 1 ;

Fig..6 shows a diagram of a cross-correlation between the radar signals received at the first one of the spatially different antennas and the radar signals received at the second one of the spat ially different antennas;

Fig. 7 shows a diagram of the calculat ion of the intersection between the radar s ignals received at the first one of the spat iall y different antennas and the radar signals received at the second one of the spatially different antennas when shifted according to a first correlation peak corresponding to the first radar signal ;

Fig. 8 shows a diagram of the calculation of the intersection between the radar signals received at the first one of the spatially different antennas and the radar signals received at the second one of the spatial ly different antennas when shifted according to a second correlation peak corresponding to the second radar signal ;

Fig. 9 shows a timing diagram of a first j ittered radar signal emitted from the fi rst one of the unknown radar transmitters of Fig. 1 ;

Fig. 10 shows a timing diagram of a second j ittered radar signal emitted from the second one of the unknown radar transmitters of Fig. 1 ;

Fig. 1 1 shows a diagram of a cross-correlation between the j ittered radar signal s received at the first one of the spatially di fferent antennas and the j ittered radar signals received at the second one of the spatiall y different antennas;

Fig. 12 shows a schematic diagram of an unknown radar transmitter and three spatially different antennas according to a third illustrative embodiment of the invention; and

Fig. 1 3 shows a schematic diagram of a radar signal separator system according to the third illustrati ve embodiment.

The drawings are for illustrative purposes onl y and are not to scale.

Detailed Description

A first il lustrative embodiment of the invention will now be described with reference to Figs. 1 - 8. Fig. 1 shows a schematic diagram of first and second unknown radar transmitters Scr_A and Src_B respectively, and first and second spatially different antennas R l and Rx2 respectively.

As shown in Fig. 1 , the first and second unknown radar transmitters, and the first and second spatial ly different antennas, are all located at different positions to one another.

S ince the antennas RX 1 and RX2 are at different positions to one another, a radar signal from the first unknown radar transmitter takes di ffering amounts of t i me to reach the antennas RX 1 and RX2. Specifically, the Scr_A radar signal takes 300p s to reach the antenna RX 1 and takes 67C^ s to reach the antenna RX2. The Time Difference Of Arri val (TDOA) of the Scr_A radar signal between the first and second antennas is therefore 370p s. Furthermore, since the first and second unknown radar transmitters are also at different positions to one another, the Scr_B radar signal takes 500p s to reach the antenna RX 1 and takes 360p s to reach the antenna RX2. The Time Difference Of Arrival (TDOA) of the Scr_B radar signal between the first and second antennas is therefore - 140p s.

S ince the radar signals travel at the speed of light c, the antenna RX 1 is a distance of 300p s.c from the first unknown radar transmitter Scr_A, and 500p s.c from the second unknown radar transmitter Scr_B . The antenna RX2 is a distance of 670ps.c from the first unknown radar transmitter Scr_A and 360ps.c from the second unknown radar transmitter Scr_B .

The timing diagram of Fig. 2 shows the Src_^A radar signal as comprising multiple pulses RP_A. The pulses RP_A are 20ps in width, and are repeated at a pulse repetition interval (PRI) of 450p s. The timing diagram of Fig. 3 shows the Src_B radar signal as comprising multiple pulses RP_B. The pulses RP_B are also 20p s in width, but are repeated at a pulse repetition interval (PRI) of 633p s. The first radar pulses RP_A and RP_B are both emitted from the radar transmitters Src_A and Scr_B at a time T = 0.

The timing diagram of Fig. 4 shows a first RF time segment taken from the antenna RX 1 , the first RF time segment running from the time T = O s to T = lOOOOp s. S ince the antenna RX 1 is 300p s.c from the radar transmitter Scr_A, the first pulse RP_A arrives at T = 300p s, and since the antenna RX 1 is 500p s.c from the radar transmitter Scr_B, the first pulse RP_B arrives at T = 500p s. Throughout the first RF time segment The pulses RP_A are received every 450p s, and the pulses RP_B are received every 633p s.

The timing diagram of Fig. 5 shows a second RF time segment taken from the antenna RX2, the second RF time segment also running from the time T = Op s to T = l OOOOp s. Since the antenna RX2 is 670p s.c from the radar transmitter Scr_A, the first pulse RP_A arrives at T = 670p . and since the antenna RX2 is 360p s.c from the radar transmitter Scr_B , the first pulse RP_B arrives at T = 360ps. Throughout the second RF time segment, the pulses RP_A are received every 450p s, and the pulses RP_B are received every 633p s.

Next, the first and second RF time segments are cross-correlated with one another to identify correlation peaks that may correspond to radar signals being received at the first and second antennas at different times to one another. The result of the cross-correlation is shown in Fig. 6. The cross-correlation runs from -250(^s to -25()0 s, as the second RF ti me segment is slid past the first RF time segment from -2500p s relative to the first RF time segment to +2500M S relative to the first RF time segment.

The large correlation peak CP_A at -370 s indicates that the first RF time segment and the second RF time segment con elate wel l with one another when the second time segment has a time shift of -370p s from the first RF time segment. This corresponds to the TDOA of the Scr_A radar signal , the radar transmitter Scr_A being 370 s.c closer to the first antenna RX l than the second antenna RX2.

The correlation peaks at -820 s, -8C^ s, and 530 s correspond to aliased TDOA' s of the Scr_A radar signal , when earlier pulses RP_A correlate well to later pulses RP_A of the Scr_A radar signal . The correlation peaks of -820μ $, - 70μ 8, -80μ ς, and 530μ , occur at intervals matching the pulse repetition interval of 45C^ s of the Src_A radar signal.

The large correlation peak CP_B at 140μ $ indicates that the first RF time segment and the second RF time segment correlate wel l with one another when the second t ime segment has a time shift of 14C^ s from the first RF time segment. This corresponds to the TDOA of the Scr_B radar signal, the radar transmitter Scr_B being 14(^ s.c further from the first antenna RX l than the second antenna RX2.

The correlation peaks at -49^ s, and 773 s correspond to al iased TDOA's of the Scr_B radar signal, when earl ier pulses RP_B conelate well to later pulses RP_B of the Scr_B radar signal. The conelation peaks of -493 s, 14C^ s, and 773μ , occur at intervals matching the pulse repetition interval of 633μ s of the Src_B radar signal.

The correlation peaks relating to the pulses RP_A are higher than the correlation peaks relating to the pulses RP_B due to the higher pulse repetition rate, and therefore the greater the number of pulses that correlate between the RF time segments, of the Src_A radar signal.

Once the cross-correlation has been calculated, a correlation peak is chosen for reconstructing the signal that is responsible for the correlation peak. Typical l y, the h ighest correlation peak is chosen first. The signal is reconstructed by performing an intersect ion between the first and second RF time segments when shifted according to the correlation peak. The diagram of Fig. 7 shows the calculation of an intersection INT1 between the first and ^■ second RF time segments when the correlation peak at -370 s is chosen. Specifically, the second RF time segment is shifted -370μ * relative to the first RF time segment, and the first and second RF time segments are multipl ied together to produce the intersection INT 1 , which coiTesponds to the radar signal from the unknown radar transmitter Src_A. The dashed l ines on Fig. 7 show how the pulses RP_A are al igned with one another at the ti meshift of -370 s. and therefore resul t in the pulses RP_A being reconstructed in the intersection 1NT 1 .

Once the Src_A radar signal has been reconstructed, and identified as having a pulse repetition interval of 450p s, the correlation peaks that appear in the Fig. 6 cross-correlation at intervals of 450p s can be discounted as aliases. Had a different one of the -820 s, -37C^ s, - 80p s, and 530 s correlation peaks been chosen first, then the Src_A radar signal wou ld sti l l have been reconstructed, albeit with earlier-transmitted pulses multiplied with later- transmitted pulses to yield the reconstructed radar signal . This should not lead to a significantl y different result; if the earlier-transmitted pulses were significantly different to the later-transmitted pulses, then the correlation peak would not have been produced in the first place.

The diagram of Fig. 8 shows the calculation of an intersection INT2 between the first and second R F time segments when the next correlation peak at \40μ $ is chosen. Specifical l y, the second RF time segment is sh ifted 140p s relative to the first RF time segment, and the first and second RF time segments are multipl ied together to produce the intersection 1NT2, which corresponds to the radar signal from the unknown radar transmitter Src_B. The dashed lines on Fig. 7 show how the pulses RP_B are al igned with one another at the timeshift of 140 s, and therefore result in the pulses RP_B being reconstructed in the intersection INT2.

Once the Src_B radar signal has been reconstructed, and identified as having a pulse repetition interval of 633μ $, the correlation peaks that appear in the Fig. 6 cross-correlation at intervals of 633 s can be discounted as aliases. Had a different one of the -493 s, 140MS, and. 773 s correlation peaks been chosen first, then the Src_B radar signal wou ld sti ll have been reconstructed, albeit with earlier-transmitted pulses multipl ied with later-transmitted pulses to yield the reconstructed radar signal . This should not lead to a significantly different result; if the earlier-transmitted pulses were significantl different to the later-transmitted pulses, then the correlation peak would not have been produced in the first place.

The reconstructed radar signals INT 1 and INT2 can then be passed to any subsequent processing stages to further characterise them if necessary. Although in this embodiment the first and second RF time segments were both synchronised to start at time T = 0, a difference between the starts of the first and second RF time segments would simply be reflected by a difference in the timing of the correlation peak, and would not therefore have any significant effect upon the reconstructed signal,

A second i l lustrative embodiment of the invention will now be described with reference to Figs. 1 and 9 - 1 1 . The second embodiment uses the same configuration of first and second unknown radar transmitters and first and second antennas as the first embodiment, however the radar s ignals that are transmitted differ from those of the first embodiment.

The tim ing diagram of Fig. 9 shows a j ittered radar signal Src_AJ that is transmitted from the first unknown radar, transmitter Src_A. The j ittered radar signal Src_AJ comprises multiple pulses that have a nominal pulse repetition interval of 450p s, but which are j ittered by pl us or minus SOp s.

The timing diagram of Fig. 10 shows a jittered radar signal Src_BJ that is transmitted from the second unknow n radar transmitter Src_B . The j ittered radar signal Src_BJ comprises multiple pulses that have a nominal pulse repetition interval of 633p s, but which are j ittered by plus or minus 50p s.

The j ittered radar signals Src_AJ and Src_BJ are received at the first and second antennas RX 1 and RX2, and first and second RF time segments are taken therefrom. Fig. 1 1 shows a cross-correlation of these first and second RF time segments, and it can be seen that the cross-correlation only produces two peaks, one peak CP_AJ corresponding to the radar signal Src_AJ at a TDOA of -370 S, and one peak CP_BJ corresponding to the radar signal Src_BJ at a TDOA of 140 s. The aliased correlation peaks that were present in Fig.6 have all disappeared due to the jittering of the pulses, and the number of radar signals can be directly counted according to the number of correlation peaks. A third embodiment of the invention wi l l now be described with reference to Figs. 12 and 1 3.

Fig. 1 2 shows a schematic diagram of an unknown radar transmitter Scr_C, and first, second, and third spatially different antennas RXA and RXB , and RXC. The three antennas each receive a radar signal from the unknown radar transmitter Src_C. The three antennas also receive radar signals from other unknown radar transmitters (not shown in Figs ).

Fig. 1 3 shows a schematic diagram of a radar signal separator system which comprises the three antennas RXA and RXB , and RXC, a radar signal separator 10, and a radar signal analyser 20. The three antennas RXA and RXB, and RXC are connected to the radar signal separator 10 at three respective inputs I_RXA, I_RXB . and I_RXC.

In use, the radar s ignal separator 10 records three RF time segments from the antennas RXA and RXB , and RXC. The connections between the antennas and the radar signal separator are of different lengths, and the radio signal separator 10 automatically offsets the recorded RF time segments from one another according to the differences between the lengths of the connections, such that the recorded RF time segments are all synchronised with one another in t ime.

The radar signal separator 10 calculates the cross-correction between the first RF time segment, the second RF time segment, and the third RF time segment, and a correlation peak designates a TDOA of the Scr_C radar signal between the first and second antennas RXA and RXB, and a TDOA of the Scr_C radar signal between the first and third antennas RXA and RXC.

The radar signal separator 10 then calculates the intersection between the first RF time segment, the second RF time segment, and the third RF time segment when the RF time segments are aligned according to the correlation peak, which yields the reconstructed Scr_C radar signal. The reconstructed Scr_C radar signal is output at an output OP to the radar signal analyser 20 for analysis.

S ince three antennas are present rather than only two, the TDOA' s can be used to determine the position of the unknown radar transmitter Scr_C by multi-lateration. Specifically, the TDOA of the Scr_C radar signal between the first and second antennas RXA and RXB defines a first hyperbola HYP 1 along wh ich the unknown radar transmitter Scr_C must lie, and the TDOA of the Scr_C radar signal between the first and third antennas RXA and RXC defines a second hyperbola HYP2 along which the unknown radar transmitter Scr_C must lie. The intersection of the first and second hyperbolas gives the location of the unknown radar transmitter Scr_C.

As an alternative to synchronising the first, second, and third RF time segments in time according to the lengths of the connections between the first, second, and third antennas and the radio signal separator 10, the RF signals may be recorded at the locations of the first, second, and third antennas, and be sent to the radio s ignal separator 10 together with timestamps indicating at which times the RF time segments were recorded.

Note that figures 2 to 10 predominantly depict an embodiment where instead of the method involving the recording or processing of the raw RF waveform, the method is appl ied to a simplified representation of the RF waveform, including information on the (at least relative) timing of the pulses arriving at each recei ver. Figures 2 to 10 are described above specificall y recording the timing of leading and trai ling edges, and this being embodied in an envelope waveform representation, however it wil l be apparent that any timing measurement relevant to the pulses can be used if used consistently, such as the midpoint of each pulse, and it wi l l be apparent that the calculation of performing the cross correlation need not involve a graphical representation, nor even an envelope waveform data fi le, and instead a list of the timings may be sufficient.

Further embodi ments fall ing within the scope of the appended claims will also be apparent to those skil led in the art.

Claims

1 . A method of receiving multiple unknown radar signals at spatially different antennas, the multiple unknown radar signals being from d ifferent radar transmitters that have different positions to one another such that the radar signals have different time difference of arrival (TDOA ) from one another with respect to the spatially different antennas, and separating and extracting an unknown radar s ignal out from the mul tiple unknown radar s ignals, the method comprising:

receiving the multiple unknown radar signals at spatially different antennas, including receiv ing a first RF time segment taken from a first antenna of the spatially different antennas and receiving a second RF time segment taken from a second antenna of the spatiall y different antennas, and

separat ing the unknown radar signal out from the multiple unknown radar signals based upon the ti me difference of arrival (TDOA) of the unknown radar signal at the spat ial l y different antennas, including:

calculating a cross-correlation between the first RF t ime segment and the second RF time segment, and

identifying a peak in the calculated cross-correlation, the peak designating a TDOA of the unknown radar signal at the first and second antennas, and,

extracting the parts of the recei ved signals that correlate with one another at the identified correlation peak to construct a radar signal,

wherein, the first and second RF time segments each comprise a simpl ified representation of respectively associated radio frequency waveform time segments, each simplified representation including summary information including timings of radar pulses of the respectively associated radio frequency waveform time segments.

2. The method of clai m 1 , wherein the TDOA of the unknown radar signal at the first and second antennas is a timeshift of the second RF time segment from the first RF time segment at which the first RF time segment and the second RF time segment correlate with one another, and wherein the separating the unknown radar signal out from the multiple unknown radar signals further comprises: - calculating an intersection of the first RF time segment and the second RF time segment when the second RF time segment is shifted from the first RF ti me segment by the timeshift; and

- outputting the calculated intersect ion as the unknown radar signal .

3. The method of claim 2, wherein the cal culat ing an intersection of the first R F t ime segment and the second RF time segment comprises calculating the product of the first RF time segment and the second RF time segment .

4. The method of claim 1 ,2 or 3 further comprising restricting a frequency range of each of the first and second RF time segments to cover ex pected radar signal frequencies, prior to the step of calculating a cross-correlation between the first RF time segment and the second RF time segment.

5. The method of claim 1 , 2, 3 or 4 wherein:

- the receiving the multiple unknown radar signals at spatially different antennas comprises receiving a th ird RF time segment taken from a third antenna of the spat ial ly different antennas;

- the calculating a cross-correlat ion between the first RF time segment and the second RF time segment comprises calculating the cross-correction between the first RF time segment, the second RF t ime segment and the third RF time segment; and

- the peak additionally designates a TDOA of the unknown radar signal at the first and third antennas.

6. The method of claim 5, wherein the TDOA of the unknown radar signal at the first and th ird antennas is a ti meshift of the third RF time segment from the first RF time segment at which the third RF time segment correlates with the first and second RF time segments.

7. The method of claim 6, further comprising determining a position from which the unknown radar signal originates based upon the timeshift of the second RF time segment from the first RF time segment, the timeshift of the third RF time segment from the first RF time segment, the relative times at which the first, second, and third RF time segments were taken from the first, second, and third antennas respectively, and the positions of the first, s'econd, and third antennas.

8. The method of claim 7, wherein the relative times at which the RF time segments were taken from the antennas are determined according to respective timestamps associated with each one of the RF time segments.

9. The method of claim 5, 6. 7 or 8 wherein the step of calculating an intersection comprises calculating the intersection of the first RF time segment, the second RF t ime segment, and the third RF ti me segment, when the second RF time segment is sh ifted from the first RF t ime segment by the timeshift of the second RF time segment from the first RF time segment, and the third R F time segment is shifted from the first RF ti me segment by the timeshift of the third RF time segment from the first RF time segment.

10. The method of any one of claims 1 - 9, further comprising identifying a further peak in the calculated cross-correlation, the further peak designating a TDOA of a further unknown radar signal at the first and second antennas.

1 1 . The method of claim 10, further comprising repeating the steps of any one of claims 2 - 9 for the further peak to output the further unknown radar signal.

12. A radar signal separator for separating an unknown radar signal out from mu ltiple unknown radar signals, the radar signal separator configured to:

- receive the multiple unknown radar signals from spatial ly different antennas; and

- separate the unknown radar signal out from the multiple unknown radar s ignals based upon the time difference of arrival (TDOA) of the unknown radar signal at the spatially different antennas.

13. The radar signal separator of claim 12. wherein the radar signal separator is configured to receive the multiple unknown radar signals from spatially different antennas by:

- receiving a first RF time segment taken from a first antenna; and

-receiving a second RF time segment taken from a second antenna that is remote from the first antenna, and wherein the radar signal separator is configured to separate the unknown radar s ignal out from the multiple unknown radar signals based upon the TDOA of the unknown radar signal at the spatially different antennas by:

- identifying a peak in the calculated cross-correlation, the peak designating a timeshift of the second RF time segment from the first RF time segment at which the first R F time segment and the second RF time segment correlate with one another;

- calculating an intersection of the first RF time segment and the second R F time segment when the second RF time segment is shifted from the first RF ti me segment b y the timeshift; and

- outputing the calculated intersection as the unknown radar signal .

14. A method of separating an unknown radar signal out from multiple unknown radar signals, comprising receiving the multiple unknown radar signals at spatiall y different antennas, and separating the unknown radar signal out from the multiple unknown radar signals based upon the time difference of arrival (TDOA) of the unknown radar signal at the spatially different antennas.

15. A method of receiving multiple unknown radar signals at spatially different antennas, the multiple unknown radar signals being from different radar transmitters that have different positions to one another such that the radar signals have different time difference of arrival (TDOA ) from one another with respect to the spatially d ifferent antennas, and separat i ng and extracting an unknown radar signal out from the multiple unknown radar signals, the method comprising:

receiving the multiple unknown radar signals at spatially different antennas, includ ing receiving a first RF time segment taken from a first antenna of the spatial l y different antennas and receiving a second RF time segment taken from a second antenna of the spatially different antennas, and

separating the unknown radar signal out from the multiple unknown radar signals based upon the time difference of arrival (TDOA) of the unknown radar signal at the spatially different antennas, incl uding:

calculating a cross-correlation between the first RF time segment and the second RF time segment, and 2 ) identi fying a peak in the calculated cross-correlation, the peak designati ng a TDOA of the unknown radar signal at the first and second antennas, and,

extracting the parts of the received signals that correlate with one another at the identified correlation peak to construct a radar signal.