GB2204757A - Radars - Google Patents

Abstract

The well-known problem of the so-called "multi-path" effect, in which radar energy signals traveling between the radar and a target can propagate along two or more different pathways so that the radar is unable to distinguish between these, is met by using a frequency-agile system. It transmits within a short time period a multiplicity of radar signal sequences each at a different frequency and then averages the returned, received, signals (or some value representative thereof) over a similar short time period to give an indication of the true position of the target relative to the reflective surface. <IMAGE>

Description

RADARS This invention concerns radars, and relates in particular to tracking radars capaxble of seeing and following a fast moving low level target.

A well-known problem in the radar Art is that caused by the so-called "multi-path" effect, in which radar energy signals travelling between the radar and a target can propagate along two or more different pathways, and the radar is unable to distinguish between these, and so cannot easily tell exactly where the target is. One classic occurrence of the multi-path effect is when the target is very close to a radar-reflecting surface - the ground, say, or the surface of the sea. Radar energy from the target - specifically, energy "shone" by the radar onto the target, and reflected off back to the radar - may return to the radar by two distinct routes, namely (a) directly, along the straight line from target to radar, and (b) indirectly, via a reflection off the surface roughly midway between target and radar.If the radar is unable to distinguish between the two energy inputs (the reflected signals) - if, for example, the radar cannot look at and see the target without at the same time seeing the reflection - then equally it may be unable to tell where the target actually is - thus, whether it is where it really is (a short distance above the reflecting surface) or where the reflection off the surface shows it to be (a short distance "below" the surface).However, because of the coherent nature of the radar energy the result in practice is not that the radar sees two targets (as would the human eye) but cannot tell the difference between the real one and the reflection, but that it sees a single target whose height apparently varies rapidly - and, though the radar cannot tell, the variation is from some indeterminate fraction of its real height to some equally indeterminate multiple (up to twice) of its real height.

Various methods of reducing multi-path effects which degrade the measurement of the elevation angle of an aerial target have been proposed. One common solution is to design the radar to have a very narrow beamwidth (in elevation), so that it is then possible for the radar to look at, and see, the target without at the same time being able to see its reflected image. Clearly, by discriminating against the image it is possible to measure the genuine elevation angle of an air target flying at low level. Unfortunately, there are several problems associated with very narrow beamwidths.

Firstly, radars using a very narrow beamwidth have difficulty in finding the target (if the beam is narrow then the radar must look in precisely the right direction to see the target, but if it is not known where the target is this is not so easy!). Further, a narrow beamwidth is usually achieved by employing a radar antenna wdith a large aperture, a very short wavelength, or a combination of both. Naturally, a large aperture antenna is physically large, bulky, and with considerable inertia - and is hardly suitable for use with a radar that is tracking a fast moving target.Again, radars using a very short wavelength tracking radars operate generally in the I and J bands, up to about locum, and in this context "very short" means the lem and less wavelengths of the Q band - to create a narrow beam have to deal with the significant attenuation of the radar energy met if the signal path'is through bad weather, rain in particular (this additional attenuation is sufficient to prevent shorter wavelength radars being described as all-weather radars).

The present invention provides quite a different solution to the multi-path problem, avoiding any necessity for very narrow beams, and requiring only a relatively wide bandwidth, frequency-agile system. It is based upon the fact that the radar energy employed, even in pulse radars, is coherent and the apparent rapid variation in target height is the result of the energy coming directly from the target interfering, constructively or destructively, with the energy reflected off the surface to a degree depending upon the variation in path length phase difference as the target travels towards the radar. This means that the real height of the target must be somewhere below its apparent maximum height (caused by constructive interference) and above its apparent minimum height (caused by destructive interference).

At any given time the apparent height of the target depends upon the phase difference in the two path lengths - whether the radar energy waveforms arrive at the radar in phase, out of phase, or somewhere in between - and also (of course) on the relative amplitude of the two signals (i.e., the "brightness" of the reflection as compared to the target as viewed directly). If the reflecting surface was a perfect mirror, then the reflected image would be as bright as the target, the two radar signals would be of equal strength, and the target's apparent height would range from zero (1800 out of phase; full destructive interference) to twice its real height (perfectly in phase; full constructive interference).

With a less-than-perfect reflecting surface the apparent height ranges from somewhere below its real height (but above zero) to somewhere above its real height (but below twice the real height). The problem, then, is how to "average" the two heights in order to deduce the real height ; and how to do it in real time so that the deduction can be done before the target gets too close.

The invention's solution to the first part - how to "average" the apparent height - is to integrate the apparent height error over at least a full cycle (from, say, the apparent maximum to the minimum height, and back). From the result can be found the real height error, and thus the real height. However, if the system merely waits sufficiently long for a full cycle to occur simply because the target has travelled the requisite distance, then the target may now be too close for the action the radar is designed to initiate. For example: a typical target might be an Exocet missile 5 miles (8 kilometres) out travelling directly towards a ship at Mach 3 (about 2,000 mph at sea level ; this is roughly 3,000 feet per second, or 1,000 metres per second) and at a height of 10 feet (3 metres).With a radar operating at 10cm wavelength and having its antenna 30 feet (10 metres) above sea level the path length phase difference would only change through a full cycle as the missile travelled 4.5 miles (7.5 kilometres), which makes it about 1/2 mile (0.8 kilometres) away before the radar knows its real height. This is rather too close! The invention deals with this, the second part of the problem, by having the radar transmit energy in bursts not of a single frequency/wavelength but of a multiplicity of frequencies/wavelengths very rapidly one after another so that the equivalent multiplicity of apparent target heights the radar sees in effect defines the full range of apparent heights that would be seen over a full cycle of some mean frequency/wavelength energy. By averaging - and specifically by integrating - these values (which may be based on energy received in as little as 1/25th of a second) the radar can deduce the missile's true height in the time it takes to travel 40 yards (roughly 40 metres), which is more than fast enough.

Accordingly, in one aspect this invention provides a radar system capable of accurately seeing a target despite the latter's proximity to a radarreflective surface, which system transmits within a short time period a multiplicity of radar signal sequences each at a different frequency and then averages the returned, received, signals (or some value representative thereof) over a similar short time period to give an indication of the true position of the target relative to the reflective surface.

The invention is based upon the idea that by using a radar with a large bandwidth at micro-wave, and hopping rapidly from frequency to frequency, it is possible to obtain a large reduction in the multipath errors and yet in the same design use both radar frequencies that are free from unacceptable attenuation due to weather, and also a beamwidth that is comparatively large so that there is no difficulty in laying the beam on target to start the tracking sequence.

As implied above, the radar parameter which provides the solution is the micro-wave bandwidth.

The mulithpath error is a function of h1 h2/#R where: hl is the radar height above the earths surface; h2 is the height of the target; Xis the radar wavelength; R is the range of the target from the radar In this equation, hl and h2 are virtually constant, R varies slowly as the target approaches or recedes from the radar, but X , the wavelength, is under the control of the radar designer. By varying the wavelength by a sufficient amount in a short time to give a large number of samples of the apparent elevation angle of the target, and then within the radar integrating these samples of apparent elevation angle, there is achieved a value of apparent elevation angle which is effectively the actual elevation of angle of the target.Thus, in essence, the method of solving the multi-path problem, reducing the multipath error to an acceptable value, is to design a radar with a very wide bandwidth at micro-wave and integrate the errors at each wavelength in the sample set to obtain a mean value which is a good measure of the elevation angle of the target being tracked. And by employing this bandwidth technique to reduce elevation errors the design concept of the radar still include a combination of the necessary operational attributes, i.e. all-weather, a multi-path solution on low-level targets, an adequate beamwidth for target acquisition, and even state-of-the-art M.T.I.

processing to reduce interference from clutter.

The invention is a radar system, and indeed specifically applicable to such a system that is a tracking radar designed to seek out, lock onto and then track a target. Tracking radars commonly operate in the I or J band (a frequency of round about 8 GigaHertz, with a wavelength of 10cm or less), and have an angular beamwidth in elevation of roughly 50.

They can often be switched from one frequency to another (to choose the frequency best suited to the expected target type or weather conditions), and may thus have a bandwidth of a few Gigahertz, but are not in use rapidly and frequently switched from one to the other so as to operate at a multiplicity of frequencies/wavelengths within a short time. An example of such a tracking radar is the Marconi Fire Control Radar MV400. This radar operates in the micro-wave bandwith (the octave from 8 to 16GHz), with an average beamwidth of 30 in both elevation and azimuth. The radar uses a pulse waveform, and transmits 12,000 pulses/sec.Pulses are transmitted in groups of 8 at the same frequency; each group of 8 is on a different micro-wave frequency, and there are 64 different micro-wave frequencies to select (after 8 x 64 pulses the radar waveform is repeated in terms of re-using the micro-wave frequencies, but the order of selection may be different). When used in the invention the complete set of 64 micro-wave frequencies can be employed in a time of 42.67 milliseconds. Each different micro-wave frequency obtains a different value for the apparent elevation angle of the target, so that in 42.67 milli-seconds 64 differennt values of apparent height are obtained that can be averaged to give a fairly accurate value for the real height.

The inventive system is for seeing a target close to a reflective surface, and one specific class of target is that of the air-to-ship or ship-to-ship missile of the Exocet or Sea Eagle variety. Other likely targets are low-flying aircraft or even individual gunfired shells.

The system of the invention allows the radar to see the target despite the latter's proximity to a radar-reflecting surface. In the normal course of events the surface will be the sea surface (or perhaps the ground surface), but other types of radarreflective surfaces can be envisaged (and are not excluded). Moreover, though a sea surface will invariably be generally horizontal, a ground surface may be at some non-zero angle to the horizon - the ground may be the sloping surface of a hill, for example. Thus, the target's true position relative to the reflective surface may be described as any appropriate combination of elevation and azimuth - but purely for convenience the invention is hereinafter discussed mostly in terms of a target above a horizontal reflecting surface, so that the target's position may be defined in terms solely of its elevation (height).

In accordance with the invention the radar system transmits a multiplicity of signals over a short time period. The length of this short time period is, as will become clear hereinafter, bound up with the number and length of each individual signal sequence, the bandwith and frequency-agility of the system, the required accuracy of the results, and so on.

Generally, however, the short time period will be of the order of 0.1 sec or less (by which time even a very fast moving target will not have traveled much more than 100 yards or so).

The system transmits a multiplicity of signals each at a different frequency. These two factors the multiplicity and the frequencies - go hand in hand. The object is to enable sufficient signals to be received in return and for long enough, giving apparent heights acceptably spread over the range between the maximum and minimum apparent heights, that there may be achieved an average height that is an accurate measure of the true height.More specifically, it is required that the multiplicity of signals define the variation of the signal value (and thus the target's apparent height) that would normally occur with time - the use of different frequencies is, as explained in more detail elsewhere, an acceptable substitute for a time period - so that the accurate average may be calculated (preferably, as discussed further hereinafter, by integrating the area under the graph obtained by plotting the signal against frequency/wavelength). Although discussions on signal sampling techniques always point out that no repetitive curve (like the sine wave of a radar signal) can be defined until it is sampled at least twice within one wavelength - thus, at at least twice the signal's own frequency - this is not nearly sufficient to allow a good representation of the wave to be constructed. Indeed, modern digital sampling may occur at five, or more, times the maximum signal frequency. In the radar system of the invention, however, it is preferred to "sample" - to switch frequencies - at least 20 times within the chosen short time period. Indeed, most preferably the multiplicity of frequencies is more than 20, and is in the range 32 to 64, and indeed is most preferably as high as 64 (though 40 to 50 is cost effective).

The number of signals in the multiplicity of signals will be at least partially constrained by the minimum length of each signal that is necessary to ensure receipt of a return allowing an accurate apparent height computation to be performed. This should not, however, set any significant limit on the number, for modern techniques can make use of a return signal lasting as little as seconds (which, even allowing for the frequency switching time, would still enable there to be differentfrequency signals in 0.1 second).

Each signal in the multiplicity of signals is at a different frequency. This is naturally necessary if there is to be simulated the manner in which interference between the direct and reflected target signals varies as the target distance varies.

However, to define best the manner in which the apparent target height varies between its maximum and minimum values these frequencies need to be spread evenly over whatever frequency range, or band, most simply simulates the interference effect of the signal from a moving target. Mathematical analysis of the interference situation shows that this band is an octave - that is, the range from any particular radar frequency to twice that frequency - while convenience dictates that the band should be centrally disposed in the range over which the radar is capable of agile operation. Thus, the different frequencies of the multiple signals are preferably evenly spread over a band that is an octave wide and centred on the radar's bandwidth centre.For a 10cm radar, for example, operating with a central frequency of 12GHz, the frequencies should be spread evenly from 8GHz to 16GHz.

The return signals received within the allotted short period, or some values representative thereof, are averaged to give a measure of the true height of the target. The averaging technique employed may be any suitable such technique, but conveniently involves what is effectively an integration of the area under the height/time curve calculatable from the signals.

One particularly convenient way of achieving this makes use of the relatively slow response - which is an integrating response - of the motors driving the radar tracking system. If these are pulsed in short equal length bursts whose magnitude represents the present apparent height (or, rather, the height error - specifically the elevation angle error) they will over a time point the radar in the average direction.

For example, if one pulse represents "too high by 0.1 " (the target is 0.10 lower than where the radar is pointing) while the next represents "too low by 0.30" (the target is 0.30 higher than where the radar is pointing) then a sluggish tracking drive motor will end up by "combining" the two pulses, acting as though it had received a "too low by 0.1 " pulse (where 0.10 is the average of the 0.10 too high and 0.30 too low pulses: 1-0.1 + 0.3]/2 = 0.1).

Naturally, the short time period over which the averaging of the return signals is performed should be of the same length as that during which the multiplicity of signals is transmitted. Using an integrating system to perform the averaging takes care of this automatically.

Certain aspects of the invention are now described, though only by way of illustration, with reference to the accompanying drawings in which: Figure 1 shows diagrammatically the multi-path problem as applied to a ship-mounted radar tracking a low-level missile approaching over the sea surface; Figure 2 is a diagrammatic graphical plot of the apparent height of a target (the missile of Figure 1) as it varies with distance; Figure 3 is a plot, similar to that of Figure 2, but showing how the height/distance curve differs for different radar frequencies; Figure 4 is a plot, similar to part of Figure 3, showing how, at a "fixed" range, the apparent height varies with radar frequency; Figure 5 is a plot, similar to that of Figure 4, showing how the height/frequency curve may be obtained with a multiplicity of discretely different radar frequncies;; Figure 6 is a real version of Figure 2; Figure 7 is a real version of Figure 5; Figure 8 is a modified version of Figure 7; Figure 9 is a real plot of the residual error left in the system giving rise to the plot of Figures 7/8; and Figure 10 is a block diagram of the radar system giving rise to the plots of Figure 7/8 and 9.

The diagrammatic sketch of Figure 1 shows a warship (11) being attacked by a low-level missile (12) approaching from the right (as viewed). The ship's radar (13) is attempting to track the missile, but its beam is so wide in elevation that radar energy can travel to and from the misile both via the direct line (14d) and by the indirect line (14r) reflected off the sea surface (15). If the radar were able to look exclusively down the indirect line 14r it would see the missile - in fact, a reflection (12r) of the missile in the sea surface - underneath the sea. It cannot do this, however, for its wide elevation angle means it looks along both lines, see "both" missiles (12 and 12r), and, being unable to distinguish between the two, thinks that the missile is really at a height somewhere between the two.

Figure 2 is a graphical plot of where (at each height) the radar thinks the missile is as the missile approaches. The dashed straight line is the missile's real height, the solid wavy line is the apparent height according to the radar. Depending upon the radar reflectivity of the sea surface (15 in Figure 1) - about 0.8 is common for a calm sea - the effect of constructive and destructive interference between the two return signals is to make the apparent height vary between half its real value and 1 1/2 times its real value, though this difference decreases as the missile gets closer (at lower ranges the radar's characteristics mean that the elevation beamwidth is effectively narrower, so it becomes easier for the radar to distinguish between the missile and the reflection).The difference decreases to zero at very close range - but by then it is usually too late to do much about it.

In Figure 3 there is shown, superimposed, the Figure-2-like plot obtained for three different radar frequencies (one the same as in Figure 2, one lower, and one higher). The plot the same as that in Figure 2 has the solid line, the lower frequency one has the dot-dashed line, and the higher frequency one has the dotted line. Looking at any particular range - for example, that depicted by the slot-like window (W) it is clear that three radar signals of these three frequencies give three different target heights. If it were imagined that the target was stationary at the window W range, and the radar frequency was varied over roughly an octave centred on the middle frequency, then the signal received would indicate an apparent height varying with frequency in the manner shown by the "perspective" plot forming part of Figure 3 - and shown in more conventional form in Figure 4.This latter plot can itself be re-drawn as in Figure 5, in terms of height error (representing the magnitude of the correction signals involved in the radar's tracking system) against the phase difference between the radar signal travelling via the direct path 14d and that via the indirect, reflected path 14r.In this plot the area underneath the curve (between the curve and the phase-difference axis) is equally split into plus and minus areas - that is, between, say, STrand 6gr the plus area above the zero error line equals the minus area below that line, as is so for the corresponding areas between 6n: and 7X , and as is also so for the combination of areas between 5nr and 7SK Indeed, this is so for any combination of areas between two points with a phase difference of a whole number of 2v radians - 2 , 4 , 6or , 8 ,etc.

It will be clear from this that averaging all the height errors obtained from such a plot will result in a zero, or near zero error as all the plus errors cancel all, or nearly all, the minus errors (if a length of plot is taken such as is exactly 2fr long then cancellation is exact, and a zero error results, but otherwise there remains a small error uncancelled; this is part of the residual error, and is discussed further below).

Figure 6 is a real plot, similar to that of Figure 2, of angular elevation error (rather than apparent height) against target range, and Figure 7 is a real plot, similar to that of Figure 5, of angular elevation error (rather than height error) against phase difference for the range window at 5Km. Figure 8 is an enlarged, slightly modified, version of Figure 7.

In the invention there is used a method of reducing the error caused by multi-path effects in tracking radars in which there is employed a very wide micro-wave bandwidth, and the radar is designed to be agile in frequency over this bandwidth. The radar uses a large number of frequencies to sample the waveform of the multi-path error as it varies with the phase difference between the direct and reflected waves from the target.

The direct wave is given by sin wt and the reflected wave is sin w(t - T). The phase difference at the radar is = wT = ? < fT (1) where f is the frequency, and T is the time difference between the direct and reflected paths. T is given by T=d c where c is the velocity of light and d is the pathlength difference.

The radar elevation error is related to the phase difference J by

(-2.elevation angle) (2) where K is the reflectivity of the reflecting plane.

Equation 2 for a particular geometry and reflection coefficient produces the waveform shown in Figure 7.

From equation (1) = 2 n' fT = 2f c For a given geometry the path difference d is

where: hl is the radar height; h2is the target height; and R is the range of the target.

where: # (the wavelength) = c/f, For the particular case in which hl = 3m, h2 = 50m, R=5000m:

If the radar is agile at selected frequencies (from, say, 8 to 16 GHz) then for this geometry the phase difference varies from = = 1.6 x 2fr toe52 = 3.2 x 2t i.e. changing the frequency of the radar at a rate virtually instantaneously compared to the change of geometry due to target motion creates a change of 1.6 cycles in the cusp waveform given by equation (2).

Referring to Figure 7, with the particular parameter values specified, the action in the radar is to select frequencies to provide a stepped phase shift from 1.6 x 21r to 3.2 x 2 , and so generate a set of samples of the error signal one at each micro-wave frequency. These samples of the cusp waveform are integrated to a mean value using sufficient samples to obtain a good mean. With a Marconi MV400 radar operating as described above, within a time of 42.67 milli-seconds, 64 different samples of elevation error are obtained. The action of the angle servo which controls the pointing of the antenna along the boresight is to average the 64 samples. This averaging process, which happens naturally with a servo bandwidth of 4Hz, results in a very small residual error.

The mean of a half cycle (or multiples thereof) is zero. Thus, if the change of radar frequency over the micro-wave bandwidth and a particular geometry creates a whole number of half cycles for integration then the error is zero. But if there are fractions of the waveform additional there is a residual error after integration. This can be understood from the plot in Figure 8, where the sampling over the 8GHz bandwidth (from 8 to 16 GHz) includes more than just a portion 2rf wide. As depicted, some parts of the extra cancel (the hatched areas), but the rest (R) remains to give rise to the residual error.

In the case of a fixed frequency radar the error is any value on the error curve for a particular geometry and the particular micro-wave frequency (see Figure 6). By using a frequency agile radar with frequency agility over an octave or more then a large number of plus and minus samples of error are produced; integration by the action of the angle servo mechanism provides the mean of the samples.

Practical values of servo bandwidth, and particular values for the excursion of frequency agility, create a situation where several mean values of the residual angular error are integrated, thus creating a mean of the samples due to frequency change and samples due to changing geometry caused by target movement in space.

The significant effect is caused by change of microwave frequency of the radar; target motion contributes to reducing the error.

A typical residual error plot is shown in Figure 9; its "derivation" may be understood from a further consideration of Figure 7. Figure 7 shows a plot of elevation error as a function of phase shift, the phase shift limits being F = 1.6 x zero radians at 8GHz and f = 3.2 x 2Tf radians at 16 GHz. The sampling frequencies are selected in the bandwidth 8 to 16 GHz, and at each frequency there is an ordinate, or error, in milli-radians; these ordinates carry plus and minus values with respect to the actual elevation of the target. Over a complete half cycle of the waveform shown in Figure 7 the ordinates add up to zero.

Errors occur when a fraction of the waveform (either positive or negative) is left over; the worst case is a quarter of a cycle.

In the example shown in Figure 7, the worst case error is The area of a quarter cycle (m rads) micro-wave bandwidth (radians) = 0.8 m rads (this is a peak value) In the case of a fixed frequency radar with both the target and the earth's surface in the beam the measurement of target height by radar is unacceptable.

For the particular case of a target flying at a constant height of 50m towards the radar, the height error varies from 70m above the target at long range to zero error at short range (when the radar discriminates between the target and its image in the surface,) as shown in Figure 6. However, by applying the inventive technique the error due to multi-path reduces considerably. For the particular case of Figures 7 and 8 the error reduces to a few metres above and below the target. In effect, the error reduces to a value comparable with the physical dimensions of a strike aircraft or helicopter.

In the MV400 the 64 samples of error are digitised with the polarity retained. By an addition process on plus and minus values a mean value is obtained before the servo integration is applied.

Averaging the errors at the input to the servo prevents saturation on the peaks of the error signal, and the action of the servo is to integrate several sets of digitally integrated values at 42.67 millisecond intervals. With a servo response time of 250 milli-seconds (servo bandwidth = 4Hz) the servo provides additional integration on nearly 6 groups of the original 64 samples; thus creating a further reduction on the error signal.

Figure 10 shows in schematic form an MV 400 radar system using the technique of the invention. The components of the system are self-evident from the Figure. The operation is as follows: A pulse waveform is selected by the waveform generator, and each pulse selects a micro-wave frequency in the bandwidth 8 to 16 GHz. The waveform is selected to obtain Moving Target Indication (M.T.I.) on pulse groups at the same frequency (8 pulses per group), and each group is at a different micro-wave frequency to select error samples across the radar bandwidth. A total of 64 error samples are obtained in 42.67 milli-seconds.

The antenna is carried on a two- axis director, i.e. with both elevation and azimuth servo control., The static split obtains monopulse processing to form a sum channel and two angle channels - elevation and azimuth.

Target echoes in the receiver are automatically detected using digital sampling of signal amplitudes.

A target signal - the strong echo - is selected for range tracking, and having measured the range or timing of the echo the angle signals are accepted in a range gate for processing in terms of angular error.

The error signal in digital form is a set of samples. When the multi-path effect is present, these samples follow a waveform similar to the one shown in Figure 7. The action of the digital smoothing is to reduce the magnitude of the error signal prior to integration by the angle servo, thus preventing unsymetrical limiting due to saturation on the peaks of the error signal, for example.

The reduced error signal is then smoothed further by the action of the servos which control the position of the antenna in both elevation and azimuth. Note that this type of multi-path reduction is employed in both angle channels because on sloping terrain the multipath effect may occur in both planes.

M.T.I. processing prior to the smoothing of the angular error is essential to prevent echoes from clutter interfering with the multi-path reduction process.

Claims (10)

1. A radar system capable of accurately seeing a target despite the latter's proximity to a radarreflective surface, which system transmits within a short time period a multiplicity of radar signal sequences each at a different frequency and then averages the returned, received, signals (or some value representative thereof) over a similar short time period to give an indication of the true position of the target relative to the reflective surface.

2. A radar system as claimed in claim 1 that is a tracking radar operating in the I or J band, and having an angular beamwidth in elevation of roughly o 5.

3. A radar system as claimed in either of the preceding claims, wherein the length of the short time period over which the system transmits a multiplicity of signals is of the order of 0.1 sec or less.

4. A radar system as claimed in any of the preceding claims, wherein the multiplicity of frequencies is more than 20.

5. A radar system as claimed in claim 4, wherein the multiplicity of frequencies is at least 64.

6. A radar system as claimed in any of the preceding claims, wherein the number of signals in the multiplicty of signals corresponds to a rate of around 10,000 signals per second.

7. A radar system as claimed in any of the preceding claims wherein the multiplicity of signals is at frequencies spread evenly over an octave centrally disposed in the range over which the radar is capable of agile operation.

8. A radar system as claimed in any of the preceding claims, wherein the short period, or some values representative thereof, are averaged by a technique involving what is effectively an integration of the area under the height/time curve calculable from the signals.

9. A radar system as claimed in Claim 8, wherein the "integrating" average is achieved by making use of the relatively slow response of the motors driving the radar tracking system.

10. A radar system as claimed in any of the preceding claims and substantially as described hereinbefore.