 |
Description  |
|
|
This invention relates to pulse doppler radar.
In such a radar, pulses of microwave energy are transmitted with a certain
pulse repetition frequency (p.r.f.). When a pulse falls on a target it is
reflected and an echo pulse is received at a time after transmission
corresponding to the range (i.e. distance) of the target from the
transmitting antenna. To analyse the received information a set of gates
are opened sequentially, each for a short period of time, after
transmission of each pulse, each gate being arranged to pass the signals
received from a predetermined range to a separate channel. Within each
channel there is included means for analysing the spectrum of the received
signal such as to determine the frequency of the received pulse. The shift
in frequency between the transmitted pulse and the received pulse due to
the Doppler effect is indicative of the velocity of any target falling
within that predetermined range.
When the p.r.f. of a pulse Doppler radar is fixed, it is inherent that
there should be a degree of ambiguity if the p.r.f. is such that
transmitted pulses follow one another in a shorter interval than it takes
for a response to be received from the furthermost target within the
overall range of the radar. This ambiguity is caused by the fact that a
received signal need not necessarily be caused by a reflection of the
immediately preceding transmitted pulse but could be a reflection from a
more distant target of an earlier transmitted pulse. Thus, any received
signal may correspond to a set of possible ranges for the target, the
separation of the individual ranges being determined by the p.r.f. and the
total number of ambiguities being a factor influenced both by the p.r.f.
and the overall range of the radar.
A fixed p.r.f. may not only result in range ambiguities but also in
velocity ambiguities. To appreciate this it is necessary to consider the
frequency content of the transmitted signal. Because the transmissions are
in the form of bursts of a fixed carrier frequency, the carrier frequency
is not the only frequency present but there are also present all
frequencies resulting from modulation of the carrier frequency by the
p.r.f. The frequency spectrum of such a transmission is a line spectrum
centred on the carrier frequency f.sub.c and having a plurality of further
frequencies each differing from the centre frequency by a multiple
(including one) of the p.r.f. The amplitudes of the frequency components
decrease gradually with increasing separation from the centre frequency.
Spectrum analysis of reflections from a stationary target will yield this
same line spectrum whereas reflections from a moving target will displace
the transmitted spectrum by an amount proportional to the target velocity.
In the presence of noise one cannot be sure that in the received signal
the component of greatest amplitude is the centre frequency and it is
therefore possible to have a plurality of possible velocities each
yielding effectively the same spectrum upon analysis.
In order to remove ambiguities in range and velocity it has already been
proposed to change the pulse repetition frequency of the radar in between
scans. That is to say, it has been suggested that one may carry out a
complete scan of the area of surveillance with one p.r.f. and then carry
out a second scan with a second p.r.f. The second p.r.f. will also have
its own range ambiguities but provided that the two p.r.f.'s are not
harmonically related one may remove ambiguities by cross-correlation. This
is because if each p.r.f. yields a set of possible ranges for a target,
there should only be one target range which occurs in both sets. Similarly
velocity ambiguities may be sorted out by cross-correlation.
This proposal, however, gives rise to its own difficulties. A first
difficulty is concerned with the fact that unambiguous information
relating to any target can only be obtained after two complete scans. This
may mean a delay of one or two seconds and such a delay is often
unacceptable, for example where the target being detected is a missile
approaching at high speed. A second difficulty arises in signal processing
since all the information derived during a scan, i.e. information relating
to all targets within an area of surveillance, must be stored for a period
of one scan so that it may be correlated with the information derived from
the subsequent scan. Even if such storage does not in itself give rise to
very severe problems, there is still difficulty in correlating the
information derived during a scan with the information stored in respect
of a preceding scan since any target of interest is likely already to have
changed its position and possibly even its speed. In the case of a rapidly
moving target, it may mean that during one scan the target gives rise to a
reflection passed by a first range gate and in the next scan by a
different range gate. Thus, any system attempting to correlate information
from one scan to the next is inherently involved, complex and consequently
expensive and more susceptible to failure.
To resolve ambiguities without the need for storage and correlation between
scans, it is necessary to change the p.r.f. of the radar sufficiently
rapidly that during a single scan any target gives rise to reflections
with two p.r.f.'s. Thus the p.r.f. should be changed during the time that
the beam takes to traverse any point in space.
Previously, fast changes in p.r.f. have been considered as undesirable as
they may give rise to an unacceptable level of spectrum spreading. Such
spectrum spreading was thought to cause difficulties in the removal of
clutter and also in spectrum analysis to determine target velocity.
Furthermore, in order to optimize the signal to noise ratio it is
necessary to integrate the responses over the full time that the beam
crosses the target which involves integrating signals of two differing
p.r.f.'s with unknown transitions from one p.r.f. to the next.
A clutter filter capable of mitigating the problems caused by spectrum
spreading is described in our British Pat. Application No. 20242/75 filed
on the same day as the present application. The clutter filter described
enables interference from clutter to be suppressed and produces for any
moving target of interest a CW signal having a frequency dependent upon
Doppler shift. The frequency is centred on the I.F. of the radar and
differs from the I.F. by less than the p.r.f.
A suitable spectrum analyser for determining the frequency content of the
signals which occur at the output of the clutter filter is described in
FIG. 1 of the accompanying drawings. Such a spectrum analyser is known in
general terms, from our British Pat. Specification No. 1,299,023. The
spectrum analyser includes a bandpass filter 10 connected to receive
pulses of known repetition frequency which changes when the p.r.f.
changes. The output of the filter 10 is applied to a dispersive delay line
12. Bandpass filter 10 is arranged to reduce the bandwidth of the input
pulses to match the bandwidth of the dispersive delay line 12, these two
units together constituting a passive network for generating a frequency
swept signal.
These frequency swept signals are applied to one input of a mixer 14 which
is also connected to receiver Doppler shifted signals occurring at i.f.
frequency. The lower sideband of the mixing products is selected by a wide
bandpass filter 16 which produces a frequency swept signal of known centre
frequency whose timing is representative of the Doppler shift in the input
signal. This frequency swept pulse is compressed in a dispersive delay
line 18 whose output signal is passed to a detector and low-pass filter 20
to produce for each Doppler shifted input signal a pulse whose time delay
in relation to the pulses applied to the bandpass filter 10 is
representative of the Doppler shift.
The bandwidth of the Doppler spectrum is dependent upon the p.r.f. of the
radar. It is preferable to so choose the p.r.f. of the pulses applied to
the bandpass filter 10 that the sweeps are contiguous. When two p.r.f.'s
of say 8 KC/s and 6.6 KC/s are used alternately, then in one case after a
sweep of 8 KC/s is finished the next sweep begins and in the other case
after a sweep of 6.6 KC/s is finished the next sweep begins. This is to
enable maximum use to be made of available processing time; the passive
generation of frequency sweeps enabling phase continuity to be preserved.
In order to detect the signals under conditions of noise and also to
measure the Doppler frequency shift, it is necessary, as earlier
mentioned, to integrate the pulses from any target over the full time that
it is exposed to the radar beam.
At this juncture it is to be pointed out that the Doppler frequency shifted
signals applied to the mixer 14 of the spectrum analyser contain ambiguous
information and that ambiguities have to be resolved if correct
measurements of Doppler shift are to be effected. In the example given
above a true Doppler shift of, 2 KC/s would appear as a 2 KC/s shift in
both p.r.f.'s. A 10 KC/s doppler shift will also give a 2 KC/s shift at
the higher p.r.f. but will give a shift of 3.4 KC/s at the lower p.r.f.,
these figures being (10-8) KC/s and (10-6.6) KC/s respectively. Similarly,
a 2 KC/s shift can be obtained with the higher p.r.f. if the true Doppler
frequency shift is 18 KC/s etc. It is therefore necessary to correlate the
frequency shifts measured in respect of each target appearing at both
p.r.f.'s and to compare these to enable ambiguities to be resolved.
The present invention seeks to provide a radar in which the problems
associated with integration and ambiguity resolution are mitigated.
According to the present invention, there is provided a pulse Doppler radar
utilising rapidly switched p.r.f.'s, having a spectrum analyser adapted to
mix a Doppler shifted frequency derived from a target return with
frequency swept signals derived by application of a train of first pulses
to a dispersive delay line, the first pulses being spaced apart by time
intervals proportional to the prevailing p.r.f., the spectrum analyser
further including means for processing the mixing products to produce a
train of second pulses of the same frequency as the first pulses and whose
timing in relation to the first pulses is representative of the Doppler
shift, the radar further including integration means for integrating the
train of second pulses, which integration means comprises a plurality of
integrators sequentially connectible by a commutator means to receive the
output signals of the spectrum analyser, each integrator being connected
to integrate spectrum analyser output pulses corresponding to a
predetermined range of Doppler shifted frequencies, and means for changing
the commutation of the integrators in synchronism with changes in the
p.r.f. of the radar such that each integrator is arranged to receive the
signals at all p.r.f.'s corresponding a predetermined unambiguous range of
Doppler shifted frequencies.
Preferably, each integrator is constituted by a first capacitor connected
in a negative feedback loop of an amplifier and a second capacitor
connected between an input of the amplifier and a line at reference
potential, the commutator means being operative to connect different
capacitors across the latter amplifier at different times following the
said first pulses.
FIG. 1 shows a spectrum analyser for determining the frequency content of
signals.
The invention will now be described further, by way of example, with
reference to FIGS. 2, 3 and 4 of the accompanying drawings, in which:
FIG. 2 shows the form of the output pulses of the spectrum analyser,
FIG. 3 shows a commutated capacitor integrating filter, and
FIGS. 4a-4d are timing diagrams used to explain the method of operation of
the integrating filter.
Referring first to FIG. 2, the output of the spectrum analyser is seen to
comprise a series of pulses following a generally Gaussian shaped
envelope. Within the envelope, the pulses adopt one of two repetition
frequencies equal to the repetition frequencies of the signals applied to
the bandpass filter 10 of the spectrum analyser described with reference
to FIG. 1. The timing of the pulses under the Gaussian envelope in
relation to the timing of the pulses applied to the bandpass filter is a
function of the Doppler shift though the information contained in any one
repetition frequency alone is ambiguous.
FIG. 4a shows the timing relationship of the output pulses of the spectrum
analyser in greater detail. The vertical lines 40 indicate the timing of
pulses applied to the bandpass filter 10 to generate a frequency sweep
extending over a bandwidth of 8 KC/s and the vertical lines 42 indicate
the timing of pulses generating frequency sweeps of bandwidth 6.6 KC/s.
Assuming that there is present a true Doppler shift of 2 KC/s. This will
generate the spikes 44 each of which follows the preceding pulse 40 or 42
by a time period equal to the time taken for the dispersive delay line 12
to sweep through 2 KC/s. The filter described in FIG. 3 seeks to integrate
all the spikes 44 which, as may be seen, are not equally spaced in time.
The commutated capacitor integrating filter in FIG. 3 includes a series of
two resistors R1 and R2 connected to the input of an amplifier 30 whose
output is connected by way of a commutator 32 and one of a bank of
capacitors 34 to the junction between the resistors R1 and R2. The input
of the amplifier 30 is further connected to a second bank of capacitors 36
which are sequentially connected to earth by means of a commutator 38
which is ganged to the commutator 32. The output from the amplifier 30 is
smoothed by means of a low pass filter 39.
Assuming that the commutators 32 and 38 are stationary and that a
predetermined one of the capacitors in each bank 34 and 36 is connected in
the circuit. This constitutes a conventional integrating circuit in that
the charge stored by the capacitor in the bank 34 is proportional to the
time integral of the applied input pulses, the integration period being
determined by the decay rate set by the capacitor in the bank 36.
The commutators are arranged to operate at such a rate that they carry out
a complete cycle in the interval between consecutive pulses. Consequently,
each capacitor is connected to receive the signals from a narrow Doppler
frequency range. This may be seen from FIG. 4b which shows the sequence of
connection of the capacitors in a bank of 10 which are designated by the
letter of the alphabet A to J. Considering now only the sequence of events
when a bandwidth of 8 KC/s is swept within the spectrum analyser, each of
the capacitors A to J is connected sequentially in circuit and is
therefore operative to receive the signals covering a bandwidth of 0.8
KC/s. Capacitor A is always connected to receive the first 0.8 KC/s of the
sweep and capacitor J the last 0.8 KC/s. Thus in the drawing which
demonstrates a Doppler shift centred on 2 KC/s, this will always be
applied to the capacitor designated C.
When the p.r.f. of the radar is changed, then so is the frequency of the
pulses applied to the spectrum analyser and likewise the commutation. The
commutator continues to dwell for the same length of time on each
capacitor but on this occasion the capacitors I and J are omitted on each
cycle, thus shortening the length of the commutation cycle to match the
frequency bandwidth which is swept by the spectrum analyser. It will be
seen, therefore, that with the 2 KC Doppler shift this will continue to
give rise to pulses which are integrated by the capacitor C.
Within the range of Doppler frequencies extending from 0 to 6.6 KC/s, the
received signals will give rise at both p.r.f.'s to an increase in the
peak charge stored across a predetermined one of the capacitors designated
A to H. This is true for the zero order ambiguity. If the true Doppler
frequency, however, lies in a range between 6.6 and 13 KC/s then this
statement will not hold true but instead one capacitor will integrate
pulses arising during one p.r.f. and another capacitor the pulses
occurring at the other p.r.f. Such an example is given in FIG. 4c of the
drawings. In this example, the true Doppler frequency is taken to be 10
KC/s. Because of the modulation of the carrier frequency by the p.r.f. in
the manner earlier described, the Doppler frequency will have harmonics
spaced apart by the prevailing p.r.f. which, in the case of the first
p.r.f. will be 8 KC/s and in the case of the second p.r.f. 6.6 KC/s.
Consequently, at the first p.r.f. the Doppler shift will give rise to a
component occurring at 2 KC/s, as occurred in the example of FIG. 4a, but
on this occasion the component detected by the frequency analyser during
the second p.r.f. will occur at 3.4 KC/s.
In order to detect frequency shifts lying within this range, a second
commutated capacitor filter identical in construction to that shown in
FIG. 3 is provided whose commutation however is modified in the manner
shown in FIG. 4d. During the first p.r.f. all the capacitors continue to
be connected tn sequence in the same manner as has been described in
connection with the zero ambiguity filter. Likewise, at the second p.r.f.,
the time that the commutator dwells on each capacitor remains the same and
two capacitors are omitted in order to shorten the total cycle time.
However, on this occasson, it is not the capacitors I and J that are
omitted but the capacitors G and H though otherwise the cyclic sequence of
commutation is maintained. It will be seen from correlation of FIGS. 4c
and 4d that the 10 KC/s Doppler shift will give rise always to maximum
charge across the capacitors C. At the same time, the zero order ambiguity
filter will develop some charge across its capacitor C but also some
charge across its capacitor E and therefore whilst the capacitor C and E
in the zero ambiguity will have a charge the first ambiguity filter will
develop twice that charge across the capacitor C. By a comparison of the
levels it is possible to decide the level of ambiguity and thereby arrive
at an accurate measurement of Doppler shift.
In a corresponding manner, if targets in the Doppler range 13 to 20 KC/s
are of interst, it is possible to provide a second order ambiguity filter
again identical with the zero and first order ambiguity filters but in
which two other capacitors are omitted during the lower p.r.f. and whose
order of commutation is designed such that in the presence of a second
order ambiguity the charge will always develop across the same capacitor.
It follows that there is a need for as many commutated capacitor
integrators as there are possible ambiguities.
It will be appreciated that the integration technique provided in the
present application may also serve to resolve range ambiguities in a
totally analogous manner. In the case of range, the target returns are
ambiguous on account of the fact that any given echo may have been caused
by more than one transmitted radio frequency pulse. Hence, one has a set
of reference times corresponding to the times of transmission and a set of
echo pulses whose timing represents the range of the target with a
possibility of ambiguity in an analogous manner to that described with
reference to FIG. 4. It is believed that the manner in which a commutated
capacitor filter may be used to overcome the problems of integration in
the range channels will be self-evident to a person skilled in the art in
view of the foregoing description.
* * * * *
|
|
 |
|
 |
Description |
|
