Constant False Alarm Rate Doppler Radar System

PD-73365 Constant False Alarm Rate Moving Target Indication Radar System (CFAR-MTI) filed concurrently herewith, Ser. No. 588,577; filing date June 20, 1975.

1. Field of the Invention

This invention relates generally to radar systems and more particularly to constant false alarm rate doppler systems.

2. Description of the Prior Art

Electronic countermeasures, commonly referred to as jamming, presents a serious problem to doppler radar systems. The jamming energy causes saturation and sensitivity degradation in addition to causing many false alarms due to the required dwell time inherent in doppler radar systems.

It is well known in the art that the doppler frequency shift resulting from a relative velocity between radar and target may be used either in pulse, or continuous-wave radar to distinguish fixed targets from those in motion. However, the distinction of moving targets in itself is not the only nor is it the most important benefit to be gained from the extraction and employment of doppler information. It is also well known in the art that even when the return or echo from fixed targets called clutter, in orders of magnitude greater, say 20 to 50 db, than the moving target echo, doppler radars can nevertheless discern the moving target through utilization of doppler information. These advantages stem from the basic physical observation that echo signals from fixed targets will not undergo a frequency shift while the echo from a target exhibiting a relative velocity v.sub.r will be frequency shifter in an amount f.sub.d described by the relation

where .lambda. is the radar wavelength. Detection of moving targets has been reported even when the target echo has been 70 to 90 db below the clutter echo.

Observation of moving targets vis-a-vis stationary targets may be made by watching the video output in a radar receiver system on an A-scope. Observation of successive sweeps reveals that echoes from fixed targets do not vary appreciably, however, echoes from moving targets do vary predictably from sweep to sweep in amplitude at a rate related to the doppler frequency. Superposition of successive sweeps results in an A-scope display that distinguishes the moving targets. Such superposition, while effective for distinguishing moving targets on an A-scope, is not, however, suitable for PPI display.

One method for processing doppler information so as to render it suitable for PPI display is through the implementation of a delay-line canceler. The delay-line canceler in its simplest form, rejects the fixed target return and passes the returns from moving targets. In performing this function, received video information is fed to two separate channels in one of which the video signal is timed delayed by one pulse repetition period. The outputs from the two channels are then subtracted one from the other thus eliminating the constant amplitude signals of echoes received from fixed targets and preserving an uncanceled residue signal resulting from the subtraction of the pulse-to-pulse varying echoes received from moving targets. Output from the subtractor in an ideal system will only be produced by moving targets.

Historically reference to a doppler radar indicates a system in which the doppler frequency measurements are unambiguous but in which ranging measurements may or may not be ambiguous. Ambiguous range means that multiple-time-around echoes are possible while ambiguous doppler means that blind speeds fall within the compass of expected target speeds.

A constant false alarm rate (CFAR) doppler radar system according to the invention provides CFAR doppler radar in various kinds of jamming or electronic countermeasures (ECM) environments. It has been noted that doppler systems, because of the inherent dwell time, are peculiarly susceptible to various kinds of jamming. It would be of great benefit to the radar art to provide a doppler radar system that maintains improved sensitivity and CFAR action in all types of jamming environments.

It is thus an object of the present invention to provide a CFAR-doppler radar system acceptably operable in various kinds of ECM.

It is a further object of this invention to provide a CFAR doppler radar system that will not suffer from saturation or incapacitating sensitivity degradation.

FIG. 1 is a simplified block diagram of the front end of a conventional radar system.

FIG. 2 is a simplified block diagram showing a continuation from FIG. 1 of a conventional radar system but in addition including the details of the invention.

FIG. 3 comprises input signal and response curves, helpful in an understanding of the operation of the invention.

FIG. 4 comprises graphical illustrations of swept jamming and the response thereto.

FIG. 5 illustrates off-frequency jamming and the response exhibited by prior art systems as compared to responses obtained from systems incorporating the present invention.

Referring first to FIG. 1, the front end of a conventional pulse radar system is shown. The transmitted pulse is conventionally developed through the action of the exciter 10 responding to pulse expansion network 2, pulse generator 4 and transmitter frequency select gate 6 and furnishing local oscillator output signals to first mixer 26 and second mixer 32. The exciter 10 also furnishes output signal to Range Counter-Clock Generator-Timing Control 8, and the Master Oscillator Power Amplifier (MOPA) chain 12. The pulse so derived is then applied to the antenna via ATR 14 and TR 18 and radiated into space via antenna 20. Assuming a target within the range of the radar system, the return echo therefrom will be applied typically through the antenna 20 and the transmit-receive (TR) switch 18, preamplified in preamplifier 22 and then processed through image rejection filter 24. Conventional first and second IF amplifications are then accomplished by the actions of first mixer 26, amplifier 28 and first IF filter 30, second mixer 32 and second IF filter 34.

Referring now to FIG. 2, the action of the circuit used to supply the normalized output to the doppler processor will be explained. A Dicke-Fix like receiver is used to process the intermediate frequency signal. This receiver, using a wide-band, hard-limiter is sometimes named after Dicke. References: Radar Handbook by M. I. Skolnik, McGraw-Hill, 1970, pp. 5-31 - 5-38; Radar Design Principles by F. E. Nathanson, McGraw-Hill, 1969, pp. 119-130.

Before describing the circuit used in the invention it may be instructive to explain the operation of a conventional Dicke-Fix receiver. It will be convenient in that connection to present some typical numerical values in order to further clarify the explanation. The normal Dicke-Fix receiver comprises a wideband amplifier with a bandwidth, for example of 20 megahertz followed by a hard limiter, limiting well into the noise, followed in turn by an optimum bandwidth linear amplifier with a bandwidth of say 300 kilohertz for a 2.0 microsecond pulse width mode. Since the hardlimiter limits below its wideband noise level the noise level in the narrow-band channel will be decreased from the wideband noise level by the wide bandwidth-to-narrow bandwidth ratio, whereas there will be no reduction in the amplitude of a pulse whose width is matched to the narrow-band filter. This ratio turns out to be approximately 66:1 or about 18 db for the 2.0 microsecond pulse mode. This result means that no matter how large a signal may be at the input to the receiver, it will only be 18 db at most above the normal noise level at the narrow-band output. Even though the signal-to-noise ratio during the signal pulse may be very large thus suppressing the noise during the pulse, the noise level before and after the pulse is only 18 db below the maximum pulse amplitude.

The normal Dicke-Fix receiver responds to different types of jamming in different ways. The major disadvantage of the Dicke-Fix receiver manifests itself in the receiver's response to off-frequency CW jamming. CW jamming off-frequency, but within the widband bandwidth will suppress the output by the same number of db that the CW jamming is above the wideband noise level. This is called jam capture. On-frequency CW jamming will saturate the narrowband channel causing large numbers of flase alarms and signal suppression. Broadband noise jamming will reduce receiver sensitivity but will not increase the false alarm rate. Swept jamming, so disastrous to most other receiver types, is combatted very effectively in the normal Dicke-Fix receiver. If, for instance, a swept jammer sweeps over a 100 megahertz bandwidth every 5 microseconds, this would cause false alarms every 5 microseconds in most other type receivers. In the Dicke-Fix receiver, signal suppression occurs during the time the jammer is within the 20 megahertz bandwidth of the wideband limiter, that is, for one microsecond every five microseconds and for the other four microseconds, normal sensitivity would prevail. Since the swept jammer only appears in the 300 kilohertz narrowband channel for about 0.02 microseconds, the mismatch is so great that no false alarms will occur. Pulse jamming that is slightly off frequency does not affect the Dicke-Fix receiver because the hard limiter places an upper limit on the signal-to-noise level into the narrow band filter of 18 db in our example. As a result, signals at frequencies other than the IF center frequency are suppressed by the filter bandpass characteristic. A signal that is attenuated by the filter skirts by the ratio of the wideband-to-narrow bandwidth, or greater, is at or below the output noise level. For instance, assume a pulse jammer with a signal-to-noise ratio or +40 db that is off-frequency enough to allow a +20 db pulse through the linear and the logarithmic channel. The output of the narrow-band channel of the Dicke-Fix receiver will only be suppressed by 40-18 or 22 db during this pulse interval. Consequently, pulse jamming must be on-frequency and reasonably matched to the narrow-band channel to cause false alarms in the Dicke-Fix receiver.

Still referring to FIG. 2, the CFAR system of the invention in which a Dicke-Fix like circuit is used may be explained showing that unlike a conventional Dicke-Fix circuit, the system of the invention will not suffer from either off-frequency capture of the hard limiter or clutter capture of the hard limiter. To explain this effect it is convenient to use some typical parameters. Assume the wide-band hard limiter 36 has a 10 megahertz bandwidth and the narr-band matched filter 38 has a 1.0 megahertz bandwidth. The ratio of these two bandwidths defines the dynamic range of the Dicke-Fix receiver as:

at the output of the narrow-band filter 38. When large signals outside the narrow-band filter but inside the wideband limiter exist, as in FIG. 5 at curve (b) they will capture the limiter. This capture causes the noise or signal level out of the narrowband filter 38 to decrease below the normal receiver noise level. We define the jamming-to-noise level of the jamming over the 10 megahertz noise bandwidth and when this is, for instance, +15 db, then the signal and noise out of the narrow-band filter 38 will be 15 db below normal. This is the capture effect of the conventional Dicke-Fix receiver. This capture effect is overcome by the proposed system as follows: the A/D converters quantize several db into the noise; the wideband-to-narrowband ratio of 66:1 or about 18 db has been assumed. The first step to overcome the limiter capture problem is to provide large dynamic range analog-to-digital (A/D) converters 48 and set their maximum level to the maximum hard limiter output which is well defined by convention as 6 db per bit. For example, assume A/D converters with 8 bits plus sign which give a total dynamic range of 48 db. This means that with no signal or jamming present, the quantization into the receiver noise has the value

An improvement in dynamic range may be realized through the implementation of a pulse compression network 39. With pulse compression techniques there may be obtained, for example, through utilization of a 10 db pulse compression ratio gain, an additional 10 db of dynamic range. A conventional coherent oscillator (coho) 46, supplies the reference signals for the system in the form of a direct signal to phase detector 40 and a signal shifted 90.degree. by phase shifter 44 to phase detector 42. Phase detectors 40 and 42 also receive the output signal from the narrowband filter 38 and process this signal with the coho signals to form the in-phase (I) and quadrature (Q) signal components applied to A/D converters 48. Consequently if the signal suppression due to jamming is less than 38 db a complete recovery of suppressed signal levels can be achieved. After recombination in recombining network 52, the maximum amplitude A has a form resulting from I and Q signal recombination. This signal is now processed through a digital FTC circuit comprising shift register 54, summer 56 and normalization converter 58. A represents the content of present range bin, B represents content of the recent range bin, C the next most recent, and D the next most recent before C. The sum, B/2 + C/4 + D/8 is applied to summer 56 and the signal resulting therefrom is applied to normalization converter 58. The output signal from normalization converter 58 has the value of an integral power of 2 as for example 2.sup.N. This signal applied to dividing circuits 60 causes division of the I and Q signals by a requisite power of 2. This is accomplished by computing the weighted average of the pulse amplitude over the previous three range bins and normalizing the I and Q signals to this weighted average. The normalization computer identifies the most significant digit (MSD) of the weighted average at each range bin. Since the weighted average is at most an 11 bit number, the following selection of the normalization number is made according to Table 1 below.

If I and Q are expressed as magnitude and sign, division by a power of 2 represents a right or left shift. The modified I and Q signals are then sent to the doppler processor. Since no phase shifts are involved in this operate the ratio of the main lobe to the side lobes in the velocity dimension will not be affected. The six most significant digits of the characteristic are decoded as shown in Table 2.

For instance the number 101 after decoding becomes 100 as:

its decoded value is 301

the sum of the decoded values gives:

3 becomes binary 011

0 becomes binary 000

1 becomes binary 001/100

The above process causes the received to behave exactly like a Dicke-Fix log receiver in jamming and yet achieve clutter cancellation down to either the receiver noise level or the in-band jamming level whichever is larger when both clutter and jamming are present.

The signal output from narrow-band filter 38 is concurrently applied to phase detectors 40 and 42 as heretofore noted and also to a blanking circuit comprising threshold circuit 64, inverter 66, clocked flip-flops 68 and 70 and logical AND gate 72. Referring now to FIG. 3 the operation of the blanking circuit of FIG. 2 will be examined. Threshold circuit 64 maintains a 0 output as long as the input signal to it does not fall, for example, by 5 db or more. The composite input signal to AND gate 72 under this condition is:

thus there will be no blanking signal applied to the doppler processor. Whenever the input signal to threshold circuit 64 falls to a predetermined level, here for our example, -5 db, the output signal from threshold circuit 64 changes from 0 to 1 thereby making the set input S.sub.1 of flip-flop 68 to true or one and the reset input R.sub.1, because of the action of inverter 66, becomes false or zero. However, since flip-flop 68 is a clocked flip-flop, it cannot change state until it receives a clock pulse. Upon receipt of the clock pulse labeled 1 in FIG. 3, Q.sub.1 of flip-flop 68 assumes the 1 state and therefore Q.sub.1 = 0 . The composite input signal to AND gate 72 is now:

thus there will still be no blanking signal. Q.sub.2 cannot change state until the next clock pulse. Upon receipt of the second clock pulse labeled 2 in FIG. 2, we have S.sub.1 still equal to one, Q.sub.1 still equal to one and Q.sub.2 now becoming 1. The composite input signal to AND gate 72 is now:

so still there is no blanking siganl to be applied. This condition of signal input to AND gate 72 will be maintained until the signal pulls up to and above the threshold level of typically -5 db. This state of affairs occurs in FIG. 3 between clock pulses 6 and 7. R.sub.1 immediately assumes the 1 state since the output of threshold circuit 64 is now zero whence S.sub.1 is also zero so that through the action of inverter 66, R.sub.1 = 1. However, the flip-flop 68 cannot change state until the clock pulse is received, in this instance, clock pulse number 7 in FIG. 3. Upon receipt of the clock pulse, since R.sub.1 is true or one, Q.sub.1 becomes true or one and Q.sub.2 remains true or one. The composite input signal to AND gate 72 is now:

thus supplying a blanking signal to the doppler processor. This blanking signal from AND gate 72 will only be applied for one range bin time and will blank that range bin for that particular pulse burst since upon receipt of the clock pulse labeled 8 in FIG. 3 Q.sub.2 will assume the false or zero state thus making the composite input signal to AND gate 72 become:

and hence supplying no blanking signal as output of AND gate 72. A blanking signal to the doppler processor of this kind is required because when the IF signal level input decreases, the receiver sensitivity decreases for a short period of time, however, when the IF signal level increases, the receiver perceives this increase as the presence of a target and therefore, absent a blanking signal, the system is presented with a false alarm.

In summarizing the response of the CFAR-doppler system of the invention it is interesting to consider some specific forms of often encountered electromagnetic counter measures (ECM) or jamming. Under the influence of a broadband noise jammer, the noise level in the conventional receiver varies considerably thus causing numerous false alarms. On the contrary, due to the hard limiter incorporated into the system of the invention, the receiver noise level remains constant regardless of the input jamming level unless the jamming level before the hard limiter is some number, say N db, above the normal receiver level in the narrow-bandwidth filter in which case the receiver sensitivity is reduced by N db. It is to be emphasized that the jamming noise must be in the narrow-bandwidth filter for this latter sensitivity reduction to occur. If the ECM environment encountered comprises a swept jammer, as indicated at FIG. 5(a) sweeping, for example, over a 200 megahertz radio frequency bandwidth once every 10 microseconds at say +60 db above rms noise, a conventional receiver would give a very large, say of the order of +40 db above noise, false alarm pulse every 10 microseconds thus rendering the receiver totally ineffective. In the system as exemplified by the invention, this swept jammer causes no output or sensitivity loss when it is outside the wide bandwidth hard limiter assumed to be 10 megahertz for the sake of example in FIG. 5(b). If, however, the jammer is within the wide band hard limiter but outside the 1 megahertz narrow band filter, a signal suppression will be suffered once every 10 microseconds for a period of about 0.5 microseconds so that for a period of time given by

of the time the system will be desensitized and the other 95% of the time the system of the invention will exhibit normal sensitivity as indicated in FIG. 4(b). In case the ECM environment encountered is a pulse jammer there are three cases to consider.

In the conventional receiver, pulses will be attenuated 20 to 40 db but large pulses will still cause false alarms. On the other hand, because of the limiting action of the invented system, no pulse of this kind would get through the receiver.

A conventional receiver will attenuate a pulse such as this only a few db, say 10 to 20 db, so very large pulses will cause false alarms. In the receiver as exemplified by the invention, the noise at the output of the narrow band filter will be reduced by about 60 db during the length of the pulse. The trailing edge of this pulse may be detected by the FTC circuit in the output, however, its effect will be greatly attenuated by the action of the blanking circuit at the output of the narrow band filter 38 of FIG. 2.

A conventional receiver would give a large pulse of the order of 60 db whereas the system of the invention would give a pulse of about 10 db due to the limiting action. It sould be noted that such pulse matching to the narrow band filter is highly unlikely.

In the final case to the considered here of off-frequency narrow band jamming, the jamming signal and the response curve of a conventional receiver are indicated in FIG. 5(a). If the jamming is, for instance, +30 db above noise, the conventional receiver will be desensitized by 25 or more db. On the other hand, the system of the invention will suffer no degradation as apparent from FIG. 5(b) except when the jammer is first turned on FIG. 5(c) and then for only a period of about 2.tau. as indicated at FIG. 5(c), where .tau. is radar pulse width.

There has thus been described a constant false alarm rate-doppler radar system capable of effective operation in various kinds of electronic countermeasures (ECM) environments.