Electronic methods of steering beams, communication between radars and advanced processing give modern radar systems considerable flexibility.
1- INTRODUCTION
In this project we look at some current radar technology and find out how it is being used. The antenna has been the subject of some of the most interesting radar developments in recent years. Domestic satellite dishes are representative of the old level of antenna technology because they must be mechanically pointed at their target. Radar versions of these parabolic reflectors are rotated by heavy and expensive (but reliable) turning gear, and a target can only be observed once every few seconds when the dish faces that direction.
At the centre of a reflecting antenna, such as a satellite dish, is the device that illuminates it, called the feed. Feeds can be dipoles, microwave horns or other devices for radiating or receiving electromagnetic waves. The purpose of the reflector is merely to increase the collecting area, and hence the gain, of the antenna. However, a large collecting area can also be achieved in an entirely different way. It is possible to arrange many dipoles on a flat plate
and connect them together in a phase-coherent manner to form a phased array. In the UK, one domestic system for receiving satellite broadcasts was constructed in this way.
Phased arrays are often considerably more expensive than reflector antennas but they have the advantage that the beam can be steered electronically rather than mechanically. If the phasing of a receiving antenna is carried out digitally by computer calculation, rather than by analogue devices (which may be controlled by a computer), the process is known as digital beamforming, and there is considerable flexibility for beamshaping and the formation of multiple beams.
Phased arrays are now in widespread use for receiving, but transmitting versions are only just coming into service because of the difficulties of building, and controlling, large arrays of small transmitting modules. These active arrays represent the current level of technology for single radar systems.
An entirely separate development area has been concerned with netting independent radars (and other sensors) together to improve target detection and tracking accuracy. We briefly examine this field, which is sometimes known as multihead radar. Finally, we take a brief overview of the recent advances made in signal and data processing, which have enabled radars to go beyond their original function of radio detection and ranging, and discussing the stealth applications.
2- PHASED ARRAYS
An array of radiating elements can be one-dimensional (linear array) or two-dimensional (planar array). Arrays work the same way on both transmit and receive, but it is often easier to follow the theory for the transmitting case because of the analogy in optics with Young's double slit experiment and diffraction gratings.
A simple linear array, with element spacing d, is shown in (Fig. 1). If all the elements are excited in phase, then a wavefront is radiated broadside to the array. If a successive time delay is introduced across the array, causing a phase change
y between elements, then the wavefront will be radiated at an angle q given by
The field Ea(q ) from N elements can be found by summing across the array. Using a little trigonometry the summation comes out as
Where E1 (
q ) = field due to a single element.
I
n optics, the field arising from a diffraction grating has two components; the diffraction pattern from a single slit and the interference pattern created by the interaction of coherent signals from different slits. It is exactly the same with antenna arrays. Each element of the array acts as a source with its own diffraction pattern E1 (q ), and the radiation from the sources interacts to form an interference pattern, which is usually called the array factor (AF) given by
The radiating elements are usually simple devices such as dipoles, slots, patches or some times small horns. All these elements have fairly wide diffraction patterns and the antenna pattern is determined by the much narrower array factor. Even so, care must be taken not to steer the array beam outside the pattern of the individual radiating elements. A normalised array pattern is shown in (Fig. 2) in comparison with the diffraction pattern of a single half-wave dipole.
Part of the purpose in rearranging (Eq. 2) to get (Eq. 3) is because the term in large parentheses in (Eq. 3) can be shown to have a maximum value of unity. This means that the field strength has been increased by£ N times the field from a single element, and therefore the power gain of the array has been increased by N. Some other properties of array patterns are now outlined.
The half-power beam width of the pattern steered off-broadside
through an angle q can be
found from (Eq. 3) by setting the expression in large parentheses to Ö 0.5. For large N, this leads to
For the broadside pattern this reduces to
Where D = overall length of the array = Nd. This, of course, corresponds to the l /d approximation for antenna beam-width. The width increases by (1 /cosq ) as the beam is steered away from broadside because the effective length of the array is reduced by (cos q ) when viewed from an angle q off-bore-sight.
A convenient approximation of (Eq. 5) in degrees, to remember for rough calculations, is
Nulls occur in the pattern at angle
q null given by
K=1,2,…
Often, element spacing near l /2 are used, giving only one main-lobe (plus one to the ear, if the array is not mounted above a reflecting plane). For wider element spacing, grating lobes (other principal maxima, or "secondary beams") occur when
K=1,2,…
With careful design, the individual element pattern can sometimes be used to null out grating lobes near 90
° , as shown in (Fig. 2). Side-lobes (secondary maxima) occur at
K=1,2,…
The first side-lobe has a power gain of - 13.4 dB with respect to the peak gain of the array. This is a well-known result arising from a rectangular illumination function. The FT of a rectangular function (in this case the aperture) gives rise to a (sin x)/x function of a similar form to the array factor. Equation (3) can be approximated, for small values of q , by an expression of the form
Thus the far-field pattern of the array can be represented by the FT of the spatial illumination function, provided that the element spacing is close enough to avoid grating lobes. For this reason, FTs are used extensively in antenna theory.
Confusion at this point may happen because we have only used the FT to transform between the time domain and the frequency domain. Table (1) shows how to relate this new way of using the FT to the more usual time/frequency form.
If all elements are fed equally, an array has a rectangular illumination creating side-lobes only 13.4 dB smaller than the main-lobe, as discussed above. Side-lobes this large are generally fatal for a radar system, owing to the difficulty of separating small targets in the main-lobe from larger targets in a side-lobe.
Low side-lobes are achieved through accurate element and array design and the application of a weighting or window function, such as triangular, raised cosine, Dolph-Chebyshev polynomials, etc., to the array. The effect of weighting is such that the elements at the centre of the array are fully utilised but those towards the edge are increasingly attenuated, thereby causing a wider beam and a loss of gain and SNR; unfortunately, these are necessary afflictions.
.
The processes of beam forming and beam-steering are identical. A progressive time delay is applied to the elements, creating the tilted wave front shown in (Fig. 4). A beam steered through an angle q s, has an array factor of
As an alternative to using a time delay, at a single frequency the steering can be applied to incrementing the phase between elements by
Grating lobes can be avoided, even when steering, if the element spacing is restricted. Rearranging (Eq. 8) gives the spacing limit for a steering angle q s
Both time-delay and phase-delay beam forming and steering are used. Analogue time-delay beam-formers are somewhat slow and clumsy, involving the use of relays to switch lengths of cable in and out, but there are no frequency restrictions and simultaneous operation over a wide band is possible. Phase delays can be introduced by a variety of means, illustrated in (Fig. 6), which include:
* Diodes to switch between alternative path lengths or to change the reactance of loading stubs.
* Ferrite phase-shifters; these use a cylindrical ferrite rod, inserted in a waveguide, which can be magnetised.
* An array of hybrids, known as a Butler matrix, arranged in a similar way to an FFT algorithm.
* A pre-set resistor matrix combining signals of different phases to give the appropriate value.
Surprisingly, it is not necessary to maintain precise phase and amplitude values at each element on the array. The summation of the field from many elements tends to cancel out small errors, and consequently phases are approximated in binary steps, often with as few as three bits giving 45° steps in phase. The loss in main-beam gain from a 3-bit phase shifter is only 0.23 dB, falling to 0.01 dB for a 5-bit device.
The problem with steering by phase rather than by time delay, is that changing the radar wavelength alters the steering angle. This frequency dependence has been exploited in the past as a method of electronic scanning (a given frequency corresponds to a given direction), but with modern electronic warfare the radar designer needs the liberty to use frequency in a much freer way. The upshot of this wavelength dependence is that most modern phased array radars are restricted in instantaneous bandwidth, with a limit (as a percentage of the carrier frequency) roughly equal to the beam-width in. This limit equates to a pulse length restriction related to the aperture size.
So far we have only discussed linear arrays, while most radars use two-dimensional planar arrays. In many respects, planar arrays are similar to linear arrays. The simplest way to develop a planar array is to form a stack of linear array planks and vary the phase between the planks to beam-form in the Second dimension. Other designs use triangular or rectangular lattice layouts Of elements on the plane. The increased complexity of these arrangements requires larger computational resources to control the phase shifters and creates something of a wiring distribution problem.
The wiring problem has been tackled in recent years by using fibre optics to control the phase shifters and, in a further development, optical beam formers have been constructed. Optical fibres avoid most of the problems of RF cables, being lightweight, lost-cost, low-loss and immune to interference and cross talk. These attractive properties of optical transmission in manipulating microwave signals have given rise to an emerging technology known as microwave optics.
One of the additional problems of planar arrays is that moving away from the normal causes a form of conical distortion because azimuth and elevation measurements are no longer independent. Corrections have to be made when tracking off-axis targets, which again adds to the complexity of the system and restricts performance at large angles off-bore-sight.
3- DIGITAL BEAMFORMING
In general, the increase in flexibility of radar antennas is matched by an increase in cost. The early mechanically rotating designs are the least expensive, followed by mechanically rotated arrays that beam-form electronically in elevation. A full three-dimensional radar (a two-dimensional array plus range) using an analogue beam-former is very expensive and the beams are relatively difficult to set up. But there is still one further degree of adaptability (and cost) to be gained from using arrays. The ultimate flexibility is to have one receiver and A/D converter per element and to form beams by performing calculations in a computer on the A/D samples. This is digital beam-forming. Sometimes, the amount of hardware needed can be reduced by combining local groups of elements to form sub-arrays, and using a receiver for each of these.
There are tremendous advantages to digital beam forming:
The disadvantages of digital beam forming are high cost and performance limitations. The latter are determined mainly by the technology of the A/D converters. Slow converters with few bits make an array narrow band with a limited dynamic range. As technology improves, so does the performance of this type of array.
A digital beam forming system usually has a relatively simple receiver for each element, which down-converts the frequency into I and Q (in-phase and quadrate) channels for the A/D converter (see Fig. 7). Real-time beam forming takes place by multiplying these complex pairs of samples by appropriate weights in multiply/accumulate integrated circuits (ICs). The array output is formed from
where Vn is the complex signal from nth channel, wn is the weighting coefficient, e-j2
p n(d/l )sinq is the steering phase shift and C is the correction factor. Corrections are necessary for several reasons. These include errors in the position of the element, temperature effects and the difference in behaviour between those elements embedded in the array and those near the edge, caused by coupling effects with neighbouring elements.Digital beam forming is now used extensively for receiving arrays, where sufficient funding is available, and it can be found in use from small HF sea-sensing systems up to large-scale surveillance radars. An example of a system in service is the Patriot radar system
4- ACTIVE ARRAYS
When an array is used for transmitting, as well as receiving, it is said to be an active array, and represents the most up-to-date (and highest-cost) form of antenna. An active array has a number of separate modules, each with its own amplifier and phase control. There are considerable advantages to the use of an array of small transmitting elements (rather than one large transmitter), including the flexibility common to receiving arrays plus the possibility of very large power x aperture products. The disadvantages are those of yet further increase in cost and complexity.
One of the main difficulties of active arrays is that not all phase shifters (ferrites, for example) are reciprocal, and a different phase is received from that transmitted. Furthermore, the transmitting beam may need to be wider than the receiving beams to encompass a block of four monopulse beams. The end result is that transmitting and receiving phase shifters need to be set up independently, leading to designs similar to that shown in (Fig. 8). These modules require solid-state gallium arsenide transmitter ICs as well as the receiving chip set. The power output per module is generally of the order of a few watts.
An example of an active array is the UK multifunction electronically scanned adaptive radar (MESAR) developed by the UK Admiralty DRA (M) and Siemens-Plessey Radar. This is an S-band array of 918 elements using 4-bit phase shifters. Each element uses GaAs ICs for the microwave circuitry and silicon ICs for control. Array thinning is used to improve the transmitted side-lobe level.
The MESAR design demonstrates the Current level Of technology. Generally speaking, very-high-speed Silicon ICs (VHSICs) are needed to carry out all the calculations and gallium arsenide microwave monolithic ICs (GaAs MMIC) are needed for the RF components and fast A/D converters. The development of these technologies, and of fibre optic control, will be crucial to the future of both digital beam forming and active arrays.
5- MULTIFUNCTION RADAR
Given the flexibility of phased array radar, the traditionally separate roles of surveillance, tracking and even radio communications can be taken over by a single multifunction radar system. The system's resources (beams, receivers, and computers) must be allocated efficiently to carry out all the required functions, which might include:
Of course, the problem is the same as with every surveillance radar-there is never enough time to do everything, and a priority list must be drawn up for the beam scheduling. Making such complex, weighted decisions in a computer is quite difficult and is generally held to be within the area of artificial intelligence (AI). Fortunately, modern software languages, are structured such that antenna array operations can be written as software modules, which can be called and run by the scheduler with a predetermined priority list.
6- MULTIHEAD RADAR
When several monostatic radars can see the same target, as shown in (Fig. 9), their outputs can be combined to improve the tracking. This method of netting independent radars together is sometimes known as multihead radar and has become popular since the development of digital communications enabled reliable low-cost data links to be formed.
Multihead radar is especially applicable for air defence and ATC radars with over lapping cover, but other situations where it may be applied include naval systems where several radars, and other sensors, may be combined within the area of a fleet.
The outputs from several sensors must be sent to a data processing centre for combining. There are then three ways in which the data may be combined:
1. Individual plots from each radar can be transmitted to the data processing centre and the tracker can make use of all the data simultaneously. This gives the best results, but requires high-bandwidth data links and a lot of computer processing power.
2. Individual plots can be sent to the centre and combined together before track forming. This requires a wide band data link but it reduces the computational load and still gives quite good results.
3. Tracks can be formed locally at each radar and sent to the centre for combining; this requires only a low-capacity data link but the improvement in tracking accuracy is modest.
There are other advantages to multihead radar in addition to improved tracking accuracy. Low-flying aircraft in mountainous country will only be detected by radars in short glimpses, but netting several radars together, to view the target across different terrain, increases the probability of detection and of forming some sort of assessment of the threat. A multihead radar system is also more jammer-resistant because each radar views the target from a different angle.
In some ways, multihead radar networks are more attractive than multistatic arrangements. Each radar operates independently and causes only a relatively small degradation to the system if it should fail or be jammed. Multihead is also much simpler than multistatic because no synchronisation is required and different radars, from different manufacturers, can readily be netted together.
Merging data in multihead systems is not without difficulties. For example, giving each measurement a weight that is inversely proportional to its variance usually combines data. In this way the tracker takes more account of good data than bad. A problem occurs when one radar correctly detects a target manoeuvre and uses a filter with a larger variance in order to keep track. The data from this radar will receive less weight than those radars that have not yet spotted the manoeuvre. This is typical of the type of quandary that can be so interesting for a radar engineer to work on.
7- HIGH RESOLUTION RADAR TECHNIQUES
Modern radar has gone beyond 'radio detection and ranging' to incorporate target Classification and recognition algorithms. These facilities require improvements in evolution to identify distinguishing features on targets. We have seen throughout this book how the resolution depends on the system parameters, but we have not yet mentioned the contributions that can be made by developing the hardware, the signal processing and the data processing software.
There are many methods of increasing the resolution of radar systems. Often, techniques were developed in other fields of research and adapted for use with radar. Some of the current approaches involving changes in hardware, or data collection strategies, is discussed below.
Wide band radar systems High range resolution can sometimes be used to identify features on a target, if a sufficiently large bandwidth can be obtained. Wide band techniques include carrier-free radar, pulse to pulse frequency modulation, noise modulation, wide band chirp, etc. The main disadvantage is that the increased noise bandwidth usually restricts this technique to short range applications.
Multifrequency illumination Rather than having a single radar channel of wide bandwidth, it is possible to select a number of narrower bandwidth channels and correlate the outputs to look for certain spatial properties of the target. This subject can be quite powerful when a priori information on the target shape is available, although it is somewhat sensitive to aspect angle.
Multipolarization techniques It has been realised that more information on many types of target can be improved if multipolarization techniques are used. This arises because the polarization signature of targets has, in principle, five degrees of freedom (three amplitudes and two relative phases). In contrast, frequency diversity yields only a single extra degree of freedom. Evaluation of the polarization characteristics of targets is an active area of research, and radars to exploit these polarization signatures are under development.
Quadpolarized systems Use of several frequencies and full polarization information is a characteristic of the latest generation of airborne remote sensing radars. These quadpolarized systems are capable of recognizing the different scattering mechanisms contributing to the backscatter. This can be utilized in target recognition and classification. Some of the most significant progress in understanding the scattering properties of extended targets and in the application of SAR is likely to stem from this advance.
Interferometry A further advance in the use of SAR is in an interferometric mode. In this mode, the phase difference between SAR images gathered on closely spaced orbits can be used to construct a map of the topography of the imaged area. Tests on Seasat and ERS-1 data have indicated that it should be a practicable technique. Very curious results were also obtained over flat agricultural land, and there is clearly more investigation needed. These results have led to a flurry of development of multi-antenna interferometric systems for aircraft operation, and SAR processors giving high phase accuracy.
Below we describe some of the software and mathematical techniques that can be used to increase radar resolution.
Inverse synthetic aperture radar ISAR is used to increase angular resolution in the same way as SAR but with the difference that the radar remains stationary and advantage is taken of the target motion.
Improved Doppler spectrum analysis There are many modern spectral estimators, such as the maximum likelihood method, maximum entropy method, minimum eigenvector method, etc. These can be used to improve the doppler resolution and to identify targets by resolving such features as turbine blade rotation, or the pitch, roll and yaw of an aircraft or ship. Improved spectral estimators are also applied to phased arrays to improve the angular information.
Radar cross-section fluctuation analysis When a target is composed of a limited number of individual scatters that characterize it, the target may be identified by statistical or spectral analysis of the RCS amplitude fluctuations.
Super-resolution Even when the shape of a target is smaller than the range resolution, some information may be recovered from the echo by carrying out a deconvolution process known as super-resolution, provided that the, "point spread function" ( = impulse response) of the system is known. Essentially, the radar bandwidth is artificially increased by restricting the possible choice of target shapes.
Topography The structure of a given target maps into a particular shape in the ambiguity diagram (the convolution of the target structure with the pulse shape in range and doppler). Different chirp rates correspond to different angles on the ambiguity plane and so may be used to gain information on the target shape, in a similar way to topographic imaging used in medicine.
Many of these techniques provide good results in computer simulations, but in practice the effects of noise and phase errors due to unresolved motions may lead to degradation of performance. The best strategies often involve the combination of several of the techniques above. If the objective is to identify a small target, such as an aircraft or ship, it can be useful to include a 'library' or 'catalogue' of possible target shapes. Some means of relating high resolution observations to the library is then required and these days there is a lot of interest in neural nets which can be trained to recognize features in the data that relate to certain types of target.
8- STEALTH APPLICATIONS
Stealth is the art of concealing targets from detectors. For submarines, it is important to be acoustically Stealthy; ships and aircraft probably need to balance acoustic, visual, infrared and radio Stealth in proportion to the threat posed by the sensors in each of these spectra. The threat of radar detection is usually a major concern to an attacking force and a lot of effort is put into trying to reduce radar signatures.
The subject of Stealth arose from the analysis of Allied bomber losses during World War II in which the tiny, but very fast, wooden Mosquitoes (carrying a bomb load almost as large as a B17) fared very much better than contemporary larger and slower metal bombers. After the war, the Northrop 'Flying Wing' and the Avro Vulcan bomber showed that the aircraft shape, as well as the construction material was important in the reduction of the Radar Cross Section (RCS).
A change in tactics also occurred after the war; the large defensive bomber formations, and escorting fighters, were replaced by solo nuclear missions in which single bombers were required to navigate through enemy defences alone. Escaping enemy ground fire by flying at very high altitudes was eventually abandoned after the U2 of Francis Gary Powers was shot down in 1960. Flying at low altitudes was dangerous because of the accuracy of radar-controlled anti-aircraft artillery, unless the aircraft flew very low to get beneath the radar defences. The solution lay in adopting an idea tried in World War 1, to make aircraft 'invisible' by covering the airframes in translucent materials, and then extending the concept to include the radar spectrum.
The maximum radar detection range can be summarized as
The RCS
s must be changed dramatically if the radar performance is to be seriously affected. Computer modelling of airframes and ship structures shows that most of the scattering comes from flat surfaces and corner reflectors, formed where different parts of the structure join together. The RCS of most structures can be dramatically reduced by choosing an inherently low RCS design and then using smooth, continuously curved shapes and avoiding any right angles. In the case of an aircraft, this avoidance of right angles may mean that a single fin cannot be used and the aircraft must be designed without one (the flying wing approach used in the B2 bomber) or with two inclined fins (as used on the F117 fighter).After shaping, any remaining 'hot spots', such as the reflections from engine air intakes on aircraft, can be covered with radar-absorbing material (RAM). RAM is generally some form of lossy dielectric material built into a wafer, which is tuned for maximum absorption in a particular radar band. Care also needs to be taken to ensure that the radar antennas on ships and aircraft do not act as RCS hot spots. In a battle scenario, Stealth is a very effective counter to radar, especially when other Electronic Countermeasures (ECM) activities are going on to reduce the radar efficiency. There are, however, a few antidotes to Stealth. Wide band carrier-free impulse radar may be used to try to exploit RCS weaknesses in the radio spectrum and, in general, radar needs to move to lower frequencies to combat Stealth.
"Over-the-horizon" radar may be effective because the scattering mechanism is more related to radar-induced currents flowing throughout the structure of the target than from the individual scattering facets on the surface that dominate the RCS at microwave frequencies. Sky wave OTH radar is particularly useful because it looks down on air targets, the aspect from which they present the largest scattering cross-section. Bistatic radar has also been cited as a possible countermeasure because it is hard to design Stealth shapes against unknown radar geometry.
9- SUMMARY
The future of radar does not lie in larger and more powerful systems, but rather in slightly smaller systems that are more agile, intelligent and difficult to detect because of larger band widths to be used. The resolution of radars, and the number of targets that can be tracked, can be expected to increase as large amounts of low cost computer power become available.
[Home] [Articles] [Resume] [DSP & ES] [Wavelet Shopping] [Wavelet Resources] [Links] [Contact Us]
Send
mail to madani@ieee.org with questions or comments
about this web site.
Copyright © 1999 by Salem AL-Madani. All rights reserved.
