Revised version of paper presented at the International Radar Conference, Brest, France, May 17 to 20th 1999
Any assessment of overall radar performance must include the performance of the radar antenna. When the antenna is installed on a moving vehicle, for example, an airborne platform, the installed rather than the free space antenna characteristics must be employed in the systems analysis. The nature and extent of the performance degradation after installation will depend both upon the type of vehicle, in particular its size and surface structure, and the siting of the antenna. Modelling the installed performance of radar antennas has proved to be very difficult owing to the difficulties in computing the complex interactions between the antenna and the structure. Some work has been reported on arrays installed on aircraft [1]. This paper considers the more difficult problem of modelling the performance of antennas installed on helicopters, and compares the performance of array and reflector antennas.
There are various problems in using radars on a helicopter. The sites which are suitable for installation are very limited and full coverage over 360.0 degrees is not achievable with one antenna unless the antenna can be accommodated under the helicopter. However, this is impractical for the antenna dimensions required for AEW or surveillance radar applications and the more usual sites are
In addition to the siting problems and the interactions between the antenna and the helicopter fuselage, the presence of the moving rotor blades will further perturb the installed antenna performance as compared to free space. In general, the bulk of the energy incident on the main rotor blades will be scattered downwards, thereby degrading the antenna performance in the direction of the earth's surface. The installed performance of the antenna, therefore has implications both for the levels and the distribution of clutter present in the radar receiver.
It is difficult to model the structure of the helicopter. The only effective method of computing the installed performance of a radar antenna on a helicopter is to use diffraction theory, for example, the program ALDAS See 1 but this method relies on modelling the structure geometry using cylinders, ellipsoids, cones and flat plates. While these are very suitable for aircraft, they are not so suitable for helicopters and some care must be taken in the modelling to ensure a good match to the structure. This can be done (See See Figure 1: ).
A sample antenna specification ( See Table 1 ) has been used to generate designs for a reflector and an array and these antennas have been modelled in free space and when installed on a helicopter. The centre of each antenna was placed at 290 mm outboard for the fuselage with the boresight pointing forward, that is, parallel to the side of the fuselage.
Antenna Design Aim |
|
| Frequency | 10 GHz |
| Half-power beamwidth | 5.0 degrees |
| First sidelobe | -25.0 dB |
| Polarisation | Vertical |
The computational modelling of array antennas, while expensive in runtime, is straightforward. For installed performance, it is necessary to model every element with its own radiation pattern and relative amplitude and phase and then sum the installed patterns of every element to obtain the final installed pattern.
The design aim results in an array of 576 (24 by 24) waveguide elements each 16.5 mm on a side. When installed on the side of a helicopter, the radiation patterns of See Figure 2: and See Figure 3: are found. Two rotor positions were used, the second one being rotated by 36.0 degrees with respect to the first set. The presence of the helicopter causes beam squint, a gain reduction of about 1 dB and degradation in the sidelobe levels.
The degradation is better viewed through contour plots. See Figure 4: shows the array in free space. See Figure 5: and See Figure 6: show the contour plots for the array when mounted looking forward and with Rotor positions 1 and 2 respectively. The effect of the rotors is very small. There are some effects at the -38 dB levels in the radiation patterns. The effects can only be seen in the contour plots when the dynamic range of the plot is increased to 60.0 dB.
To model an installed reflector antenna is more difficult than modelling an array and several methods have been developed and have been compared See 2 . These methods vary in accuracy, complexity and in runtime. The one adopted here uses the nearfield planar output from the reflector analysis program, REFLECT See 4 . This program uses Physical Optics (PO) and Physical Theory of Diffraction (PTD) to generate currents on the reflector which are then processed to provide far or nearfield data.
Trials were carried out using as inputs to the diffraction program, ALDAS See 5 , an array of elements in a plane 300 mm from the reflector aperture and the amplitude and phase computed at these points by REFLECT. The results from ALDAS agreed with the farfield results from REFLECT to an accuracy of +/-0.5 dB at -25 dB wrt peak gain, +/- 1.0 at -30 dB and +/- 3 at -40.0 dB. When a plate was included in both computations, the agreement was the same. To meet the design aim of See Table 1 , an offset reflector with an aperture diameter 400mm and focal length 255 mm and an offset angle of 45.0 degrees was employed. The reflector feed was a Potter horn See 3 . A total of 1681 elements had to be used to provide the accuracy quoted above.
See Figure 7: and See Figure 8: compare the patterns found in free space with those when installed on the side of a helicopter. The degradation in Elevation is similar to that undergone by the array but the degradation is worse in Azimuth due to the larger area of elements which had to be included in modelling a reflector.
The runtimes on a PENTIUM Pro 266 MHz are compared in See Table 2 for two principal plane cuts. The runtime for the computation of the array was 145 hours. However the computations need not be carried out in one run but can be subdivided and run on several machines simultaneously. The ALDAS contour plotter includes the ability to add several runs together to form one contour plot.
Comparison of Runtimes (minutes) |
||
| Parameter | Array | Reflector |
|---|---|---|
| No of elements | 576 | 1681 |
| Time (free space) | 3.3 | 1.1 |
| Time (installed) | 106.0 | 498.0 |
A typical operating environment for a radar mounted on a helicopter is maritime AEW See 6 . This involves target detection in the presence of sea clutter. The characteristics of sea clutter are dependent upon a variety of factors including grazing angle, polarisation, wind speed, sea state and the size of the resolution cell and hence upon the pulse width and Doppler filter extent. For large resolution cells, the sea surface scattering behaviour may be represented approximately by an average reflectivity versus grazing angle model, whereas as the resolution cell size becomes smaller, the nature of the clutter tends towards that of discrete targets which vary with time and which are commonly known as sea spikes. Various statistical distributions have been used to represent the spatial and temporal amplitude distributions of sea clutter; these include the Rayleigh and log-normal distributions and more recently the Wiebull and K distributions See 7 . In general, sea surveillance radars employ small range cell sizes to reduce the extent of the clutter return. Fast scanning techniques are employed to provide temporal decorrelation and frequency agility can be used to reduce the coherency of the clutter returns See 7 .
The differences between the free space and installed antenna characteristics for both reflector and array antennas are small in system terms when considering their effects upon the average clutter performance. Although the installed performance reveals main beam squint and loss of gain together with considerable alteration to the sidelobe structure with some regions of increased sidelobe level, the sidelobe envelope for the installed antenna corresponds fairly closely with that for the free space antenna. This is demonstrated in the calculations of average clutter for the free space and installed reflector antenna performance shown as contours of clutter in unambiguous range-velocity space in See Figure 9: and See Figure 10: respectively, and using the reflector antenna models calculated above. These calculations were performed for a pulse-Doppler radar system using the clutter modelling software CASPAR See 8 , with a clutter reflectivity versus grazing angle model representative of sea state 5 and the following radar and platform parameters:
In See Figure 9: and See Figure 10: , only a subset of the total unambiguous range-Doppler extent is shown. For a platform altitude of 1 km, the range to the radar horizon is approximately 130 km assuming a 4/3 effective earth radius and the depression angle at the horizon is 0.88 deg. The PRF of 10 kHz corresponds to an unambiguous range of nearly 15 km. Thus there will be typically 8 or 9 ambiguous range contributions from 1 or 2 Doppler intersections to the clutter return in each range-Doppler cell. The contributions, therefore, are distributed over the visible range extent. The primary differences between the free space and installed antenna performance are an increase in the sidelobe contributions for the installed antenna and a broadening of the main beam clutter. However, these increases are relatively small and increase only marginally the extent of the obscuration due to clutter. This result is in contrast to similar calculations performed for aircraft where, for the cases in which the boresight of an antenna mounted above the fuselage was normal to the direction of motion, the sidelobe blockage provided by the aircraft wing reduced the clutter for the installed antenna as compared to the free space antenna performance See 1 . The effect of the rotors was negligible.
The calculations of installed performance show that taking the antenna boresight so close to the fuselage does degrade performance. The beam is squinted in Elevation due to the presence of the sponson and in Azimuth due to the presence of the fuselage. The degradation is very similar for the use of an array or a reflector antenna, although the runtimes are very much greater for a reflector.
In general, for the antennas considered, the reduction in antenna performance when installed on a helicopter will not degrade significantly the performance in the presence of sea clutter in terms of the average properties of sea clutter. However, the sidelobe structure of the rotating antenna will typically vary with instantaneous position. In some cases, therefore, clutter cancellation or rejection techniques based upon models of the spatial and temporal fluctuations of sea clutter may not be as effective in situ as predictions based upon free space antenna characteristics. The position of the rotor blades has a relatively small effect on the antenna sidelobes but the rotation effects are likely to influence the temporal clutter characteristics. The reduction of gain and the main beam squint associated with the installed antenna will also typically be subject to variation as the antenna rotates. In addition to reduced detection capability, this may limit the accuracy with which target positional information can be obtained.