Volume Surveillance Radar Frequency Selection

VOLUME SURVEILLANCE RADAR FREQUENCY SELECTION Dx Robert J. Galejs, MIT Lincoln Luboratory, Lexington, MA, USA

Abstract

stressing the clutter and interference rejection capabilities of these radars. Bandwidth requirements are very modest at 1 MHz, if no missions other than volume surveillance are required. Very low life-cycle costs are desired as well.

The US Navy developed the volume surveillance radar concept to fill the need for voIume surveillance in future non-AEGIS ships. This new radar is envisioned as a lowcost replacement for the SPS-48/49 radars. A phased array architecture is desirable for flexibility, ship integration ease and ship signature goals.

In order to size the radars across frequency, an assumption must be made as to the T/R module output power. Table 1 outlines the T/R module power levels assumed. These powers are based on readily available commercialoff-the-shelf (COTS) components. Advances continue to be made in this area and this table will likely soon be out of date. However, the important factor, as far as frequency trades go, is that the average power of the assumed T/R modules scales as llfrequency.

This paper discusses the many trade-offs one must consider when deciding on the frequency choice for a volume surveillance radar. Some considerations favor higher frequencies and some favor lower frequencies and depending on how one weights these factors, different conclusions can be made as to the “optimum” frequency choice.

Figure 1 illustrates the RCS assumptions mentioned above. The RCS is assumed to be constant above 1 GHz and to rise as 1/f2 below 1 GHz.

The frequency choice in this instance was driven by three main factors: cost, pulse repetition frequency (PRF) selection and beamwidth. The trade-off between these factors resulted in L-band as the preferred frequency.

A further assumption is made that the radar noise figure plus losses is constant across frequency. This is certainly not an exact assumption but small differences in sensitivity will not ultimately decide what frequency band is best.

Assumptions For the purposes of this frequency trade, it was assumed that radars at all frequencies would be 4-faced active phased arrays. In order to keep costs down, only lowcost T/R module designs (roughly US$lk/module) were considered. Since power amplifier cost is a major contributor to the module cost, this assumption results in consideration only of power levels consistent with a single amplifier in the final amplification stage.

Compiling all of the preceding assumptions into a radar size is done in Table 2. Here it is shown that the PA product is constant for frequencies above 1 GHz and reduced at lower frequencies due to the target RCS. The radar size peaks at 10 m2 at 1 GHz and falls at both higher and lower frequencies.

The radar designs considered here all have free-space sensitivities equal to that of a 16 kW 10 m2 L-band array. For frequencies above L-band, this restriction amounts to having constant power-aperture (PA) product. Below Lband, the PA product is allowed to go down due to the assumption of an enhanced radar cross section (RCS) in this regime. The arrays that result will be compared with a variety of measures of performance.

There are several factors that favor lower frequency operation. These include: cost, PRF choice for clutter rejection, atmospheric and rain attenuation and rain clutter. Of these, cost and PRF selection are of critical importance, attenuation is very important and rain clutter is of lower importance.

The basic requirements that this volume surveillance radar must meet are as follows: A detection range of 75 km is desired -with an instrumented range of 460 km. Noise limited performance is desired in the littoral environment,

There are also several factors which favor higher frequency. These include beamwidth, land clutter and bandwidth. Of these three, beamwidth is of critical importance while the others are of lower importance.

Frequency Selection Factors

This work was sponsored by the United States Navy under Air Force Contract FI 9628-95-C-0002. Opinions, interpretations. conclusions and recommendations are those of the author and are not necessarily endorsed by the United States Air Force or United States Navy.

187 0-7803-5776-O/OO/$lO.OO 0 (2000IEEE)

IEEE INTERNATIONAL RADAR CONFERENCE

Atmospheric and Rain Attenuation

Module Cost Total radar costs are very hard to quantify, so only the contribution from T/R modules is considered. Under the assumptions of fully populated 4-face active arrays, the number of T/R modules is easily found. Under the low-cost T/R module assumption of US$lOOO per module, the total module cost is easily found. This cost is plotted in Figure 2. In order to make some decisions about which frequencies are preferable over others, some thresholds must be defined. In this case, it was desired for the total cost of the volume surveillance radar to be below US$lOM. If the module cost exceeded half of the total cost, that was judged to be “unacceptable”. If it fell to 20% of the total cost, that was judged to be “very good.” Between these limits was ”acceptable.” The frequencies at which these thresholds are crossed are indicated with dashed vertical lines. Thus, one sees that above 2 GHz, module cost becomes excessive and below 1 GHz it is easily affordable.

PRF Selection There are many considerations which influence the choice of PRF. First, there is the desire to maintain a relatively large unambiguous velocity to restrict the region where rain and bird clutter exists to roughly 25% of the unambiguous Doppler space. This yields a limit of approximately 200 m/s as the minimum acceptable velocity ambiguity. This restriction amounts to a requirement that the PRF grows linearly with frequency. In addition, a relatively large number of pulses per coherent processing interval are desired (8 in this case) to allow adequate Doppler filtering. This yields a PRF requirement that varies as f2/A where A is the antenna area. The highest PRF requirement of these two is plotted vs. frequency in Figure 3. The slope discontinuity at 2.5 GHz is the crossover between these two requirements.

Atmospheric and rain attenuation become significant only at relatively high frequencies. Figure 4 plots atmo-, spheric and rain attenuation due to oxygen and water vapor. At L-band and higher, 0 2 absorption is nearly constant and water vapor absorption only becomes significant at X-band. Nearly all frequencies are “acceptable,” while frequencies below 1 GHz are “very good.” Figure 5 plots rain attenuation for rain rates of 1,4 and 16 mm/hr. Rain attenuation is not significant below C-band. Thus, frequencies below 5 GHz are judged to be “very good,” 5-7 GHz are “acceptable” and above 7 GHz is “unacceptable.”

Clutter Levels Figure 6 plots the single pulse clutter levels due to rain and land vs. frequency. Rain clutter dramatically increases at higher frequencies but is overwhelmed by the land clutter requirement. The land clutter decreases slightly with frequency but is not significant enough to distinguish between frequency bands. All frequencies are “acceptable” when considering single pulse clutter returns.

Beamwidth The beamwidth of the radar as outlined in the introduction is plotted in Figure 7. The beamwidth decreases with increasing frequency from about 5 degrees at 1 GHz down to about 1 degree at 10 GHz and increases dramatically to 15 degrees at UHF. A smaller beamwidth has many advantages: It allows accurate handover to a fire control radar, reduces vulnerability to main beam jamming of the radar and enhances elevation estimation for low altitude targets in multipath. Figure 8 plots the number of 2’ fire control radar (FCR) beams required to search the handover uncertainty from the surveillance radar to achieve a high-confidence handover of 95%.The best one can achieve is a single beam handover which occurs above 900 MHz, the “very good” region. Only a few beams are required down to UHF. Below UHF is “unacceptable”.

To establish thresholds in this situation, consider the effects of range ambiguities. If the unambiguous range exceeds the instrumented range, then there are no clutter folding effects to consider at any range. PRFs that meet this requirement are considered “very good.” If the unambiguous range exceeds the 75 km detection range, then there are no clutter folding effects out to the desired detection range. This is termed “acceptable.” When the ambiguous range becomes shorter, near-in clutter starts to compete with targets within the required detection range, dramatically increasing the required clutter rejection requirements (roughly 25 dB here). This is “unacceptable.” Above 1.5 GHz is “unacceptable” and below 250 MHz is “very good.”

Figure 9 plots the angle off of boresite that a 1 kW/ MHz jammer at 100 km range causes a 25 dB jammer to noise ratio in the radar. The thresholds for this measure of performance are likely to be the most controversial. I took the relatively modest threshold of 5 degrees as the “very good” point and 10 degrees as “unacceptable”. This results in the acceptable region spanning from 800 MHz to 2 GHz.

188 IEEE INTERNATIONAL RADAR CONFERENCE

Figures

In many situations, a 20 degree jammed sector would likely be “unacceptable”, but that was chosen here. If a much smaller jammed sector was desired and this factor weighed heavily in the frequency choice decision, then perhaps higher frequencies would be chosen as optimum. 15

When a low altitude target is being tracked, multipath errors will corrupt the elevation information about that target. To illustrate the point, a simple model of a target flying level at 1000 m above the ocean with a 0.15 m RMS waveheight is shown in Figure 10. A flat Earth was assumed with a radar height of 15 m. The maximum elevation bias at any range is plotted vs. frequency. When that bias exceeds the FCR beamwidth of 2 degrees, it was deemed “unacceptable.” This occurs below 800 MHz. “Very good” performance is found at 2 GHz and up.

10 5

0 -5

1.o Frequency (GHz)

0.1

Bandwidth

10.0

Figure 1. Target RCS Assumptions

As the radar frequency increases, the available bandwidth in radar bands generally increases as well. However, since the bandwidth requirement is quite minimal at 1 MHz, all frequency bands have adequate bandwidth for the volume surveillance radar. The NTIAl-allocated radar bandwidths are plotted in Figure 11 to illustrate this point.

10 JlOOOlmodlrle

Summary

Fully Populated Aperture ConstamFrsaSpace

*

Search Sensitivity

-$ 6

One can see from all of the plots presented so far that there is no radar frequency that allows performance in the “very good” regime for all selection factors. Thus, rather than maximizing the good aspects, one is left with the case of minimizing the bad. Figure 12 plots the “unacceptable” regions for the 6 most important factors considered: electronic countermeasures (ECM),cost, handover, clutter rejection, multipath bias and rain attenuation. The volume surveillance radar frequency should not be higher than Lband due to module cost and clutter rejection considerations. Below L-band, the radar beamwidth becomes very large. One can see from this figure that L-band is a reasonable compromise between many competing factors for a volume surveillance radar.

0.1

1 I

1 1

I

1.o Frequency (GHz)

10.0

Figure 2. T/RModule Cost

1. National Telecommunications and Information Admin-

istration.

189 IEEE INTERNATIONAL RADAR CONFERENCE

Unacceptable

0.1

1.o

Frequency (GHz)

Frequency (GHz)

10.0

Figure 5. Rain Attenuation Figure 3. PRF Requirements

6or=====l

h

40 L

0.1

AcceDtable

20

-

0.1 __ -

1.o

10.0

Frequency (GHz)

Figure 6. Clutter Levels Figure 4. Atmospheric Attenuation

20 15

-

10

-

5

-

Search Sensitivity

-

,- .. 0

0.1

1.o Frequency (GHz)

Figure 7. Beamwidth

190

IEEE INTERNATIONAL RADAR CONFERENCE

10.0

10000 From NTlA 1996 US Frequency Allocation Cha Radiolocation Bands

Frequency (GHz)

Frequency(GHz)

Figure 8. Handover to Fire Control

Figure 11. Frequency Bands Allocated to Radar

h Handover

Clutter

Multipath Bias

0.1

1.o Frequency (GHz)

10.0

Figure 12. Frequency Trade Summary

1 .o

0.1

10.0

Frequency (GHz)

Figure 9. Jammed Sector

Frequency (GHz)

Figure 10. Multipath Elevation Bias

191

IEEE INTERNATIONAL RADAR CONFERENCE

Tables TABLE 1. T/RModule Power Assumptions

I UHF I L-Band I S-Band I C-Band I X-Band I I

Peak Power (W)

I

1000

I

180

Duty Factor

0.06

0.10

Average Power (W)

60

18

~~

~~

I

I

83 0.10

I

I

50

1

10

0.10

0.25

5

2.5

I 8.3

~

TABLE 2. Radar Sizing

UHF L-Band Number of modules per m2

S-Band

C-Band

X-Band

87

400

1111

4444

Antenna Area (m2)

8.0

10.0

6.8

5.3

3.7

Power Aperture (kWm2)

35

160

160

160

160

192

IEEE INTERNATIONAL RADAR CONFERENCE