RADAR EQUATIONS FOR CLUTTER AND JAMMING
3.5 EFFECTS OF DISCRETE CLUTTER
Two types of discrete clutter cause problems in radar that differ from the surface and volume clutter discussed in the previous sections:
Echoes from moving objects such as birds or land vehicles;
Large echoes from fixed objects on the surface, primarily manmade struc-tures.
Radial velocities of moving objects may lie in the response band of Doppler pro-cessors and be passed to the output as targets, overloading the traffic capacity of human operators or track-while-scan channels. Echoes from large fixed objects may exceed the cancellation capability of the signal processor, producing visible clutter or alarms that impede detection and tracking of desired targets. The dis-crete nature of these echoes prevents their rejection by cell-averaging CFAR de-tectors, and other means of rejection may raise the threshold to levels that
sup-Radar Equations for Clutter and Jamming 89 press target detection in several resolution cells including and surrounding that containing the strong clutter.
Models of both types of discrete clutter are discussed in Chapter 9, along with processing methods for its rejection. The effect of discrete clutter is not measured by a reduction in maximum detection range Rm on the desired targets. Instead, if not adequately suppressed in the signal processor, the false-alarm probability is increased, placing a burden on subsequent data processing. Adequate clutter sup-pression is inevitably accompanied by decreased target detection probability in some fraction of the coverage area, caused by increased loss of the signal in the suppression process.
3.5.1 Effect of False Alarms
In a track-while-scan or multifunction radar system, the data processor assigns a tracking channel to attempt track initiation at the location of an alarm that is not correlated with an existing track file. This channel is occupied for several scans after the alarm. For reliable track initiation, the number of tracking channels nch in the data processing system must be increased beyond the number ntr of actual tar-gets present:
ntin = number of track initiation scans following a noise alarm;
nr = number of range cells within Rm;
naz = number of azimuth cells in the scan sector Am; Pfa = false-alarm probability for noise;
ntic = number of track initiation scans following a clutter alarm;
np = number of unsuppressed clutter points;
Ac = area of the resolution cell given by (3.18);
AmR2m/2 = surface area covered by scanning an azimuth sector to range Rm. The first term on the right-hand side of (3.67) is the number of target tracks ex-pected, the second is the average number of tracking channels consumed in at-tempts to validate and acquire track on noise alarms, and the third is the average number of channels consumed by similar attempts on discrete clutter. When track-ing at out to an instrumented range Rinst > Rm is desired on targets having large RCS, the range Rinst replaces Rm in (3.67).
3.5.2 Required Noise False-Alarm Probability
To avoid excessive burden on the data processor, Pfa is normally set so that the second term in (3.67) is a small fraction of the first. For example, if the maximum
number of targets expected is ntr = 100, Pfa is set to make the second term < 10 channels. Noise alarms are uncorrelated and widely scattered in the search sector, so the number of track initiation attempts following a noise alarm can be limited to ntin = 2. Thus, in our example, about five noise alarms are allowable per scan:
nrnazPfa 5. In a typical search radar scanning 360, nr = 2,000 and naz = 200, giv-ing nrnaz = 4 105 resolution cells in the search sector, and requiring Pfa 5/nrnaz
= 1.25 105. Use of higher resolution in search requires reduction in Pfa, increas-ing the required signal-to-noise ratio.
3.5.3 Requirements for Rejection of Discrete Clutter
The third term in (3.67) must be controlled by design of the signal processor.
Since clutter alarms are correlated from scan to scan, attempted track initiation or recognition of a fixed echo typically requires ntic = 5 scans after a clutter alarm.
Accordingly, np 2 clutter alarms per scan are permitted, if the track processor load for discrete clutter is to be less that 10 channels.
An important point in design of track-while-scan and multifunction radar sys-tems is that every detectable target within the surveillance coverage must either be assigned a track file or placed in a clutter map. Otherwise there will be repeated alarms requiring attempts at track initiation. Bird statistics are such that most de-tections result from upward fluctuations of echoes whose average is too low to provide reliable tracking or mapping. Only discrimination using differences be-tween target and bird RCS, altitude, or true velocity can avoid random false alarms from loading the data processor. Shrader [7, p. 2.87] describes a sensitivi-ty-velocity control (SVC) system that combines measurements of RCS and true (unambiguous) radial velocity to perform this discrimination, using PRF diversity over several CPIs. He suggests that it is most applicable to radars operating below 1.4 GHz. For radar frequencies in the UHF or lower bands, most birds lie in the Rayleigh region where RCS varies as f04, reducing the number of detectable ob-jects. Operation in these low bands also discriminates against low-altitude objects, but these may include targets as well as birds and land vehicles.
Data from Table 9.3 show that metropolitan areas have fixed clutter sources exceeding +40 dBsm with densities up to 0.2 per km2. The maximum range hav-ing such density is typically Rcmax = 20 km. The area within 20 km over a 360
scan is some 1,250 km2, and so 250 clutter points exceeding +40 dBsm may be expected. Virtually all of these must be prevented from sending alarms to the data processor.
Suppression of fixed discrete clutter cannot be provided by cell-averaging CFAR threshold control, since the cells containing the clutter are scattered over the search area. Suppression must be provided by some combination of velocity discrimination and clutter mapping in the signal processor, as described in Section 9.6.5. Table 9.3 is based on data gathered prior to deployment of wind turbines
Radar Equations for Clutter and Jamming 91 (Section 9.4.4), which combine large RCS, fixed locations, and large Doppler shifts. Clutter mapping appears to offer a solution to this type of fixed clutter.
3.5.4 Summary of Discrete Clutter Effects
The is no equation that describes the effect of discrete clutter on radar detection range. Discrete clutter controls the design of the signal processor, and the resulting processor losses are included in calculating the effective detectability factor Dx. High-resolution clutter maps offer a means of suppressing strong fixed clutter points at the expense of suppressing some fraction of target detections. To the extent discrete clutter passes the signal processor, the subsequent data processor must be designed to handle tracking loads greater than the number of targets actu-ally present in the surveillance area, to avoid saturation that would result in failure to respond to actual target detections.