ISARLAB - Inverse Synthetic Aperture Radar Simulation - Defence ...

DSTO-GD-02102. Since we are modelling in the high frequency (Fresnel) region, the target isrepresented as a finite number of discrete scatterers in 3-D space.3. We assume the scatterer positions are more important for classification thantheir respective amplitudes. This does not mean that we fail to model relativeamplitude differences between scatterers but that the exact level of thebackscatter amplitude is not critical to our modelling.4. Target classification will be made on the basis of major target scatterers hencemodelling minor target features is lower priority.We supplement these assumptions with the following conventions:1. The positions of the target scatterers are defined with respect to a localcoordinate system xyz. The origin of this local system is referred to as theposition of the target or the target’s origin in the global coordinate system XYZ.Ideally the target origin should correspond to the effective center of rotation ofthe target with respect to the radar. If this is not the case, then the Dopplerreturns from the target will be shifted. (This is not a critical issue because inreal target imaging the apparent centre of rotation will normally shift duringthe course of relative motion).2. The scatterer positions should be defined so that the target is “forward” alongthe x-axis and that the target is “level” in the x-y plane. This information isused in realigning the orientation of the target as its velocity changes directionand to define the state of equilibrium about which the induced motion applies.See Section 3.1 for more information about target alignment.3.2.1 The Target Scatterer ModelThe above assumptions were implemented by modelling the target scatterers as a setof points such that for each scatterer we have defined:• x-coordinate position (m);• y-coordinate position (m);• z-oordinate position (m); and• reflectivity or radar cross-section (dBsm).Since target scattering is heavily dependent on the aspect angle at which the radarilluminates the target, it was necessary to allow a separate point scatterers’ model foreach angle in a range of aspect angles. Consequently, the whole model is composed ofa 2-D array of points with one dimension being the set of points (position andreflectivity) defining the target scattering at a particular aspect angle and the otherdimension being all the valid aspect angles for which there are corresponding models.Aspect angle is actually composed of two components, azimuth and elevation, but tosimplify the modelling we consider aspect angle to be synonymous with azimuthangle. It is therefore assumed that one would need a different target model for eachelevation angle to be simulated. Since scattering does not change significantly with the12