Target Tracking - MATLAB Simulations for Radar Systems Design - page 416

that the detection is assigned to the proper track. A simplified TWS data pro-
cessing block diagram is shown in Fig. 9.17.

Choosing a suitable tracking coordinate system is the first problem a TWS
radar has to confront. It is desirable that a fixed reference of an inertial coordi-
nate system be adopted. The radar measurements consist of target range, veloc-
ity, azimuth angle, and elevation angle. The TWS system places a gate around
the target position and attempts to track the signal within this gate. The gate
dimensions are normally azimuth, elevation, and range. Because of the uncer-
tainty associated with the exact target position during the initial detections, a
gate has to be large enough so that targets do not move appreciably from scan
to scan; more precisely, targets must stay within the gate boundary during suc-
cessive scans. After the target has been observed for several scans the size of
the gate is reduced considerably.

establish time
and radar
coordinates

smoothing
& prediction

deleting files
of lost targets

pre-processing
gating

correlation /
association

radar
measurements

establish
track files

Figure. 9.17. Simplified block diagram of TWS data processing.

Gating is used to decide whether an observation is assigned to an existing
track file, or to a new track file (new detection). Gating algorithms are nor-
mally based on computing a statistical error distance between a measured and
an estimated radar observation. For each track file, an upper bound for this
error distance is normally set. If the computed difference for a certain radar
observation is less than the maximum error distance of a given track file, then
the observation is assigned to that track.

All observations that have an error distance less than the maximum distance
of a given track are said to correlate with that track. For each observation that
does not correlate with any existing tracks, a new track file is established
accordingly. Since new detections (measurements) are compared to all existing
track files, a track file may then correlate with no observations or with one or
more observations. The correlation between observations and all existing track
files is identified using a correlation matrix. Rows of the correlation matrix