Multiple target tracking using parallel processing

Tracking of a single target, in the ideal situation with one noisy measurement obtained at each radar scan can be achieved using standard Kalman filtering techniques. In the multi-target case, an unknown number of measurements are received at each radar scan and each measurement has to be associated with an existing or new target. In addition, false measurements may exist and some target measurements may be missed. When targets are very close to each other, however, more than one measurement may fall in a track gate or acceptance region. Several approaches exist to deal with this case, one possibility being for the tracking filter to accept only the `best´ measurement, another, which is called a track splitting algorithm, is for all the measurements which fall in the acceptance region to be used to generate additional tracking filters. That is if a filter accepts n measurements then it branches into n filters. This later approach appears suited to parallel processing. The author reports on an investigation using transputers programmed in OCCAM