All elements of the model cooperate properly but the particular function needs more refinement and experimentation. Achievement of good results in ES A Multitarget Tracking is not a simple process. In particular, the problem of crossing trajectory correlation in 2D needs more attention. Additional information about target elevation is very useful even then accuracy of this measurement is not the best. In the 3D case, results are better but of course, it demands more computational burden. The observed too high percent of lost tracks needs a new solution adequate to work in a high-density environment including not only the real targets but also false alarms as a result of countermeasures. The next problem is dedicated to group targets. They haven't been taken into consideration in present level of research. When we compare the frame time with minimal tune necessary for search we can easily note that the time for additional looks (as a answer for tracking unit request) is limited to 1.5 second. It offers possibilities of repeating observations of a particular azimuth only 25 times. When only a small number of targets are in a search volume, the algorithms have to become more complicated. We expect the requirements for radar to reach up to 100 tracks. It is not possible to make all additional observations according to tracking unit expectations. So as a result the algorithm has to take into consideration the present density of targets. In the extreme case the performance of tracking will be on similar level to the MSA MTT ones. The analysed problem is similar to a multi criterion optimisation. Solving one problem generates problems in another process. To achieve final results, which will be good enough in different situations, we need more experiments with a different algorithms. Prepared software is a tool for such experiments.
展开▼