A netting approach to automatic radar track initiation, association, and tracking in air surveillance systems

One of several promising concepts for netting radars in a tactical air surveillance system employs track-while-scan radars having scan periods of from 4 to 12 seconds. Such scan periods are too long to yield an acceptably high probability of automatic association of measurements with tracks when the surveillance region contains many highly maneuverable targets. As presented, the radars overcome this limitation by being connected into a non-hierarchical net over which measurement data is pooled. In the system concept described, most targets will be seen by twenty or more radars. If all radars able to see a target shared the resulting data with the entire system, the communications bandwidth required would be excessive. To avoid this problem, an algorithm was devised that dynamically selects a "best" subset of the system's radars to track each target, thus, simultaneously achieving a high probability of correct association (and hence of maintaining track) while requiring reasonable communications bandwidths. This performance is achieved without centralized control. The concept was verified using a detailed computer simulation called TACRAN (Tactical Air Control Radar Net); some simulation results are presented.