Some problems of the application for multiple target tracking algorithm

Abstract

In this paper, some problems of the realization in the airborne multiple target tracking data processing are discussed. In order to reduce the computation load and the storage requirement in the airborne multiple target tracking (MTT) system, nearest neighbour algorithm that is easy realized should be applied to assign the observations to the tracks. When the decoupled Kalman filters in the line-of-sight (RHV) coordinate is adopted to maintain tracks, rectangular gate can be used to relate in each decoupled axis respectively. The statistical distance required in association must be calculated jointly in all the axes. The filter data obtained from data processing algorithm must be converted to the kinetic parameters of the targets in the earth coordinate, i.e. the speed and yaw of the targets when they are output to the command and attack systems. The computing formula of the speed and yaw of the targets and the relation between their estimate error and the filter errors are presented in this paper. The above-mentioned methods are realized in a digital signal processor (DSP) chip TMS320C30. In the laboratory environment, a real-time signal simulator is used as a substitute for radar, and the radar video signal of its output is provided for the experiments. It is shown by the results of the experiments that multiple targets can be tracked steadily by those methods. In the condition of signal-to-noise rate SNR=3:1, the range filter error is 1/3 of range cell and azimuth filter error is 1/70 of the beam width in the range of 80 km.