A new Gaussian mixture method with exactly exploiting the negative information for GMTI radar tracking in a low-observable environment

This paper investigates the problem of ground vehicle tracking with a Ground Moving Target Indicator (GMTI) radar. In practice, the movement of ground vehicles may involve several different maneuvering types (acceleration, deceleration, standstill, etc.). Consequently, the GMTI radar may lose measurements when the radial velocity of the ground vehicle is below a threshold when it stops, i.e. falling into the Doppler blind region. Besides, there will be false alarms in low-observable environments where there exist high noises interferences. In this paper, we develop a novel algorithm for the GMTI tracking in a low-observable environment with false alarms while exactly incorporating the 'negative information' (i.e., the target is likely to stop when no measurements are recorded) based on the Bayesian inference framework. For the Bayesian inference implementation, the Gaussian mixture approximation method is adopted to approximate related distributions, while different filtering algorithms (including both extended Kalman filter and its generalization for interval-censored measurements) are applied for updating the Gaussian mixture components. Target state estimation can be directly obtained through the Gaussian mixture model for the GMTI tracking at every time instance. We have compared the developed method with other state-of-the-art ones and the simulation results show that the proposed method substantially outperforms the existing methods for the GMTI tracking problem.