Adaptive Sequential Gaussian Sum Approximation for High-Resolution Radar Target Tracking with Non-Gaussian Multi-Detection

This paper proposes a practical nonlinear target tracking filter using high-resolution radar measurements with non-Gaussian distribution. To cope with the numerical instability and computational burden of the existing nonlinear filters, an adaptive sequential Gaussian sum approximation is used in designing a tracking filter. As an efficient way to restrain the mode inflation of the a posterior probability density function, the modes with the same measurement update history are merged at every time instant. Accordingly, the proposed filtering scheme is suitable for real-time implementation while maintaining satisfactory target tracking performance. Through simulations of a typical re-entry ballisitic target tracking scenario, the superior performance and computation efficiency of the suggested filter are demonstrated.