Non-parametric multi-target tracking with Doppler measurements

This paper investigates the problem of incorporating Doppler measurements to improve multi-target tracking performance. The clutter density in both spatial domain and Doppler direction is assumed to be unknown and non-homogeneous. Instead of modeling the clutter density as the product of the clutter spatial density and the clutter Doppler probability density function (pdf), this paper introduces a hyper-spatial density concept, where the Doppler direction is considered as an additional pseudo position dimension. The spatial sparsity estimator is extended to the hyper-spatial density estimation and is combined with the Joint Integrated Probabilistic Data Association (JIPDA) tracker to deal with non-parametric multi-target tracking problem with Doppler measurements. Monte Carlo simulation results demonstrate the effectiveness of the proposed approach.