Multi-dimensional multiple hypothesis tracking with a Gaussian mixture model to suppress grating lobes

A new approach for multi-target tracking under grating lobe environment is presented. Ambiguous angle measurements are obtained by a wide-array antenna whose elements have a long interval. Angle measurements different from the true target angles are obtained because of the grating lobes; therefore, it is difficult to initialize and maintain target tracking accurately. In this paper, we propose a multi-dimensional multiple hypothesis tracking method using a Gaussian mixture model, which is an extension of multiple hypothesis tracking for multiple targets. The proposed method assumes distribution of angle measurement error including grating lobes as a Gaussian mixture model. From this assumption, combinations of measurements, tracks, and Gaussian components are generated as a three-dimensional (3-D) hypothesis. The algorithm finds multiple measurement-to-track-to-grating hypotheses from a 3-D association matrix efficiently. Simulations are performed to demonstrate the effectiveness of the proposed algorithms.