PARAFAC-based Robust Localization for Bistatic MIMO Radar with partially available measurements

This article tackles the problem of parameter estimation of bistatic multiple-input multiple-output radar in the presence of impulsive noise or outlier. It is assumed that only part of data entries are recorded, and a new approach based on parallel factor decomposition (PARAFAC) is proposed for the joint estimation of directions-of-departures and directions-of-arrivals. Since most of the state-of-the-art algorithms fail to give reasonable results under impulsive noise, we utilize the $l_{p}$-norm with $0p\leq 1$ to against the outliers. By employing the data model of low-rank higher-order tensor with partly observed entries and outliers, an alternating iterative approach based on augmented Lagrangian multiplier is designed to recover the signal subspaces, and then a standard ESRPIT scheme is suggested for the final target localization. Experimental results show that the proposed algorithm outperforms the state-of-the-art algorithms in terms of estimation accuracy under Gaussian mixture noise.