Robust Vehicle Localization Exploiting Two Based Stations Cooperation: A MIMO Radar Perspective

Autonomous vehicles depend on global positioning systems' aided by motion sensors to estimate its position within the traffic network. However, all the driving vehicles cannot be ensured to have satellite visibility. Therefore, in order to increase the accuracy and robustness of vehicle localization, it is important to have an assistant localization method for these systems by using some environmental sensing, such as cameras and radars. In this paper, we look at using multiple-input multiple-output (MIMO) radar for vehicle localization. The performance of cross localization in MIMO radar relies on the accuracy of the direction of arrival (DOA) estimation. But the performance of most existing DOA estimation methods based on sparse signal recover is affected by the unknown nonuniform noise and mutual coupling. The proposed method uses a linear transformation to eliminate the influence of mutual coupling by utilizing the banded complex symmetric Toeplitz structure of the mutual coupling matrices in both transmit and receive arrays. Then, a real-valued covariance vector-based sparse Bayesian learning framework is formulated for DOA estimation by utilizing the unitary transformation, in which the variances of nonuniform noise can be updated by using the least square strategy. The proposed method can work well and provide better DOA estimation performance than the existing sparse signal recover-based algorithms in unknown mutual coupling and nonuniform noise. Simulation results are provided to demonstrate the advantage of the proposed method.