Efficient Implementation for SBL-Based Coherent Distributed mmWave Radar Imaging

In a distributed frequency-modulated continuous waveform (FMCW) radar system, the echo data collected are not in azimuth direction, so imaging effect of traditional range-Doppler (RD) algorithm is poor. Sparse Bayesian learning (SBL) an optimization based on theory that has been successfully applied to high-resolution because its strong robustness and high accuracy. However, SBL highly computationally complex. Fortunately, with FMCW data, most time-consuming operations involve Toeplitz-block Toeplitz (TBT) matrix. this article, advantage, we propose fast can be used obtain high-angular-resolution images, which inverse TBT matrix transposed as sum products block lower triangular circulant by using new decomposition method, some multiplications quickly computed Fourier transform (FFT), decreasing computation time several orders magnitude. Finally, simulations experiments were ensure effectiveness proposed algorithm.