This work deals with the problem of fast direction-of-arrival (DOA) estimation. A low complexity iterative off-grid method is proposed, which employs the approximate message passing with unitary transformation based sparse Bayesian learning (SBL) to obtain initial estimates of the signals and their corresponding DOAs, and then refines the estimates iteratively using the Jacobi or Gauss-Seidel iteration with low complexity. Both general array and uniform linear array (ULA) are considered. Simulation results demonstrate that, with much lower complexity, the proposed method outperforms state-of-the-art methods, and its performance can approach the Cramer-Rao bound closely.

这项工作解决了快速到达方向（DOA）估计的问题。提出了一种低复杂度的迭代离网方法，该方法采用近似消息传递并结合基于unit变换的稀疏贝叶斯学习（SBL）获得信号及其对应DOA的初始估计，然后使用Jacobi或Gauss迭代地优化估计。 -Seidel迭代，复杂度低。通用阵列和统一线性阵列（ULA）都被考虑了。仿真结果表明，该方法具有较低的复杂度，其性能优于最新方法，并且其性能可以接近Cramer-Rao界。