Spatially sparse emitters localization with QVBEM algorithm

We study the estimation of the spatially sparse radio emitter locations from space, via the proposed Quad-tree variational Bayesian expectation maximization (QVBEM) algorithm. Firstly, we assume that the emitters are approximately lie on a uniform grid points in the region under surveillance. The VBEM algorithm is applied and the points exceeding the threshold level are considered as potential targets. Then, the grids are refined around the potential targets via the Quad-tree algorithm and the process is iterated. It allows us to find the location of sparse emitters with much less computational complexity due to the use of fewer grid points.Â我们针对天基空间稀疏无线电辐射源的定位估计问题，采用本文提出的四叉树变分贝叶斯期望最大化（Quad-tree variational Bayesian expectation maximization，QVBEM）算法展开研究。首先，我们假设监视区域内的辐射源近似分布于均匀网格点之上。随后应用变分贝叶斯期望最大化（VBEM）算法，将超出阈值水平的网格点视为潜在目标。接着通过四叉树算法对潜在目标周边的网格进行细化，并迭代执行上述流程。由于仅需使用更少的网格点，该方法能够以更低的计算复杂度实现稀疏辐射源的定位检测。