Sparse Bayesian Inference-Based Direct Off-Grid Position Determination in Multipath Environments
Hao Ke-gang, Qun Wan
Abstract
The classical super-resolution Direct Position Determination (SR-DPD) algorithms fail to suppress the coherent Non-Line-of-Sight (NLOS) interference due to the lack of independent measurements. The existing Sparse Signal Reconstruction (SSR) based DPD approach suffers from the intractable complexity since it needs to solve a Second-Order Cone Programming (SOCP) problem. Besides, the Grid Quantization Error (GQE) exists in all above on-grid model based algorithms inherently. The proposed Sparse Bayesian Inference (SBI) based off-grid DPD algorithm is easy to implement as the Expectation-Maximization (EM) method is applied to decouple the multi-dimensional optimization problem. In addition, the GQE is also eliminated by introducing the Gradient Descent (GD) mechanism into the EM steps to update the grid point coordinates of interest iteratively.