Enhanced Direct Position Determination of Multiple Sources Based on One-Bit Sparse Bayesian Learning

The conventional direct position determination (DPD) algorithms impose significant communication overhead, since the raw signals are required to be transmitted from all sensor stations to a central processor. To address this challenge, we propose a communication-efficient approach for multi-source localization, named as OBSBL-DPD, which integrates one-bit quantized measurements with Sparse Bayesian Learning (SBL). The utilization of one-bit quantization reduces the communication load between the central processor and distributed stations, while SBL improves the localization accuracy by exploiting signal sparsity. In contrast to existing l₁-penalty-based DPD approaches, the proposed framework employs SBL to solve the localization problems with partially unknown dictionaries in a sparsity-driven way. Moreover, the traditional single-snapshot model is extended to support multiple-snapshot scenarios, which can further enhance robustness in practical applications. Additionally, we introduce an alternating minimization strategy to improve the convergence of the proposed algorithm. Simulation results demonstrate that the proposed OBSBL-DPD method exhibits superior localization accuracy and reduced computational complexity compared to conventional DPD approaches.