Distributed Direct Position Determination of Moving Sources in Asynchronous Sensor Networks

Existing direct position determination (DPD) methods for moving radiation sources based on time difference of arrival (TDOA) and frequency difference of arrival (FDOA) typically assume synchronous sensor networks, rendering them unsuitable for asynchronous networks affected by clock and frequency biases. To address this limitation and extend applicability to large-scale sensor networks, this paper proposes a distributed DPD algorithm for moving sources in asynchronous sensor networks based on calibration sources. To achieve distributed global optimization of parameters including clock and frequency offsets between neighboring sensors as well as the position and velocity of the source, this study introduces a distributed multiple particle filtering (MPF) framework. Utilizing the coarse estimates provided by the distributed MPF, the algorithm then applies a diffusion-based gradient method to obtain the refined estimates. Throughout the entire process, the radiation source is regarded as an increasingly reliable calibration source. The algorithm jointly utilizes information from the radiation source and the calibration sources to estimate clock and frequency biases. Meanwhile, information exchange in the algorithm is restricted to neighboring sensors, eliminating the need for multi-hop transmission. Experimental results demonstrate consistent convergence of the estimated parameters across sensors and validate the superior performance of the proposed algorithm, even under challenging conditions of low signal-to-noise ratio (SNR) and limited observation duration.