High-accuracy Multi-node Passive Localization Based on Group Sparse Total Least-squares

‍ ‍With the development of human economy and society， the noise of the ocean environment increases year by year， coupled with the development of noise reduction technology in various countries， the range and detection accuracy of a single node （underwater array） decreases year by year. On the other hand， a single node can usually only obtain target direction or distance information， and cannot accurately know the target position， which has become increasingly difficult to meet the practical applications. Therefore， the cooperative detection with multiple nodes is becoming a popular research topic for underwater passive positioning， where how to build the received signal model and design the high-precision localization algorithm for multi-nodes is difficult to be addressed. Based on the sparseness of the target in space， this paper establishes a multi-node passive positioning model based on group sparsity， and adopts group sparse optimization technology to achieve direct positioning. To solve the problem of missing localization accuracy caused by grid mismatch， this paper constructed an off-grid localization model by first-order Taylor expansion， and propose the group sparse total least-squares algorithm to achieve high-precision estimation of the target position. Simulation experiments show that the proposed model and algorithm are effective and robust， and can correctly solve the multi-target position in the case of four groups of nodes. Compared with the on-grid model， the off-grid model can further reduce the localization error by about 10 m. Compared with the traditional two-step methods and the classical direct localization methods， the proposed group sparse total-least square algorithm has higher localization accuracy when the signal-to-noise ratio is not less than -5 dB for single-source and multi-source cases.