Abstract: Existing localization algorithms for underwater acoustic sensor networks need the presence of beacon nodes and assume that measurement noises follow Gaussian distributions, resulting in high cost and low accuracy. To address these problems, a self-localization algorithm based on maximum a posteriori is proposed for drifting-restricted underwater acoustic sensor networks under mixed measurement noises. We analyze nodes' mobility patterns to obtain the prior knowledge for localization, and characterize distance measurements under the assumption of additive and multiplicative noises as the likelihood information for localization. Under the Bayesian framework, the priori and likelihood information are fused to derive localization objective function by maximum a posteriori probability. Then Broyden, Fletcher, Goldfarb and Shanno quasi-Newton method is resorted to solve the objective function. The simulation results show that compared with similar localization methods, the proposed method does not need the presence of beacon nodes, and it has the advantages of high localization accuracy, fast convergence speed, and being robust to changes in measurement noises.

Key words: underwater acoustic sensor networks, beacon-free localization, maximum posterior probability estimination, quasi-Newton optimization