Mobility-aware and multipath-resilient node localization using MAL-CNN in UASNs

초록

Accurate localization in Underwater Acoustic Sensor Networks (UASNs) is challenged by sensor-node mobility, multipath propagation, Doppler effects, and environmental variability. This paper proposes a Mobility-Aware Localization Convolutional Neural Network (MAL-CNN) integrated with Adaptive Orthogonal Multipath Signal Design (AOMSD) for mobile sensor-node localization. AOMSD enhances acoustic arrival-time measurements through parameter-domain channel estimation, time-orthogonal beacon allocation, adaptive power control, multipath and Doppler correction, and guard-interval adjustment. The enhanced time-difference-of-arrival measurements are converted into range differences using environmentally adjusted sound speed. MAL-CNN employs two branches to extract spatial-ranging and temporal-environmental features from anchor measurements, position history, and environmental information. These features are used to predict a residual correction to a geometric position estimate, followed by Kalman-based refinement under an innovation consistency test. The proposed framework is evaluated using ns-3.38 and compared with MPL, MPOL, and HLCM across 30 paired simulation runs per experimental condition. Under the nominal configuration of 500 nodes and 100 anchors, MAL-CNN-AOMSD achieves a mean three-dimensional localization error of 0.928 m, an RMSE of 3.317 m, and a successful localization rate of 96.30% among valid estimates within a 2 m error threshold. Comparative robustness experiments under varying node mobility, multipath severity, Doppler stress, and environmental uncertainty show that MAL-CNN maintains the highest threshold-based localization success among the evaluated methods. MPOL achieves lower conventional RMSE under the tested robustness conditions, while the proposed framework requires greater communication energy than the comparison methods. These findings highlight a trade-off between threshold-based localization success, aggregate localization error, and communication energy. The Authors This is an open access article under the CC BY license. http://creativecommons.org/licenses/by/4.0/

키워드