Bridging Geometric Modeling and Data-Driven Methods for Movable Anchor Localization in Wireless Sensor Networks

Accurate localization remains a significant challenge in wireless sensor networks (WSNs), particularly in environments with limited infrastructure and high noise levels. This study introduces a novel method for enhancing node position estimation using received signal strength (RSS) measurements from a single movable anchor node. The proposed approach initiates with the anchor node at an initial position, estimating its distance to a target sensor node through a simplified log-normal shadowing model that utilizes a known path-loss exponent. This model effectively captures shadowing effects while omitting small-scale fading, resulting in robustness in open areas typical of WSN deployments, but with limitations in complex multipath environments. Subsequently, the anchor node relocates along a constrained axis to an optimized position, determined through geometric analysis and real-time RSS feedback, with the objective of minimizing localization error. In contrast to traditional methods that require multiple static anchors, this methodology exploits the mobility of a single anchor and geometric trilateration to achieve robust localization with minimal hardware. The results indicate that models relying solely on raw distance features exhibit limited predictive power, highlighting the potential of geometric constraint embedding to improve interpretability and accuracy. This research demonstrates the value of integrating domain-specific geometric modeling with data-driven techniques, providing practical guidance for the design of efficient, low-cost, and scalable localization systems in wireless sensor networks.