Topology-Aware Anchor Selection Using AI for Range-Free Localization in Wireless Sensor Networks

This paper presents an AI-assisted distributed decision-making approach for improving range-free localization in wireless sensor networks (WSNs) by extending the classical DV-Hop method with intelligent anchor selection. The proposed framework introduces a dynamic anchor elimination mechanism guided by machine learning classifiers, which enables each unknown node to select the most informative anchors in a distributed manner. A tailored feature extraction scheme captures spatial relationships, network connectivity metrics, and distance estimation errors, allowing data-driven and context-aware decision making across the network. Extensive experiments on 40 WSN scenarios, including both uniform and non-uniform anchor deployments, validate the effectiveness of the proposed approach. Results show that removing a single poorly informative anchor identified through AI-based distributed processing can significantly enhance localization accuracy, with the best-performing configuration (LM+LGBM under uniform anchor deployments) achieving nearly a $$30%$$ reduction in RMSE compared to conventional full-anchor configurations. The analysis further highlights the influence of anchor topology, revealing performance degradation when anchors are concentrated in limited regions of the deployment area. Overall, this work demonstrates that AI-driven, topology-aware distributed decision-making strategies provide an effective alternative to anchor-intensive localization schemes, challenging the assumption that a larger number of anchors necessarily leads to improved localization performance in WSNs.