Fuzzy trust-based secure localization with CNN intrusion detection and RANSAC-GDA for wireless sensor networks

Wireless Sensor Networks (WSNs) have made possible many applications like environment tracking, healthcare, smart cities, etc. However, reliable node localization remains challenging due to energy constraints, noisy range estimates, and malicious adversaries. This paper proposes a hybrid secure and energy-efficient localization framework that integrates Distributed Energy-Efficient Clustering (DEEC), Random Sample Consensus (RANSAC), and a Gaussian Distribution Approach (GDA) to enable robust position estimation. In addition, a lightweight Intrusion Detection System (IDS) based on a SMOTE-assisted Convolutional Neural Network (CNN) is deployed at the pre-localization stage to identify and exclude malicious nodes, including Sybil behavior and falsified localization information. DEEC-RANSAC-GDA facilitates energy-efficient selection for cluster head formation, mitigates localization outliers via RANSAC, and optimizes localization using Gaussian probabilistic models. Implementation of the proposed IDS on the CIC-IDS-2017 benchmark dataset achieved 99.83% accuracy, 99.00% precision, 99.84% recall, and 99.42% F1-score, along with minimizing localization errors falling within the range of 0.46 m-0.53 m compared to DV-Hop and Weighted Hyperbolic Localization schemes.