Abstract :

Wireless Sensor Networks (WSNs) enabled with the Internet of Things (IoT) are susceptible to hostile node behavior that compromises routing integrity and localization reliability. This paper describes a hierarchical malicious-node-detection and secure localization model for WSNs using IoT technologies and blockchain. To achieve practical viability, sensor nodes undertake lightweight sensing and forwarding, while preprocessing, consensus, and learning are performed at the edge and cloud layers. The UNSW-NB15 and CIC-IDS-2017 benchmark datasets are used to assess detection performance and model malicious traffic behavior. Missing or corrupted records are reconstructed using a denoising autoencoder, and optimized feature selection is performed. An authorized Byzantine Fault Tolerant blockchain system ensures the integrity of records and eliminates malicious behavior when nodes communicate. A dual-domain attention-based convolutional neural network is used to identify malicious nodes, and the network memorizes both spatial and frequency-domain interaction anomalies. The localization accuracy is evaluated independently based on Euclidean distance errors in dense, sparse, and mobile WSN deployments. Simulation experiments under controlled conditions demonstrate that, with a network density of 0.5, the detection accuracy is approximately 99%, and the localization error is between 0.42 m and 0.74 m. Though Evaluation is performed using benchmark intrusion datasets and simulation-based localization, it shows that considering trust-aware blockchain logging alongside attention-driven detection can enhance resilience against malicious-node behavior without impacting scalability or energy efficiency in IoT-WSN settings.