Simulating Hybrid Deep Learning Mechanism For Security Enhancement During Blockchain Based Wireless Sensor Network (WSN) Authentication
Tejbir Singh, Rohit Vaid, Sanjeev Kumar Rana · 2024
In WSN, the processing power, memory, and battery life of sensor nodes are limited. Due to their widespread usage in untrusted situations, WSNs are vulnerable to a variety of threats. The trustworthiness of the data it gathers is also called into doubt due to the challenges in getting a WSN. The authentication process of WSNs allows for verification of the authenticity of data and resources. Authentication examines the data's provenance and permits only approved modifications, protecting data in WSNs from manipulation. On the other hand, ID spoofing attacks and other vulnerabilities exist in existing authentication techniques. Blockchain is just one more exciting new development in the realm of cyber defense. A new blockchain-based authentication method for WSNs was developed in this study. Sensor nodes and the blockchain were integral parts of the study's system design, with users and a private blockchain playing vital roles. A comprehensive security review was conducted on the study's data. The classification of safe and insecure records via blockchain-based WSN has been accomplished using a deep learning model. Following optimization-based filtering, data stored on blockchain is categorized. After optimization, the performance of the attack classification system much improves; general accuracy increases to $\mathbf{9 6 . 9 5 \%}$, precision, recall, and F1-score rises to about 97%. Successful optimization is shown by these changes.