Enhancing WSN Performance with a Hybrid Key Management Protocol Using Multitask Deep Learning in IoT Cloud Environments

WSNs help the Internet of Things (IoT) progress by gathering data in areas with limited resources, such as the civilian and military sectors. Authentication and key management at the edge devices are crucial for the security of real-time processing in the IoT. Unfortunately, conventional cryptography techniques are ineffective, leaving WSNs open to eavesdropping. Key sizes that are currently used for key management are quite large, which increases computing and communication costs, and is necessary because sensor nodes have limited resources. In order to lessen the burden on resources while increasing security, this study suggests a Hybrid Key Management Protocol. To reduce features, the system uses KMC-IG, a preprocessing method that combines Standard Scaler and Label Encoding. For limited datasets, the prediction accuracy can be enhanced by utilizing the Boosted Tree Model with DL architecture, which merges GBDT with Minimum MDL. After receiving BTMDL data from MDL, they construct BTMDL models and augment them with features. By outperforming existing methods, the suggested solution enhances WSN performance, reaching an accuracy of 95.75% and an F1-score of 95.87%. An efficient and secure WSN security solution that is computationally cheap is proposed in this work.