Development of Adversaries Detection Framework using Ensemble of Residual Autoencoders with Attention Mechanism in Wireless Body Area Networks

Wireless Body Area Networks (WBANs) have a high potential for transforming the healthcare sector by combining implantable and wearable healthcare systems for continuous monitoring. Nevertheless, the revolutionization poses some important security limits such as vulnerabilities in the environment of WBAN, data privacy and authentication problems. Conventional security methods encounter problems in resolving these concerns efficiently. The tiny sensor systems garner the health attributes and send the data to a remotely placed medical support. These types of resource-constrained networks are susceptible to routing threats that affect the overall network. The cryptographic models are ineffective in fighting over the routing threats, while the machine learning-assisted intrusion identification approach can recognize uncommon and typical security attacks. Hence, in this work, an advanced ensemble deep learning model is developed for detecting adversarial attacks in WBAN. The data from the WBAN is collected from the online sources. The gathered data is directly fed to the Ensemble of Residual Autoencoders with Attention Mechanism (ERA-AM) for the detection of adversaries present in the network. Hence, the security of the WBAN system is highly protected through this developed ensemble deep learning-based adversary detection mechanism. Finally, the effectiveness of the system is investigated by contrasting it with other related approaches.