Attack detection and mitigation scheme through novel authentication model enabled optimized neural network in smart...

Area of Science:

• Computer Science
• Biomedical Engineering
• Cybersecurity

Background:

• The Internet of Things (IoT) is integral to daily life, enabling seamless data access and management.
• Data security in cloud storage is a critical concern, especially within healthcare applications.
• Protecting sensitive patient data transmitted via IoT networks is paramount for reliable healthcare services.

Purpose of the Study:

• To propose a secure authentication system for protecting sensitive healthcare data within IoT networks.
• To ensure the secure storage and access of Electrocardiography (ECG) signals for arrhythmia analysis.
• To enhance the accuracy of attack detection and mitigation in healthcare IoT systems.

Main Methods:

• Utilized modified Elliptic-curve Diffie-Hellman (ECDH) encryption for secure storage of ECG signals in the cloud.
• Employed a neural network (NN) for classifying attacks on healthcare data.
• Tuned NN weights using a novel hybrid tempest brain optimization algorithm integrating collaborative and hybrid search agents.

Main Results:

• Achieved a detection accuracy of 95% for identifying attacks.
• Successfully managed 7150 genuine users.
• Minimized information loss to 111 units.
• Demonstrated the proposed method's superiority in attack detection and mitigation.

Conclusions:

• The proposed secure authentication method effectively protects sensitive healthcare data in IoT environments.
• The integration of modified ECDH encryption and a hybrid-optimized neural network significantly improves security and attack detection.
• This approach offers a robust solution for secure data management and analysis in healthcare IoT applications.