An intelligent attention based deep convoluted learning (IADCL) model for smart healthcare security

Area of Science:

• Cybersecurity
• Internet of Things (IoT)
• Healthcare Technology

Background:

• Rapid expansion of IoT in smart infrastructures like healthcare systems.
• Existing security measures are inadequate for resource-constrained IoT devices against evolving cyber threats.

Purpose of the Study:

• To propose an enhanced security architecture, Intrusion Attack Detection and Classification (IADCL), for smart healthcare IoT systems.
• To develop a robust cyber threat detection framework using publicly available datasets.

Main Methods:

• Utilized CIC-IDS 2017, CIC-IDS 2018, CIC-Bell DNS 2021, and NSL-KDD datasets for framework development.
• Implemented an Intrusion Reduction and Knowledge Optimization (IRKO) method for feature selection.
• Employed an Adaptive Convolutional Bayesian Network (AConBN) classifier for traffic classification.
• Optimized classification using a Sine-Annealing-Harris Hawks Optimization (SA-HHO) algorithm.

Main Results:

• The IADCL framework demonstrated high accuracy in detecting cyberattacks.
• Performance evaluations confirmed the system's effectiveness across key metrics.
• Feature reduction by IRKO improved the efficiency of the detection process.

Conclusions:

• The proposed IADCL framework offers a robust solution for securing smart healthcare IoT devices.
• The system shows significant potential in mitigating cyber threats within connected healthcare environments.