IntruSafe: a FCNN-LSTM hybrid IoMT intrusion detection system for both string and 2D-spatial data using sandwich architecture

Abstract The Internet of Medical Things (IoMT) is a resource-constrained device with limited computational capabilities. However, the market worth of this section is booming rapidly. The IoMT manufacturers need to offer their products at a competitive price, which forces them to use simplified architecture, leaving limited and, to some extent, no scope to employ sophisticated cybersecurity algorithms. As a result, IoMT has become a lucrative practice ground for cybercriminals. The IoMT sector deals with valuable, confidential healthcare-related data and offers convenient, personalized healthcare services. That is why the market demand and IoMT intrusion are experiencing massive growth. An innovative Intrusion Detection System (IDS), IntruSafe, has been studied, developed, and presented in this paper that combines Fully Connected Convolutional Neural Network (FCNN) and Long Short-Term Memory (LSTM) to protect the IoMT network from malicious signals. The IntruSafe combines FCNN and LSTM to ensure the detection of both malicious text and image data. It detects and simultaneously protects the IoMT network from further intrusion with only a 0.18% service interruption rate. This high-performing IDS detects intrusion with 97.66% accuracy, 98.50% precision, 97.33% recall, and 97.85% F1-score. With outstanding performance, IntruSafe is a promising IDS that will facilitate further growth of the IoMT sector while minimizing the risks of a successful intrusion.