Detection of Middlebox-Based Attacks in Healthcare Internet of Things Using Multiple Machine Learning Models

Table 1

Comparative analysis.

Reference	Dataset	IoMT	Technique	Internal attacks	External attacks	Packets flow anomaly	Outcomes	Accuracy (%)	Limitations
Fujita et al. [40]	Real time data	✔	Machine learning	No	✔	✔	Anomaly detection and attacks protection	89	No detection using features
Manimurugan et al. [15]	Sensors data	✔	Machine learning	No	✔	✔	Early attack detection	88.67	No feature scoring
Saheed and Arowolo [6]	Cloud based data	✔	Machine learning	No	✔	✔	Anomaly detection	90	No real-time system
Manimurugan et al. [15]	Sensors data	✔	Machine learning	No	✔	✔	Anomaly detection	91	No detection using features
Aljumaie et al. [41]	Real time data	✔	Machine learning	No	✔	No	Anomaly detection	92	No feature scoring
Meng et al. [3]	Real time data	✔	Deep learning	No	✔	✔	Anomaly detection	91	No real-time system
Ali and Mahmoud [42]	Real time data	✔	Effective NN	No	✔	✔	Anomaly from real-time	89.5	No detection using features
Salem et al. [43]	Sensors data	✔	Efficient NN	No	✔	No	Anomaly from sensors data	92.06	No feature scoring
Sehatbakhsh et al. [16]	Sensors data	✔	Deep learning	No	✔	No	Jamming attacks in WBANS	90	No real-time system