MidSiot: A Multistage Intrusion Detection System for Internet of Things

Table 11

Comparing the proposed system with other one-stage systems.

Dataset	Work	Attack detection model	Binary attack detection	Multiclass attack detection	Prediction time
CICIDS-2017	MidSiot	Multistage	99.99	99.97	1.41
	Gamage and Samarabandu [35]	Random forest	N/A	99.86	60.48
	Vinayakumar et al. [36]	Deep neural network	93.10	95.60	N/A
	Elmrabit et al. [37]	Decision tree	99.90	99.90	N/A
	Manimurugan	Deep belief network	99.37	97.73	N/A
IOTID20	MidSiot	Multistage	99.98	99.88	1.01
	Ullah and Mahmoud [21]	Decision tree	99.94	99.69	N/A
	Alkahtani and Aldhyani [38]	Long short-term memory	98.20	N/A	N/A
	Song et al. [39]	Autoencoder	93.76	95.20	N/A
	Hussein et al. [40]	Random forest	99.90	99.90	N/A
	Ullah and Mahmoud [41]	Convolution neural network	99.98	97.76	N/A
	Islam et al. [42]	Decision tree	N/A	100.00	1139.37
BOT-IOT	MidSiot	Multistage	99.99	99.99	0.39
	Ferrag et al. [43]	Rules and decision tree	N/A	97.00	1.54
	Ferrag et al. [44]	Deep autoencoder	N/A	98.39	1916.55
	Dwibedi et al. [45]	Support vector machine	99.99	N/A	N/A
	Pokhrel et al. [46]	K-Nearest neighbor	92.10	N/A	N/A
	Ge et al. [47]	Support vector machine	99.74	99.03	693 040
	Ullah and Mahmoud [41]	Convolution neural network	99.90	99.97	N/A