B-CAT: a model for detecting botnet attacks using deep attack behavior analysis on network traffic flows

Journal of Big Data

Table 7 Comparison with other studies

Detection model	Dataset used	Botnet detection result (%)			Average result (%)			Botnet characteristic detection
Detection model	Dataset used	Accuracy	Precision	Recall	Accuracy	Precision	Recall	Botnet characteristic detection
Khan et al. [23]	CTU-13	98.70	–	–	98.70	–	–	No
Joshi et al. [7]	CTU-13	99.94	–	–	99.94	–	–	No
Dollah et al. [42]
Decision tree	CTU-13	92.20	99.93	84.47	97.27	99.83	94.59	No
	NCC-1	99.63	99.85	99.41
	NCC-2	99.98	99.70	99.88
k-NN	CTU-13	75.16	73.18	51.52	90.56	90.21	81.54
	NCC-1	96.67	99.15	94.18
	NCC-2	99.85	98.31	98.91
Naïve Bayes	CTU-13	69.34	62.28	99.45	74.69	54.08	92.48
	NCC-1	65.38	66.77	82.82
	NCC-2	89.36	33.20	95.17
Random Forest	CTU-13	73.83	49.99	47.67	74.99	66.47	49.99
	NCC-1	51.16	49.55	2.34
	NCC-2	99.99	99.86	99.96
Hostiadi et al. [16]	CTU-13	99.18	42.29	91.55	86.33	39.26	96.19	No
	NCC-1	99.73	75.14	99.29
	NCC-2	60.09	0.36	97.73
Proposed Method	CTU-13	99.97	99.89	97.38	99.82	96.79	94.64	Yes
	NCC-1	99.60	90.49	89.16
	NCC-2	99.87	100	97.38