Obfuscated Tor Traffic Identification Based on Sliding Window

Table 6

Result of the experiment for the parameters of the sliding window. Each item in the table is expressed in the form of precision rate/feature extraction time/model prediction time. We comprehensively consider three indicators (the importance of precision rate, feature extraction time, and model prediction time decrease in order) of each pair .

	n = 50	n = 100	n = 200	n = 500
m = 5	0.9984/1112/12	0.9991/1348/12	0.9990/1996/9.2	0.9991/3691/8.5
m = 10	0.9989/560/4.9	0.9993/698/5.0	0.9992/994/4.1	0.9993/1843/4.4
m = 15	0.9987/370/4.0	0.9991/470/2.9	0.9992/473.2.9	0.9993/1257/2.7
m = 20	0.9988/281/2.3	0.9990/350/2.0	0.9991/504/1.9	0.9992/939/2.0
m = 50	0.9984/113/0.8	0.9991/143/0.8	0.9990/206/0.7	0.9991/386/0.8