A Survey of User Authentication Based on Channel State Information

Table 2

Activity-based user authentication applications.

System	Users	Signal/preprocessing	Experimental scene	Purpose/classification	Performance
WiAU [53]	Office: 12; Corridor: 14 (two users are twins)	Amplitude/Butterworth low-pass filter	An office and three corridors of different floors	Identity recognition stranger identification/CNN, ResNet	Legal user (12 or 14 people): 98%; Illegal user: 92%
WiID [67]	10	Amplitude and phase/low-pass filter, PCA, and DWT	Two laboratories	Identity recognition/SVM and HMM	10 people: 90%
Shi et al. [63]	Office: 11; Apartment: 5	Amplitude and phase/band-pass filter	An office (); An apartment ()	Identity recognition stranger identification/DNN, SVM	Walking activities: 94%; Stationary activities: 91%; Illegal user: 89.7%
Regani et al. [54]	5	Phase/PCA	In-car scenarios	Identity recognition stranger identification/KNN, linear SVM, SVM-RBF, and NN	Single driver: 90.66%; Two drivers: 99.36%