An Efficient Normalized Rank Based SVM for Room Level Indoor WiFi Localization with Diverse Devices
Table 1
Percentage of accurately estimated locations based on linear SVM adopting multiple combinations of constraints violation along with comparison between RBF kernel and regularized Logistic Regression; the radio map is defined upon RSSI values.
Devices
Classification type
RBF kernel
L2-R L2-N
L2-R L2-N
L2-R L1-N
L2 LR
L1-R
L1
(primal)
(dual)
(dual)
(primal)
L2-N
LR
Oppo A31c
BBK
94.17%
95.20%
95.20%
95.42%
Coolpad 8730L
90.62%
90.42%
Gionee
99.17%
99.37%
99.37%
99.37%
HTC One E8
Huawei GRA-CL00
97.92%
98.12%
98.12%
98.33%
Lenovo A788t
94.58%
94.78%
94.78%
95.00%
Meizu
99.58%
99.58%
99.58%
99.58%
Oppo R7c
98.12%
98.12%
98.12%
97.92%
Xiaomi
Xiaomi Cancro
Samsung klteduoszn
92.92%
92.50%
92.50%
93.12%
Meizu M2 note
97.71%
97.08%
97.08%
98.33%
Samsung trlteduosctc
93.95%
93.75%
93.75%
93.96%
BBK Vivo
96.87%
97.08%
97.08%
97.29%
Standard device accuracy
91.87%
91.87%
91.87%
92.29%
Recall (ratio)
0.92
0.92
0.92
0.97
0.79
0.79
0.55
Precision (ratio)
0.95
0.95
0.95
0.97
0.86
0.83
0.77
() R stands for regularization; () N stands for norm; ( ) LR: stands for logistic regression. We put in bold the percentages lower than 90% for easiness of observation.