Research on Fingerprint Location Algorithm Based on OCAE-SOM

Xinchun Li; Xiaolu Ji; Wu Wei; Liyan Wang; Yongyan Gu; Dayan Cao

doi:10.3788/LOP202158.0812001

Journals >Laser & Optoelectronics Progress >Volume 58 >Issue 8 >Page 0812001 > Article

Laser & Optoelectronics Progress
Vol. 58, Issue 8, 0812001 (2021)

Research on Fingerprint Location Algorithm Based on OCAE-SOM

Xinchun Li¹, Xiaolu Ji^2、*, Wu Wei², Liyan Wang², Yongyan Gu², and Dayan Cao¹

Author Affiliations

¹School of Electronic and Information Engineering, Liaoning Technical University, Huludao, Liaoning 125105, China

²Graduate School, Liaoning Technical University, Huludao, Liaoning 125105, China

show less

DOI: 10.3788/LOP202158.0812001 Cite this Article Set citation alerts

Xinchun Li, Xiaolu Ji, Wu Wei, Liyan Wang, Yongyan Gu, Dayan Cao. Research on Fingerprint Location Algorithm Based on OCAE-SOM[J]. Laser & Optoelectronics Progress, 2021, 58(8): 0812001 Copy Citation Text

show less

Fig. 1. Basic model of convolutional autoencoders

Download full size

Fig. 2. Flow chart of overall algorithm

Download full size

Fig. 3. Different environment layouts. (a) LOS layout; (b) NLOS layout

Download full size

Fig. 4. Loss comparison curves of different sizes

Download full size

Fig. 5. Contrastive curves of accuracy in different modes

Download full size

Fig. 6. Comparison of positioning accuracy of four indoor positioning algorithms in different environments. (a) NLOS layout; (b) LOS layout

Download full size

Layer	Parameter	Activation function	Output size
Convolutional layer	3×3 filter kernel16 feature images	ReLU	64×64×16
Max pooling layer	2×2	--	32×32×16
Convolutional layer	3×3 filter kernel8 feature images	ReLU	32×32×8
Max pooling layer	2×2	--	16×16×8
Convolutional layer	3×3 filter kernel8 feature images	ReLU	16×16×8
Max pooling layer	2×2	--	8×8×8

Table 1. Network structure of CAE

Parameter	Meaning	Value
α	Learning rate	0.01
β₁	Exponential decay rate of 1^stmoment estimate	0.9
β₂	Exponential decay rate of 2^ndmoment estimate	0.999
ε	Constant	1×10^-8

Table 2. Parameter setting of Adam algorithm

Method	Environment	Maximum positioning error /m	Average positioning error /m	Training time /s
CAE	LOS	4.5382	1.2725	3.41
CAE	NLOS	7.9603	1.9341	6.32
PCA	LOS	6.0362	1.7260	7.29
PCA	NLOS	9.5374	2.3071	12.53
Statistical average	LOS	6.5607	1.9564	9.72
Statistical average	NLOS	10.6243	3.0472	16.74

Table 3. Comparison of location accuracy of feature extraction methods

Algorithm	Probability distributions /%		Training time /s
Algorithm	Accumulate within 1 m	Accumulate within 2 m	Training time /s
CAE-SOM	62.71	85.36	10.42
OCAE-SOM	70.54	96.11	1.86

Table 4. Performance comparison of proposed algorithm before and after improvement

Algorithm	CSI-improved WKNN	smoothing+SVM	CSI-XGBoost	OCAE-SOM
Average location response time /ms	10.249	8.935	8.027	6.914

Table 5. Average location response time of different algorithms

Xinchun Li, Xiaolu Ji, Wu Wei, Liyan Wang, Yongyan Gu, Dayan Cao. Research on Fingerprint Location Algorithm Based on OCAE-SOM[J]. Laser & Optoelectronics Progress, 2021, 58(8): 0812001

Download Citation

Set citation alerts for the article

Tools

Set citation alerts for the article

Save the article for my favorites

Paper Information