Floor positioning method indoors with smartphone’s barometer

Figures & data

Table 1. Error Distribution Data.

	Error distribution
Cell phone types	≤0.5 m	≤1 m	≤1.375 m	≤2 m
Huawei Mate8	63.3%	69.2%	84.2%	96.7%
Samsung Note2	37.5%	59.2%	74.2%	89.2%

Figure 1. The cumulative error distribution of relative height estimates by Huawei Mate8 and Samsung Note2.

Figure 2. Pressure changes over time in four states.

Table 2. Average pressure differences of four states at the intervals of 2 s (Air pressure units: hpa).

Average pressure differences	Activity
	Static	Flat walking	Upstairs	Downstairs
t= 2	0.01	0.00	0.12	0.13
t= 4	0.01	0.01	0.11	0.14
t= 6	0.00	0.02	0.12	0.12
t= 8	0.01	0.02	0.08	0.12
t= 10	0.01	0.02	0.13	0.11

Figure 3. Flow chart of going-upstairs and going-downstairs behavior detection.

Figure 4. Monitoring area schematic diagram.

Figure 5. Monitoring area flow-process diagram.

Figure 6. Different movement route schematic diagram.

Figure 7. The flow chart of “entering” detection algorithm.

Figure 8. Barometric difference between Huawei Mate8 (right) and Samsung (left).

Figure 9. Experimental buildings.

Figure 10. Precision of going upstairs and going-downstairs detection.

Figure 11. Precision of “entering” detection.

Table 3. Correct number of the floor identification.

Location	Cell phone type
	Samsung Note2	Xiaomi 5S	Huawei Mate8
Xiansu Hall	14	17	18
Library	16	16	17
Dormitory	18	18	20
Total	48	51	55

Table 4. Accuracy comparison of four floor positioning methods.

Method	WiFi AP deployment quantity requirement (points)	Fingerprint database workload (sampling points or fingerprints)	Correct rate (%)
WFL	28	210	85.7
MFL	0	84	80.0
NFL	28	210	91.4
MBFL	3	27	94.2