Indoor fingerprint location based on WiFi in large-scale indoor parking lots is more and more widely employed for vehicle lookup. However, the challenge is to ensure the location functionality because of the particularity and complexities of the indoor parking lot environment. To reduce the need to deploy of reference points (RPs) and the offline sampling workload, a partition-fitting fingerprint algorithm (P-FP) is proposed. To improve the location accuracy of the target, the PS-FP algorithm, a sampling importance resampling (SIR) particle filter with threshold based on P-FP, is further proposed. Firstly, the entire indoor parking lot is partitioned and the environmental coefficients of each partitioned section are gained by using the polynomial fitting model. To improve the quality of the offline fingerprint database, an error characteristic matrix is established using the difference between the fitting values and the actual measured values. Thus, the virtual RPs are deployed and C-means clustering is utilized to reduce the amount of online computation. To decrease the fluctuation of location coordinates, the SIR particle filter with a threshold setting is adopted to optimize the location coordinates. Finally, the optimal threshold value is obtained by comparing the mean location error. Test results demonstrated that PS-FP could achieve high location accuracy with few RPs and the mean location error is only about 0.7m. The cumulative distribution function (CDF) show that, using PS-FP, 98% of location errors are within 2m. Compared with the weighted K-nearest neighbors (WKNN) algorithm, the location accuracy by PS-FP exhibit an 84% improvement.
Weibo WANG
Xihua University
Jinghuan SUN
Xihua University
Ruiying DONG
Xihua University
Yongkang ZHENG
State Grid Sichuan Electric Power Research Institute
Qing HUA
Shandong Normal University
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copy
Weibo WANG, Jinghuan SUN, Ruiying DONG, Yongkang ZHENG, Qing HUA, "The Development of a High Accuracy Algorithm Based on Small Sample Size for Fingerprint Location in Indoor Parking Lot" in IEICE TRANSACTIONS on Communications,
vol. E101-B, no. 12, pp. 2479-2486, December 2018, doi: 10.1587/transcom.2018EBP3004.
Abstract: Indoor fingerprint location based on WiFi in large-scale indoor parking lots is more and more widely employed for vehicle lookup. However, the challenge is to ensure the location functionality because of the particularity and complexities of the indoor parking lot environment. To reduce the need to deploy of reference points (RPs) and the offline sampling workload, a partition-fitting fingerprint algorithm (P-FP) is proposed. To improve the location accuracy of the target, the PS-FP algorithm, a sampling importance resampling (SIR) particle filter with threshold based on P-FP, is further proposed. Firstly, the entire indoor parking lot is partitioned and the environmental coefficients of each partitioned section are gained by using the polynomial fitting model. To improve the quality of the offline fingerprint database, an error characteristic matrix is established using the difference between the fitting values and the actual measured values. Thus, the virtual RPs are deployed and C-means clustering is utilized to reduce the amount of online computation. To decrease the fluctuation of location coordinates, the SIR particle filter with a threshold setting is adopted to optimize the location coordinates. Finally, the optimal threshold value is obtained by comparing the mean location error. Test results demonstrated that PS-FP could achieve high location accuracy with few RPs and the mean location error is only about 0.7m. The cumulative distribution function (CDF) show that, using PS-FP, 98% of location errors are within 2m. Compared with the weighted K-nearest neighbors (WKNN) algorithm, the location accuracy by PS-FP exhibit an 84% improvement.
URL: https://global.ieice.org/en_transactions/communications/10.1587/transcom.2018EBP3004/_p
Copy
@ARTICLE{e101-b_12_2479,
author={Weibo WANG, Jinghuan SUN, Ruiying DONG, Yongkang ZHENG, Qing HUA, },
journal={IEICE TRANSACTIONS on Communications},
title={The Development of a High Accuracy Algorithm Based on Small Sample Size for Fingerprint Location in Indoor Parking Lot},
year={2018},
volume={E101-B},
number={12},
pages={2479-2486},
abstract={Indoor fingerprint location based on WiFi in large-scale indoor parking lots is more and more widely employed for vehicle lookup. However, the challenge is to ensure the location functionality because of the particularity and complexities of the indoor parking lot environment. To reduce the need to deploy of reference points (RPs) and the offline sampling workload, a partition-fitting fingerprint algorithm (P-FP) is proposed. To improve the location accuracy of the target, the PS-FP algorithm, a sampling importance resampling (SIR) particle filter with threshold based on P-FP, is further proposed. Firstly, the entire indoor parking lot is partitioned and the environmental coefficients of each partitioned section are gained by using the polynomial fitting model. To improve the quality of the offline fingerprint database, an error characteristic matrix is established using the difference between the fitting values and the actual measured values. Thus, the virtual RPs are deployed and C-means clustering is utilized to reduce the amount of online computation. To decrease the fluctuation of location coordinates, the SIR particle filter with a threshold setting is adopted to optimize the location coordinates. Finally, the optimal threshold value is obtained by comparing the mean location error. Test results demonstrated that PS-FP could achieve high location accuracy with few RPs and the mean location error is only about 0.7m. The cumulative distribution function (CDF) show that, using PS-FP, 98% of location errors are within 2m. Compared with the weighted K-nearest neighbors (WKNN) algorithm, the location accuracy by PS-FP exhibit an 84% improvement.},
keywords={},
doi={10.1587/transcom.2018EBP3004},
ISSN={1745-1345},
month={December},}
Copy
TY - JOUR
TI - The Development of a High Accuracy Algorithm Based on Small Sample Size for Fingerprint Location in Indoor Parking Lot
T2 - IEICE TRANSACTIONS on Communications
SP - 2479
EP - 2486
AU - Weibo WANG
AU - Jinghuan SUN
AU - Ruiying DONG
AU - Yongkang ZHENG
AU - Qing HUA
PY - 2018
DO - 10.1587/transcom.2018EBP3004
JO - IEICE TRANSACTIONS on Communications
SN - 1745-1345
VL - E101-B
IS - 12
JA - IEICE TRANSACTIONS on Communications
Y1 - December 2018
AB - Indoor fingerprint location based on WiFi in large-scale indoor parking lots is more and more widely employed for vehicle lookup. However, the challenge is to ensure the location functionality because of the particularity and complexities of the indoor parking lot environment. To reduce the need to deploy of reference points (RPs) and the offline sampling workload, a partition-fitting fingerprint algorithm (P-FP) is proposed. To improve the location accuracy of the target, the PS-FP algorithm, a sampling importance resampling (SIR) particle filter with threshold based on P-FP, is further proposed. Firstly, the entire indoor parking lot is partitioned and the environmental coefficients of each partitioned section are gained by using the polynomial fitting model. To improve the quality of the offline fingerprint database, an error characteristic matrix is established using the difference between the fitting values and the actual measured values. Thus, the virtual RPs are deployed and C-means clustering is utilized to reduce the amount of online computation. To decrease the fluctuation of location coordinates, the SIR particle filter with a threshold setting is adopted to optimize the location coordinates. Finally, the optimal threshold value is obtained by comparing the mean location error. Test results demonstrated that PS-FP could achieve high location accuracy with few RPs and the mean location error is only about 0.7m. The cumulative distribution function (CDF) show that, using PS-FP, 98% of location errors are within 2m. Compared with the weighted K-nearest neighbors (WKNN) algorithm, the location accuracy by PS-FP exhibit an 84% improvement.
ER -