A completion algorithm based on an improved K-nearest neighbor (KNN) algorithm and extremely randomized trees (ERT) algorithm is proposed to address the problem of data loss in indoor fingerprint localization fingerprint databases. To improve the quality of fingerprint data, a hybrid filtering strategy combining median filtering and Gaussian filtering is employed in the preprocessing process. During the model training phase, the fingerprint data is divided into a training set and a testing set. The KNN algorithm, which combines Euclidean distance and Manhattan distance, is applied to determine the set of neighboring samples. Then, the ERT model is trained this set, and the outputs of all decision trees are integrated to generate the final prediction. Finally, perform data augmentation operations on the completed fingerprint database. The experimental results show that the proposed algorithm is superior in data completion compared with other algorithms, with a maximum completion error reduction of 19% and a prediction accuracy of up to 93%. The research conclusions provide a reference for improving the integrity of the fingerprint database.
在室内定位方法中,基于接收信号强度指示(received signal strength indication,RSSI)的Wi-Fi指纹定位技术,凭借无需额外硬件改造、成本低廉且易于部署等优势,成为室内定位领域的研究和应用热点[1]。该技术的核心在于构建一个能够准确反映室内空间信号分布特征的离线指纹数据库[2]。然而,在实际应用过程中,指纹数据库中的RSSI数据极易出现缺失现象[3],这主要是复杂室内环境的固有特性所致,如环境噪声、多径效应、设备性能差异以及信号遮挡等[4-5]。这些缺失值不仅直接破坏了数据的完整性和可靠性,更严重削弱了指纹库的内在关联性和结构一致性,导致后续定位匹配算法的精度显著下降。
ZHANGL P, CHENY, GAOX L, et al. An algorithm for indoor positioning based on LightGBM+ExtraTrees chaotic weighted ensemble: evaluation and comparison[J]. The Journal of Supercomputing, 2025, 81(7): 812.
[2]
LUY W, MAY T, HANY X, et al. DMLF: device-free multitarget adaptive backscatter link fingerprint localization[J].IEEE Transactions on Instrumentation and Measurement,2025,74: 5501111.
GUYufeng, DUYujie, WANGYuyang,et al.Study on UWB positioning algorithm based on neural networks and self-adjusting Kalman filters[J].China Mechanical Engineering,2023,34(12):1504-1511.
[5]
ZHANGT W, ZHANGP, KALATHASP, et al. A machine learning approach to improve ranging accuracy with AoA and RSSI[J]. Sensors, 2022, 22(17): 6404.
[6]
ZHANGL C, ZHANGW C, BAOX. Visible light fingerprint database recovery algorithm based on CP decomposition[J]. Optics express, 2023, 31(14): 22885-22902.
ZHANGWentao, WUFei, ZHUHai,et al.Indoor fingerprint localization method based on FDE-IRF[J].GNSS World of China, 2021, 46(4): 117-126.
[11]
FOUADK M, ISMAILM M, AZARA T, et al. Advanced methods for missing values imputation based on similarity learning[J]. PeerJ Computer Science, 2021, 7: e619.
[12]
SANJARK, BEKHZODO, KIMJ, et al. Missing data imputation for geolocation-based price prediction using KNN–MCF method[J].ISPRS International Journal of Geo-Information,2020,9(4):227.
[13]
NIKFALAZARS, YEHC H, BEDINGFIELDS, et al. Missing data imputation using decision trees and fuzzy clustering with iterative learning[J]. Knowledge and Information Systems, 2020, 62(6): 2419-2437.
[14]
SONIYAS L, RAJT A B. An intelligent system for lung CT image denoising using a hybrid WT-NLM filter[J]. Automatika, 2025, 66(2): 188-200.
CHANDRANS, GOKARAJUR, NARENDRAK. An extended impedance-based fault location algorithm in power distribution system with distributed generation using synchrophasors[J]. IET Generation, Transmission & Distribution, 2024, 18(3): 479-490.
[18]
AMINIFARA, SHOKRIM, RABBIF, et al. Extremely randomized trees with privacy preservation for distributed structured health data[J]. IEEE Access, 2022, 10: 6010-6027.
[19]
ÇETINA İ, BÜYÜKLÜA H. A new approach to K-nearest neighbors distance metrics on sovereign country credit rating[J]. Kuwait Journal of Science, 2025, 52(1): 100324.
[20]
WANGX W, YUEY, ZHANGF, et al. Active detection of interphase faults in distribution networks based on energy relative entropy and Manhattan distance[J]. Electric Power Systems Research, 2025, 241: 111397.
[21]
LIH L, QIANZ H, LIUG Q, et al. NQRELoc: AP selection via nonuniform quantization RSSI entropy for indoor localization[J]. IEEE Sensors Journal, 2022, 22(10): 9724-9732.
SHIJun, WANGTiantong, ZHUZiqi, et al. Deep learning-based medical image segmentation methods[J]. Journal of Image and Graphics, 2025, 30(6): 2161-2186.
ZHANGLieping, CHENYao, ZHENGXinpeng, et al. Information completion algorithm based on improved KNN-RF[J]. Electronic Measurement Technology, 2024, 47(13): 74-80.
[26]
DONGW N, FONGD Y T, YOONJ S,et al. Generative adversarial networks for imputing missing data for big data clinical research[J]. BMC Medical Research Methodology, 2021, 21: 78.